LTE- Long Term Evolution
1
LTE/SAE Introduction
2
Orthogonal FDM
3
Multi-antenna Systems
4
E-UTRA Physical Layer
5
E-UTRA Layer 2 and 3
6
The X2- and S1-interface
7
The Evolved Packet Core
8
LTE/SAE Signalling Procedures
9
10
11
12
13
Apis Technical Training AB Tjärhovsgatan 21, 5th floor SE-116 28 Stockholm Sweden Tel +46-8-555 105 00 Fax +46-8-555 105 99 E-mail
[email protected] www.apistraining.com
14
15
Foldouts
The use of a term in this document should not be interpreted in a manner that would affect the validity or legal status of any proprietary proprietary rights that may be attached to that term. This is a training document and as such simplifies what are often highly complex technological issues. The system or systems described here should therefore be seen in that light, i. E. as simplifications rather than generalizations. generalizations. Due to ongoing progress in methodology, design, its contents are furthermore subject to revision without prior notice. Apis Technical Training AB assumes no legal responsibility for any error or damage resulting from the use of this document. Copyright © Apis Technical Training AB 2007. All rights are reserved. This training material is the confidential and proprietary property of Apis Technical Training AB. It is to be used solely for the purpose for which it was produced and is not to be copied or otherwise reproduced without the prior written permission of Apis Technical Training AB. To our best knowledge, the information in this document is accurate as per the date of publication. Other than by prior written agreement, Apis Technical Training AB will not update or otherwise advise of errors in the document which may be brought to our attention. All trademarks are trademarks of their respective owners. Apis Technical Training AB. welcomes customer feedback as part of a process of ongoing development of our documentation in order to better meet our customers' needs. Please submit your comments to our Head Office in Stockholm. Apis Technical Training AB th Tjärhovsgatan 21, 5 floor SE-116 28 Stockholm Sweden E-mail:
[email protected]
1
LTE/SAE Introduction Introduction 1.1 BACKGROUND ....................... ................................... ....................... ....................... ................. ..... 1-2 1.2 EVOLVED UTRA & UTRAN...................... UTRAN.................................. ..................... ......... 1-3 1.2.1
Network Architecture................................................................1-3
1.2.2
Requirements Requirements on E-UTRA/UTRAN.......................................... E-UTRA/UTRAN..........................................1-4 1-4
1.2.3
Overview of Technical Solutions..............................................1-5
1.3 EVOLVED PACKET CORE, EPC.............. EPC.......................... ....................... ............. .. 1-7 1.3.1
Network Architecture................................................................1-7
1.3.2
Requirements Requirements on the EPC .......................................................1-8
1.4 EVOLVED HSPA (HSPA+) ....................... ................................... ..................... ......... 1-9 1.5 REFERENCES ...................... ................................. ....................... ....................... ................. ...... 1-10
Apis Technical Training AB LTE - LTE/SAE Introduction Copyright © Apis Technical Training AB 2007. All rights reserved.
1-1
1.1
Background 3GPP Long 3GPP Long Term Evolution (LTE) is the name given to a project within the Third Generation Partnership Project (3GPP) to improve the UMTS 3G mobile system standard to cope with future requirements. Goals include improving efficiency, lowering costs, reducing complexity, improving services, making use of new spectrum opportunities and better integration with other open standards (such as WLAN and WiMAX). Thus, the term ‘LTE’ really means a standardisation project . The final outcome from this project will be a new set of standards defining the functionality and requirements of an evolved, packet based, radio access network and a new radio access. The new radio access network is referred to as the ‘Evolved UTRAN’ (E-UTRAN) and the new radio access is referred to as the ‘Evolved UTRA’ (E-UTRA). The LTE project belongs to 3GPP Release 8. The term ‘LTE’ has recently become more or less synonymous to the (proper) terms ‘Evolved UTRA’ (the new radio access) and ‘Evolved UTRAN’ (the new radio access network). With this in mind, t he author has taken the freedom to use the terms ‘LTE’ and ‘E-UTRA’ interchangeably for the new OFDM-based radio interface. The term ‘E-UTRAN’ explicitly means the whole radio access network (i.e. it includes the eNBs, the X2interface and the S1-interface). The work on LTE started with a workshop, 2-3 Nov 2004 in Toronto, Canada. The workshop was open to members and non-members of 3GPP. Operators, vendors and research institutes presented contributions with views and proposals on the future evolution of 3G. A set of high level requirements were initially identified: Reduced cost per transmitted bit More services at lower cost with better user experience Flexibility of use of existing and new frequency bands Simplified architecture, open interfaces Reasonable terminal power consumption. • • • • •
It was also recommended that the E-UTRAN should bring significant improvements to justify the standardization effort and that it should avoid unnecessary options. A feasibility study on the UTRA & UTRAN Long Term Evolution was then started in i n December 2004. The objective was "to develop a framework for the evolution of the 3GPP radio access technology towards a high data rate, low latency and packet optimized radio access technology". The study focused on supporting services exclusively from the Packet Switched (PS) domain. In parallel to, and coordinated with, the LTE project there is also a 3GPP standardisation project relating to the core network. This project is called System Architecture Evolution (SAE) and aims at standardising the Evolved Packet Core (EPC). The SAE project was started in December 2004, with the objective to “develop a framework for an evolution or Apis Technical Training AB LTE - LTE/SAE Introduction Copyright © Apis Technical Training AB 2007. All rights reserved.
1-2
migration of the 3GPP system to a higher data rate, lower latency, packet optimized system that supports multiple RATs”. The EPC will be a fully IP-based core network (‘all-IP’) supporting access not only via GERAN, UTRAN and E-UTRAN but also WiFi, WiMAX and wired technologies such as xDSL. The SAE project also belongs to 3GPP Release 8. A short introduction to the Evolved UTRA/N can be found in section 1.2 in this chapter, and an introduction to the EPC in section 1.3. The Stage 2 set (general architecture, protocol structure and key concepts) of LTE standardisation documents is, according to 3GPP, to be completed at the time of writing this document (Oct 2007). The completion date for the Stage 3 work (i.e. detailed protocol specifications) is still a bit uncertain, but a reasonable estimate is ‘early 2008’. The Stage 2 set of SAE standardisation documents are (again according to 3GPP) to be completed by March 2008, with Stage 3 following shortly afterwards. One should be aware that major updates/changes/additions to the E-UTRAN/EPC specs are expected throughout 2008-09. Real-life deployment of LTE/SAE networks should therefore not be expected until 2009-10. The reader is strongly encouraged to regularly check the 3GPP website (www.3gpp.org) www.3gpp.org) for new versions of the standardisation documents referenced at the end of each chapter in the current document.
1.2
Evolved UTRA & UTRAN
1.2.1
Network Architecture Evolved UTRAN
Evolved Packet Core
eNB
X2 S1 X2
eNB
MME SGW
X2
eNB
Figure 1-1: The Evolved UTRAN architecture
The Evolved UTRAN consists of the evolved NodeB (eNB), providing the E-UTRA User Plane (UP) and Control Plane (CP) protocol terminations towards the UE. The eNBs are interconnected with each other by means of Apis Technical Training AB LTE - LTE/SAE Introduction Copyright © Apis Technical Training AB 2007. All rights reserved.
1-3
the X2-interface , e.g. for support of handovers without data loss. The eNBs are connected by means of the S1-interface to the EPC. The S1-interface supports a many-to-many relation between eNBs and MME/SGWs (see section 1.3.1). The X2- and S1-interfaces are described in in chapter 6. The eNB can be seen as a combination of the UMTS NodeB and Radio Network Controller, hosting functions like dynamic resource allocation (through packet scheduling) and radio resource management.
1.2.2
Requirements on E-UTRA/UTRAN At the onset of the LTE project a series of requirement targets relating to performance, complexity and interworking were defined. Some of these are listed below: Peak data rate: rate: at least 100 Mb/s DL and 50 Mb/s UL (assuming 20 MHz system bandwidth). Control Plane (CP) latency: latency : transition time less than 100 ms from an idle state to an active state, and less than 50 ms between a dormant state (such as R6 CELL_PCH) and an active state. User Plane (UP) latency: latency : less than 5 ms in unloaded condition (single user with single data stream) for small IP packet. CP capacity: capacity: at least 200 users per cell should be supported in the active state (5 MHz system bandwidth). Mobility: Mobility: E-UTRAN should be optimized for low mobile speed (015 km/h) and higher speeds (15-120 km/h) should be supported with high performance. Mobility shall be maintained between 120350 km/h (up to 500 km/h depending on the frequency band). Coverage: Coverage: the throughput and mobility targets above should be met for 5 km cells with a slight degradation for 30 km cells. Cells range up to 100 km should be possible. Spectrum flexibility: flexibility : E-UTRA shall operate in different spectrum allocations of different sizes, including 1.25, 1.6, 2.5, 5, 10, 15 and 20 MHz in both UL and DL. Operation in paired (FDD) and unpaired (TDD) spectrum shall be supported. Interworking: Interworking: co-existence in the same geographical area and colocation with GERAN/UTRAN on adjacent channels. E-UTRAN terminals supporting also UTRAN/GERAN operation should be able to support measurement of, and handover from/to, both UTRAN and GERAN. The interruption time during a handover of real-time services between E-UTRAN and UTRAN/GERAN should be less than 300ms. Architecture : the E-UTRAN architecture shall be packet based, supporting real-time and conversational class traffic. The architecture shall minimize the presence of "single points of failure". Complexity: Complexity: minimised number of options and avoidance of redundant mandatory features. •
•
•
•
•
•
•
•
•
•
Apis Technical Training AB LTE - LTE/SAE Introduction Copyright © Apis Technical Training AB 2007. All rights reserved.
1-4
1.2.3
Overview of Technical Solutions The E-UTRA radio interface makes exclusive use of shared channels for both data and signalling transfer. In this respect the E-UTRA is similar to the 3GPP R5/R6 High Speed Packet Access, HSPA. The selected radio access technology, however, is very different to HSPA. Where HSPA uses WCDMA, the E-UTRA uses Orthogonal Frequency Division Multiplexing (OFDM). OFDM splits the available system bandwidth into hundreds, or even thousands, of narrow-band so-called ‘sub-carriers’. This means that a high bitrate data stream to a given UE is split by the eNB into a large number of narrow-band, low bitrate, data streams. The received parallel data streams (sub-carriers) are then ‘multiplexed’ by the UE in order to re-create the original high bitrate data stream. This has several advantages over WCDMA: •
•
•
Better spectral efficieny. efficieny. More information can be sent using the same system bandwidth as compared to a single-carrier system. Flexible/scalable spectrum allocation. allocation. The system bandwidth can be expanded in increments (by ‘adding’ more sub-carriers) as more spectrum becomes available to the operator. For example, the initial system roll-out may use a system bandwidth of 1.25 MHz and at a later stage this may be increased to, say, 2.5MHz (and then 5MHz, 10MHz and so on). Better performance under multipath fading conditions. conditions . Multipath effects leads to so-called frequency selective fading, which is much more damaging to a wideband signal than to a narrowband signal (the sub-carrier).
There are, of course, drawbacks with OFDM as well. One such drawback is that an OFDM signal exhibits a very high peak-to-average power ratio (PAPR). This is not really a problem on the network side, but leads to very inefficient use of power amplifiers, and hence high power consumption, in a mobile terminal. The E-UTRA system therefore uses a variant of OFDM for uplink transmission that reduces PAPR. This variant of OFDM is called Single-Carrier Frequency Division Multiple Access (SC-FDMA). Despite the name, there is very little that differentiates SC-FDMA from ‘classic’ OFDM. Chapter 2 contains more information on OFDM. The use of Multiple Input Multiple Output antenna arrays (MIMO) is an integral part of the E-UTRA standard. The standard supports up to four transmit/receive antennas while the expected baseline configuration is two transmit antennas at the eNB and two receive antennas at the UE. In short, MIMO can be used in two different ways:
Apis Technical Training AB LTE - LTE/SAE Introduction Copyright © Apis Technical Training AB 2007. All rights reserved.
1-5
•
•
To transmit more information over the radio interface without using more bandwidth than a single antenna system. The number of antennas used increases the system capacity in a linear manner, i.e. two antennas allows twice the amount of information to be transmitted (or, equivalently, the bitrate is doubled). To transmit the same information, with the same bitrate as a single antenna system, but with less output power (or, equivalently, with higher reliability).
An overview of various MIMO techniques and the mechanisms selected for E-UTRA can be found in chapter 3. The E-UTRA physical layer channel processing chain (channel coding and modulation) is very similar to what is used today for HSPA. It was agreed at an early stage in the standardisation process that Turbo coding should be used for error correction purposes and that the supported data modulation schemes should be QPSK, 16QAM, and 64QAM for downlink and uplink.
Figure 1-2: Constellation diagrams for QPSK (left), 16QAM (middle) and 64QAM (right)
The mapping of modulation symbols onto physical channel resources is very different compared to HSPA though. The nature of OFDM gives rise to the concept of 2-dimensional radio resources. The information to be transmitted over the radio interface is mapped onto a 2-dimensional timefrequency ‘resource grid’. The E-UTRA physical layer is described in all its glorious detail in chapter 4. (A common misunderstanding is that OFDM, by itself, makes very high bit rates possible. This is not true. Rather, the very high bit rates mentioned for E-UTRA are made possible first and foremost by a higher transmission bandwidth (up to 20MHz), higher order modulation (64QAM) and the support for MIMO with up to t o four antennas). The channel and protocol architecture for E-UTRAN layer 2 and layer 3 is quite similar to the one used in UTRAN today. For example, the UE protocol stack is close to identical and the channel hierarchy is still divided
Apis Technical Training AB LTE - LTE/SAE Introduction Copyright © Apis Technical Training AB 2007. All rights reserved.
1-6
into logical, transport and physical channels. The E-UTRAN protocol/ channel architecture is described in chapter 4. The exact functionality of the layer 2 and layer 3 protocols in E-UTRAN is, at the time of writing, far from decided. An overview of the expected functionality can be found in chapter 5.
1.3
Evolved Packet Core (EPC)
1.3.1
Network Architecture GERAN/ UTRAN
Gb/Iu SGSN
S3
S1-MME
MME
S4
S11
S5
S1-U E-UTRAN
SGW
SGi
PGW
IMS / Internet /…
S2 Non-3GPP access
Figure 1-3: the Evolved Packet Core network architecture
Figure 1-3 shows the network architecture of the Evolved Packet Core (EPC). The EPC consists of three main nodes: the Mobility Management Entity (MME), the Serving Gateway (SGW) and the Packet Data Network Gateway (PGW). The MME may be co-located with the SGW, and the SGW may be co-located with the PGW. Hence, the standard allows a completely collapsed ‘one-node’ core network or a distributed (easily scalable) core network, or any possible ‘combination’ in-between. The MME connects to the E-UTRAN via the S1-MME interface and is present solely in the CP. It is responsible for handling mobility and security procedures, such as network Attach, Tracking Area updates (similar to Location/Routing Area updates) and authentication. The MME also connects to the SGSN via the S3-interface . The SGW connects to the E-UTRAN via the S1-U interface and is present solely in the UP. Its prime responsibility is routing and forwarding of user IP-packets. It acts as a UP anchor when the UE moves between 3GPP radio access technologies ( S4-interface ).
Apis Technical Training AB LTE - LTE/SAE Introduction Copyright © Apis Technical Training AB 2007. All rights reserved.
1-7
The PGW connects to the SGW via the S5-interface and to external packet data networks (or IMS) via the SGi-interface . It is responsible for the enforcing of QoS and charging policies. It also acts as a UP anchor when the UE moves between 3GPP and non-3GPP radio access ( S2-interface ). It should be noted that additional network nodes/functions, not shown in figure 1-3, might be present as well. For example, a Packet Data Gateway (PDG) is needed for non-trusted IP access and a Policy and Charging Rules Function (PCRF) is required for IMS controlled QoS and charging mechanisms. The EPC is described further in chapter 7.
1.3.2
Requirements on the EPC A (rather long) list of general requirements has been set up as guidelines for the standardisation work related to the EPC. Some of those are: 3GPP and non-3GPP access systems shall be supported. Scalable system architecture and solutions without compromising the system capacity (e.g. by separating CP from UP). CP response time shall be such that the UE can move from an idle state to one where it is sending/receiving data in less than 200 ms. Basic IP connectivity is established during the initial access phase (hence, the UE is ‘always-on’). The Mobility Management (MM) solution shall be able to accommodate terminals with different mobility requirements (fixed, nomadic and mobile terminals). The MM functionality shall allow the network operator to control the type of access system being used by a subscriber. MM procedures shall provide seamless operation of both real-time (e.g. VoIP) and non real-time applications. In order to maximise users' access opportunities, the architecture should allow a UE that is roaming to use a non-3GPP access (e.g. WLAN) network with which the VPLMN has a business agreement. For example, it should be possible for a user to use a WLAN access network with which only the visited operator has a direct relationship (not the home operator). The evolved system shall support Ipv4 and Ipv6 connectivity. Access to the evolved system shall be possible with R99 USIM. (Please note that this does not explicitly allow access using SIM) The authentication framework should be independent from the used access network technology. Radio interface multicast capability shall be a built-in feature. The SAE/LTE system shall support network-sharing functionality. It shall be possible to support service continuity between IMS over SAE/LTE access and the Circuit Switched (CS) domain. It shall be possible for the operator to provide the UE with access network information pertaining to locally l ocally supported 3GPP and non3GPP access technologies. • •
•
•
•
•
•
•
• •
•
• • •
•
Apis Technical Training AB LTE - LTE/SAE Introduction Copyright © Apis Technical Training AB 2007. All rights reserved.
1-8
1.4
Evolved HSPA (HSPA+) Iu
cSGSN
RNC Iur
Gn Iu/Gn Iu/Gn
NB
G Gii
xGGSN
IMS / Internet /…
Figure 1-4: Evolved HSPA network architecture
A parallel 3GPP R8 project to LTE and SAE is the Evolved High Speed Packet Access, eHSPA, project (also referred to as HSPA+). The proposed eHSPA features represent a logical evolution from today’s HSDPA and HSUPA systems. Roughly speaking, the eHSPA project focuses on three areas: •
•
•
Optimising HSPA for real-time packet data services, like VoIP. A large part of achieving this goal relates to a more efficient use of the HSPA control channels. Increasing the system and user throughput. This is achieved by the introduction of higher order modulation (64QAM) and MIMO for HSPA. The theoretical maximum bit rate is around 40Mb/s for the DL and around 20Mb/s for the UL. Simplifying the network architecture. The eHSPA NodeB will take on parts of, or all of, the functionality of the RNC. In addition, the SGSN will be removed from the User Plane path (the so-called ‘one-tunnel solution’) allowing IP packets to be routed directly between eHSPA NodeB and GGSN. This can be seen in figure 1-4 above, where ‘cSGSN’ is the SGSN Controller , and ‘xGGSN’ is the enhanced GGSN .
Thus, E-UTRA/E-UTRAN and Evolved HSPA have many concepts in common (collapsed architecture, 64QAM, MIMO). As a matter of fact, the performance (bit rates, spectral efficiency etc) of eHSPA is very close to the performance of E-UTRA with 5MHz channel bandwidth. This has led to some level of debate whether to refer to eHSPA as a ‘migration path’ or a ‘complement’ or a ‘competing technology’.
Apis Technical Training AB LTE - LTE/SAE Introduction Copyright © Apis Technical Training AB 2007. All rights reserved.
1-9
1.5
References 23.401 23.402 23.882 25.912 25.913 25.999 36.300
GPRS enhancements for Long Term Evolution (LTE) 3GPP SAE: Architecture enhancements enhancements for non-3GPP accesses 3GPP SAE: Report on technical options and conclusions Feasibility study for E-UTRA and E-UTRAN Requirements for E-UTRA and E-UTRAN High Speed Packet Access (HSPA) evolution, FDD E-UTRA/E-UTRAN; overall description; Stage 2
Apis Technical Training AB LTE - LTE/SAE Introduction Copyright © Apis Technical Training AB 2007. All rights reserved.
1-10
2
OFDM 2.1 OFDM BASICS ....................... ................................... ....................... ....................... ............... ... 2-2 2.1.1
Introduction ..............................................................................2-2
2.1.2
Sub-carriers Sub-carriers and Multiplexing ..................................................2-3
2.1.3
Orthogonality............................................................................2-4
2.1.4
Cyclic Prefixes..........................................................................2-6
2.2 OFDM SIGNAL GENERATION ...................... .................................. ................... ....... 2-7 2.3 SC-FDMA....... SC-FDMA................... ....................... ....................... ....................... ....................... ............... ... 2-8 2.4 REFERENCES ...................... ................................. ....................... ....................... ................. ...... 2-10
Apis Technical Training AB LTE - OFDM Copyright © Apis Technical Training AB 2007. All rights reserved.
2-1
2.1
OFDM Basics (The theory behind OFDM is very mathematical in its nature. The following is just a brief overview in ‘layman’s terms’ to convey the basic characteristics of OFDM. For a deeper understanding of OFDM it is recommended to consult a textbook on the subject)
2.1.1
Introduction Orthogonal Frequency Division Multiplexing (OFDM) is a digital multicarrier modulation scheme that uses a large number of closely-spaced orthogonal sub-carriers . Each sub-carrier is modulated with a conventional modulation scheme (such as 16QAM) at a low symbol rate, maintaining data rates similar to conventional single-carrier modulation schemes in the same bandwidth. The primary advantage of OFDM over single-carrier schemes is its ability to cope with severe channel conditions without complex equalization filters. Low symbol rate makes the use of a guard interval between symbols affordable, making it possible to handle time-spreading and inter-symbol interference (ISI). OFDM has only become widely used during the last decade or so, but the technology as such is about 50 years old (it was first used around 1957 in an experimental communications system developed for the US Navy). During the 70’s and 80’s several important theoretical contributions from various sources made it possible to implement more efficient and robust OFDM-based systems. Today, OFDM has proved itself as the preferred radio access technology in a wide variety of communication systems. Some examples of OFDM use: IEEE 802.11a/g (WLAN/WiFi), IEEE 802.16 (WiMAX), Digital Audio Broadcasting (DAB), Digital Video Broadcasting (DVB-T and DVB-H) and Asynchronous Digital Subscriber Line (ADSL). Some advantages of OFDM: Allows adaptation to severe channel conditions without very complex equalization methods Robust against narrow-band co-channel interference Robust against Inter-Symbol Interference (ISI) and fading caused by multipath propagation High spectral efficiency Efficient implementation using FFT Low sensitivity to time synchronization errors Facilitates Single Frequency Networks (i.e. (i .e. synchronised broadcast from several transmitters). •
• •
• • • •
Some disadvantages of OFDM: Sensitive to Doppler shift Sensitive to frequency synchronization problems High peak-to-average power ratio (PAPR), requiring more expensive transmitter circuitry and lowering power efficiency. • • •
Apis Technical Training AB LTE - OFDM Copyright © Apis Technical Training AB 2007. All rights reserved.
2-2
2.1.2
Sub-carriers and Multiplexing As we have already mentioned in chapter 1, OFDM spreads the data to be transmitted over a large number of sub-carriers- typically more than a thousand. The data rate to be conveyed by each of these sub-carriers is correspondingly reduced, reduced, thus transforming a single high bitrate channel to many low bitrate channels. It follows that the modulation (OFDM) symbol length is in turn extended, which dramatically reduces the system’s sensitivity to inter-symbol interference due to multipath effects (i.e. different versions, or ‘echoes’, of the same signal travelling different paths over the radio interface and arriving at the receiver at different points in time, causing interference). This is true as long as the maximum delay of the ‘echoes’ is smaller than the OFDM symbol time duration. LONG delay (or short symbols)
n-1
Main Path:
Symbol n
Symbol n-3
Delayed Path:
SHORT delay (or long symbols)
n+1
n-1
Symbol n-2
Symbol n
Symbol n-1
Both cause ISI
Causes ISI
n+1
Symbol n
Adds constructively or destructively
Figure 2-1: multipath delays versus symbol length
With hundreds or thousands of sub-carriers available it becomes quite straightforward how to multiplex users on the radio interface. Simply allocate different sets of sub-carriers to different users (this is the ‘FDM’ in OFDM). More complex multiplexing schemes can be implemented by allowing users to share the available sub-carriers both in the frequency domain (FDM) and the time domain (TDM). f
g n i x e l p i t l u m r e i r r a c b u S
Sub-carrier n
: M D F
Sub-carrier 1 Radio frame 1
Radio frame m
t
TDM: radio frame multiplexing
Figure 2-2: OFDM multiplexing using both FDM and TDM
Apis Technical Training AB LTE - OFDM Copyright © Apis Technical Training AB 2007. All rights reserved.
2-3
Furthermore, sub-carrier frequency hopping schemes may be applied to reduce fading effects that are frequency selective. The transmitter (base station) can also order the receivers (mobile stations) to send feedback information in the form of channel quality reports. This allows dynamic channel dependent scheduling in the base station, making sure that each mobile station is always allocated a subset of sub-carriers where it experiences the least amount of interference. A graphical representation of frequency selective fading effects can be seen in figure 2-3 below.
Figure 2-3: frequency selective fading effect on OFDM sub-carriers
E-UTRA combines OFDM with FDM and TDM multiplexing schemes as well as sub-carrier frequency hopping and channel dependent scheduling.
2.1.3
Orthogonality In traditional FDM different users are allocated different frequencies, or channels, for their transmission (e.g. analog 1G systems such as NMT). To avoid interference between these channels the FDM frequencies must be spaced apart- there must be a guard band between them. This leads to waste of the available frequency spectrum. FDM: FDM: guard band between carriers
OFDM: OFDM: carriers can be packed tighter
Saving bandwidth
Figure 2-4: FDM versus OFDM (spectrum efficiency)
In OFDM, the frequencies of the individual sub-carriers are chosen in such a way that they do not interfere with each other- they are orthogonal (this is the ‘O’ in OFDM). The demodulator for one sub-carrier does not 'see' Apis Technical Training AB LTE - OFDM Copyright © Apis Technical Training AB 2007. All rights reserved.
2-4
the modulation of the others, so there is no crosstalk between sub-carriers even though their spectra overlap. This allows us to ‘pack’ the sub-carriers much more densely than in a traditional FDM system, thus increasing spectrum efficiency. All the sub-carriers allocated to a given user are transmitted in parallel. Fortunately, the apparently very complex processes of modulating (and demodulating) thousands of sub-carriers simultaneously are equivalent to Discrete Fourier Transform (DFT) operations for which efficient Fast Fourier Transform (FFT) algorithms exist, allowing affordable massproduced transceivers. To ensure orthogonality the sub-carriers must have a common, precisely chosen frequency spacing (‘carrier spacing’). This frequency spacing is exactly the inverse of the OFDM symbol duration, called the active symbol period , over which the receiver will demodulate the signal. In the case of E-UTRA the carrier spacing is 15kHz (7.5kHz for MBMS dedicated cells).
Received signals are evaluated at their maximum
Figure 2-5: orthogonal OFDM sub-carriers (frequency domain)
Figure 2-5 shows a few sub-carriers represented in the frequency domain (compare figure 2-3). The receiver will demodulate (or sample) each subcarrier precisely where it has it maximum value. Due to the ‘preciselychosen frequency spacing’ all other sub-carrier have the value zero at this precise frequency, despite that they overlap, thus not creating any interference at all. However, this nice relationship between sub-carriers can be destroyed, resulting in loss of orthogonality and severe bit error rates as a result. The bit errors can be rectified to some extent with the use of error correcting codes, so called Forward Error Correction (FEC). A combination of FEC and OFDM is called Coded OFDM (COFDM). In E-UTRA, OFDM is combined with Turbo coding. Apis Technical Training AB LTE - OFDM Copyright © Apis Technical Training AB 2007. All rights reserved.
2-5
Such loss of orthogonality can be caused by frequency synchronisation errors due to slight differences in the local oscillators, used for frequency generation, in the transmitter and the receiver. Another cause is Doppler effects arising from the relative motion between the transmitter and the receiver- an effect that must be taken seriously in any mobile system!
2.1.4
Cyclic Prefixes As mentioned earlier, OFDM is robust against multipath fading due to the long OFDM symbol duration. However, there will always be some intersymbol interference due to multipath echoes, even for OFDM. A further refinement therefore adds the concept of a guard interval. Each OFDM symbol is transmitted for a total symbol period that is longer than the active symbol period by a period called the guard interval or guard period.
Guard period
Main Path:
Delayed Path:
n-1
CP
Useful part
Symbol n
n+1
n
n-1
ISI only during CP
Figure 2-6: guard interval with cyclic prefix
This means that the receiver will experience neither inter-symbol nor intercarrier interference provided that any echoes present in the signal have a delay that does not exceed the guard interval. Naturally, the addition of the guard interval reduces the data capacity by an amount dependent on its length. Different systems use different (relative) lengths of the guard interval, common values being 5-25% of the OFDM symbol length. There are several ways to ‘fill’ the guard interval with information (to avoid turning the transmitter on and off abruptly). A common mechanism is the use of a so-called cyclic prefix . A cyclic prefix (CP) is created simply by selecting the last part of an OFDM symbol, make a copy of it and place the copy in front of the symbol (hence the term ‘prefix’). The concept of a guard interval is illustrated in figure 2-6 above.
Apis Technical Training AB LTE - OFDM Copyright © Apis Technical Training AB 2007. All rights reserved.
2-6
By doing so, a continuous signal is created (easier to implement) and any multipath delayed echoes will only cause interference in the CP part of the received OFDM symbol. The receiver treats the CP portion of the OFDM symbol as ‘rubbish’ and removes it prior to demodulating the information. E-UTRA defines a ‘normal’ length and an ‘extended’ length of the CP, to cater for the different requirements of small versus large cells. There are also different CP lengths defined for MBMS transmission, when multiple synchronised eNBs act as a Single Frequency Network (SFN).
2.2
OFDM Signal Generation S Coding Modulation
…
P
I F F T
Add CP
RF
fo
Figure 2-7: the OFDM transmitter using IFFT
There are several ways to realize an OFDM transmit-receive chain. For example, the addition of a cyclic prefix is not mandatory and filtering/ equalization of the baseband signal (the ‘RF’ box in fig 2-7) can be done in many different ways. Thus, figure 2-7 below does not represent the way to create an OFDM signal. Coding and Modulation: Modulation : this step is any conventional Forward Error Correction (FEC) mechanism, such as convolutional coding, and any conventional modulation scheme, such as QPSK or 64QAM. Serial-to-Parallel : Serial-to-Parallel : a group of modulation symbols are ‘fed’ to the Inverse Fast Fourier Transform (IFFT) in parallel. The number of modulation symbols fed to IFFT equals the number of assigned sub-carriers. Inverse Fast Fourier Transform: each modulation symbol is used for modulating one sub-carrier, in effect acting as a complex wight setting the amplitude and phase of the sub-carrier. These modulated sub-carriers are then summed together, creating one OFDM symbol. Cyclic Prefix: Prefix : the last portion of the OFDM symbol is copied and appended at the ‘front’ of the symbol. This creates a guard interval with well-defined content.
Apis Technical Training AB LTE - OFDM Copyright © Apis Technical Training AB 2007. All rights reserved.
2-7
RF processing processing : the OFDM symbols are used for modulation of the actual carrier frequency. In addition, various pulse shaping or filteri ng techniques may be applied at this stage. The receiving side uses the process in reverse. The IFFT process must be inverted in order to retrieve the information content of the individual subcarriers. This is done with the Fast Fourier Transform (FFT). Hence, the inverse of the Inverse-FFT is, of course, the FFT.
2.3
SC-FDMA In chapter 1 it was mentioned that one drawback with OFDM was its high peak-to-average power ratio (PAPR). This is a direct consequence of using DFT (FFT) to create the OFDM symbols. The DFT effectively stacks sinewaves ‘on top’ of each other. It can then of course happen that a large portion of the used sub-carriers happen to have their maximum value at the same time, resulting in a dramatic peak in the total amplitude (or power) of the signal. This puts very high demands on the power amplifier in the signal processing chain and is not desirable in a small portable device, such as a mobile phone, with limited battery capacity.
S Coding Modulation
…
P
F F T
…
M a p p i n g
0 0 0 0
…
I F F T
0 0
Add CP
RF
fo
Figure 2-8: the SC-FDMA transmitter using FFT spreading
The solution for this in the E-UTRA specification is to use a variant of OFDM called Single-Carrier Frequency Division Multiple Access (SCFDMA) for uplink transmission. Figure 2-8 shows the processing cain for an SC-FDMA transmitter. A comparison to figure 2-7 reveals that two additional steps have been added to the processing chain: an FFT transform and a sub-carrier mapping stage (the dotted box in figure 2-8). As for ‘classic’ OFDM, a block of modulation symbols are fed in parallel into the transform stage, which is now FFT instead of IFFT. The FFT process will now spread each modulation symbol over all sub-carriers instead of using 1-to-1 mapping. In other words, the input signal (modulation symbols) will be spread over the available bandwidth (the available sub-carriers) very much like in a single-carrier system. (It could be worth knowing that SC-FDMA is also referred to as ‘DFT-spread OFDM’) Apis Technical Training AB LTE - OFDM Copyright © Apis Technical Training AB 2007. All rights reserved.
2-8
The sub-carrier mapping stage (‘mapping’ in figure 2-8) then feeds the FFT-output to a subset of the IFFT-inputs, with all other inputs set to zero. Thus FFT has size=N and the IFFT has size=M, with M>N. The output from the IFFT stage is now called an SC-FDMA symbol, to which we add a cyclic prefix in the normal manner. Thus, the result of the additional FFT stage is that the created signal exhibits single-carrier properties (the ‘SC’ in SC-FDMA). Furthermore, different users will be ordered to transmit on different, orthogonal, ‘chunks’ of subcarriers (the ‘FDM’ in SC-FDMA). There are different ways of selecting which specific sub-carriers that should be part of the ‘chunk’ for a given user. For localized SC-FDMA, a set of consecutive sub-carriers are selected (e.g. use sub-carrier number 20-40). For distributed SC-FDMA, the sub-carriers are evenly distributed (e.g. use sub-carrier 5, 10, 15 etc). A third option is to select sub-carriers that are neither consecutive nor evenly distributed, but selected according to some other pattern. This gives us randomized SC-FDMA. Currently, the only option allowed for E-UTRA uplink is localized SC-FDMA.
Apis Technical Training AB LTE - OFDM Copyright © Apis Technical Training AB 2007. All rights reserved.
2-9
2.4
References 25.912 36.211 36.300
Feasibility study for E-UTRA and E-UTRAN E-UTRA; Physical channels and modulation E-UTRA E-UTRAN; overall description; Stage 2
http://en.wikipedia.org/wiki/OFDM This Wikipedia article gives a good introduction to OFDM and contains numerous references to other websites, white papers and textbooks about OFDM and related topics.
Apis Technical Training AB LTE - OFDM Copyright © Apis Technical Training AB 2007. All rights reserved.
2-10
3
Multi-antenna Systems 3.1 MULTIPLE ANTENNA SYSTEMS ....................... .................................. ................. ...... 3-2 3.1.1
SIMO/MISO: Receive/Transmit Receive/Transmit Diversity.................................3-2 Diversity .................................3-2
3.1.2
MIMO: Multiple Input Multiple Output.......................................3-4
3.2 MIMO TECHNIQUES ...................... ................................. ....................... ..................... ......... 3-5 3.2.1
Spatial Multiplexing Multiplexing (SM) .........................................................3-5
3.2.2
Space-Time Coding (STC).......................................................3-5
3.2.3
MU-MIMO and SU-MIMO.........................................................3-6 SU-MIMO.........................................................3-6
3.2.4
MIMO and OFDM.....................................................................3-6
3.3 MIMO FOR E-UTRA E-UTRA ....................... .................................. ....................... ................... ....... 3-7 3.4 REFERENCES ....................... .................................. ...................... ....................... ................... ....... 3-9
Apis Technical Training AB LTE - MIMO Copyright © Apis Technical Training AB 2007. All rights reserved.
3-1
3.1
Multiple Antenna Systems (The theory behind MIMO is, like OFDM, highly mathematical in its nature. This chapter provides an overview in ‘layman’s terms’ to convey the basics of MIMO. For a fuller/ deeper understanding of MIMO it is recommended to consult a textbook on the subject)
Any wireless communications system with one transmit (Tx) antenna and one receive (Rx) antenna is referred to as operating in Single Input Single Output (SISO) mode. More antennas (Tx and/or Rx) can be added in order to increase either throughput or reliability. Systems with multiple Tx/Rx antennas are divided into Single Input Multiple Output (SIMO), Multiple Input Single Output (MISO) or Multiple Input Multiple Output (MIMO). Simple multi-antenna systems have been around, in one form or another, for over 50 years (Guglielmo Marconi used multiple antenna transmission to transmit a Morse signal across the Atlantic Ocean, from England to Newfoundland, in 1901). But until quite recently, the amount of signal processing needed has been too expensive to be practical for large-scale deployment and implementation in small mobile devices. Important factors driving MIMO acceptance today is the advent of in-expensive high-speed Digital Signal Processors (DSPs) and significant research breakthroughs in information theory over the last decade. MIMO is currently used in various WLAN systems (IEEE 802.11 family) and in WiMAX (IEEE 802.16 family) to name a few. MIMO is also an integral part of the 3GPP R8 standards pertaining to eHSPA and LTE.
3.1.1
SIMO/MISO: Receive/Transmit Diversity
TX
RX
Figure 3-1: a SIMO system
In a SIMO system the transmitter has one antenna and the receiver has two, or more, physically separated antennas (the physical separation distance has a direct relationship with the wavelength of the carrier). This allows for receive diversity (Rx diversity). With Rx diversity the receiver picks up two (or more) ‘versions’ of the same transmitted signal. The receiver may then either:
Apis Technical Training AB LTE - MIMO Copyright © Apis Technical Training AB 2007. All rights reserved.
3-2
•
•
select the best input (from one of the antennas), for example based on signal-to-noise ratio (SNR). This is called switched diversity . combine the input from all antennas, for example through a process called Maximum Ratio Combining (MRC).
With MRC, channel compensation is applied to each received signal before being linearly combined to create a single, composite, received signal. Rx diversity using MRC is used in challenging propagation conditions when signal strength is low and/or multipath conditions are severe. MRC uses the fact that it is statistically very unlikely that a given signal will undergo deep fading on all Rx channels simultaneously. The possibility of deep frequency selective fades in the composite signal is therefore significantly reduced. Thus, MRC enhances link reliability but it does not increase the nominal system data rate.
TX
RX
Figure 3-2: a MISO system
In a MISO system the transmitter has two or more physically separated antennas and the receiver has one antenna. This allows for transmit diversity (Tx diversity). With Tx diversity, the transmitter sends redundant ‘copies’ of a signal to the receiver. Tx diversity is based on the ‘hope’ that at least one of the copies will be received in a good enough state to allow reliable decoding. One way of realizing Tx diversity is to make use of Space-Time Coding (STC). STC allows the transmitter to use diversity in both the space and time domain. This means that the signal is transmitted through multiple antennas (space) at different points in time (the ‘ST’ in STC). In the simple case of two antennas, a super-position of 2 modulation symbols (e.g. QPSK symbols) is transmitted on both antennas simultaneously. The very same modulation symbols are then transmitted again with a (very) slight delay. In addition, the modulation symbols will be coded, or weighted, differently for the second, slightly delayed, transmission (the ‘C’ in STC). As for Rx diversity, Tx diversity enhances link reliability but does not increase the nominal system data rate. However, with a more reliable data channel it is possible to use less output power (resulting in higher system capacity) or to use less robust channel coding (resulting in higher effective throughput). Apis Technical Training AB LTE - MIMO Copyright © Apis Technical Training AB 2007. All rights reserved.
3-3
3.1.2
MIMO: Multiple Input Multiple Output As might be suspected from the discussion above, a MIMO system uses multiple antennas at transmitter and receiver. Just like the diversity schemes mentioned above the whole point of using multiple antennas is to increase the system capacity (in terms of number of users/cell, coverage, throughput etc) without having to increase the system bandwidth. Using the laws of information theory, it can be shown that the diversity schemes mentioned in section 3.1.1 increase the system capacity logarithmically with the number of antennas. Symmetric MIMO configurations (same number of Tx and Rx antennas) on the other hand, increase the system capacity linearly with the number of antennas. As an example, with four Tx/Rx antennas one can quadruple the system throughput! The cost, of course, is increased complexity.
TX
RX
Figure 3-3: a MIMO system with two transmit and two receive antennas (2x2)
The ‘magic’ of MIMO lies in its ability to take multipath reception, which used to be an unavoidable and undesired by-product of radio propagation, and convert it into an advantage that actually multiplies transmission speed and improves throughput. A MIMO system uses the additional signal paths to transmit more information and recombine the signals on the receiving end. It follows naturally that the diversity modes, mentioned in section 3.1.1, as well as ‘true’ MIMO mode can be used in a system with multiple Tx and Rx antennas. For MIMO, mathematical algorithms are used in order to spread the user data across multiple transmitting antennas. The signals transmitted are defined in 3 dimensions: time, frequency and space. At the receiver, the different signals from each antenna must be identified and separately decoded during the recombination process. Hence, the (mathematical) technique of separating out different paths on the radio link is what allows a MIMO system to transmit multiple signals at the same time on the same frequency, in effect multiplying the capacity of the channel with the number of antennas.
Apis Technical Training AB LTE - MIMO Copyright © Apis Technical Training AB 2007. All rights reserved.
3-4
3.2
MIMO Techniques
3.2.1
Spatial Multiplexing (SM) Spatial Multiplexed (SM) MIMO systems increase spectral efficiency by utilizing signal processing algorithms to exploit multipath propagation on the MIMO communication link. Independent data streams, using the same time-frequency resource, are sent over different transmit antennae. The receiver is able to separate the multiple data streams by using (known) channel information about each propagation path. The multiple streams of a SM transmission must also be orthogonal to each other. Catastrophic interference would follow otherwise. The orthogonality is achieved by multiplying the transmitted streams with a so-called linear precoding matrix. One should not confuse spatial multiplexing with spatial diversity (which is not ‘true’ MIMO). The purpose of spatial diversity is to increase the diversity order of a link to mitigate fading by coding a signal across space and time so that a receiver could receive the replicas of the signal and combine those received signals constructively. Spatial multiplexing transmits not replicas of the same signal but different signals. SM achieves higher data rates by re-using the same frequency resources over multiple spatial dimensions, in effect creating (spatially separate) parallel channels ‘for free’ on the same bandwidth. For example, using two transmit and receive antennas one could transmit two separate data streams, each with data rate 5 Mb/s, resulting in a data rate of 10 Mb/s to the same terminal. At the receiver, multiple antennas are required to demodulate the data streams based on their spatial characteristics (the number of receive antennas is equal to the possible number of separate data streams).
3.2.2
Space-Time Coding (STC) Space-Time Coding (STC) was mentioned above, in section 3.1.1, as a way of realizing Tx diversity. The same theoretical framework is used also for true MIMO operation. STC-MIMO systems provide diversity gain to combat unwanted multipath effects. As already mentioned, time-delayed copies of the same signal, coded differently, are sent over the transmit antennae. The use of multiple antennas at both ends of the link creates additional independently faded signal paths, thereby increasing the maximum diversity gain that can be achieved. Space-Time Codes are split into two main types: trellis codes and (less complex) block codes. Depending on the specific STC method used, the receiver may or may not have to be aware of the characteristics of the channel in order to detect the multiple data streams properly. The use of SM or STC in MIMO is not mutually exclusive, some systems allows dynamic switching between the two modes.
Apis Technical Training AB LTE - MIMO Copyright © Apis Technical Training AB 2007. All rights reserved.
3-5
3.2.3
MU-MIMO and SU-MIMO MIMO transmission can be divided into multi-user and single-user MIMO (MU-MIMO and SU-MIMO). The difference between the two is that in SU-MIMO all the streams carry data to/from the same user while in the case of MU-MIMO the data for different users is multiplexed onto a single time-frequency resource. Hence, SU-MIMO is used either to increase the reliability of the channel (i.e. diversity) or to increase the throughput to a single user in a multiplicative manner while MU-MIMO can be seen as yet another way of multiplexing data to/from different users. In the case of, say, a 2x2 antenna MU-MIMO configuration, two mobile terminals can transmit/receive their data streams simultaneously using the same physical radio resource. Clearly, MIMO is a very powerful way of serving more users without increasing the system bandwidth. Obviously, some form of ‘stream identifier’ is needed in both cases. For SU-MIMO, the receiver must be able to separate one antenna stream from the other(s) in order to t o perfom efficient combining. In the SU-MIMO case, this is achieved by adding a code layer to the transmission- in effect using a basic Code Division Multiplexing (CDM) scheme. For MU-MIMO, the transmitter (mobile station) must be able to ‘tag’ its stream(s), allowing the reciever (base station) to figure out who was transmitting. This is done using a mobile specific reference signal that is transmitted together with the actual data. The use of transmitter specific codes and reference signals does not only allow the receiver to figure out who was transmitting. It also enables accurate channel estimation, which is crucial in MIMO systems.
3.2.4
MIMO and OFDM OFDM is particularly well matched to MIMO applications since it is very tolerant to the effects of multipath fading (remember that MIMO actually creates and relies upon multipath signals). OFDM in the downlink direction can be designed to provide good support for MIMO technologies, as mechanisms such as frequency domain scheduling allow for creating signal-to-noise conditions where MIMO works much better compared to more uniform interference conditions. In short, it can be said that the (mathematical) MIMO model is based on a narrow band non-frequency selective channel, which is precisely what OFDM (with a proper guard interval) offers.
Apis Technical Training AB LTE - MIMO Copyright © Apis Technical Training AB 2007. All rights reserved.
3-6
3.3
MIMO for E-UTRA All of the previously mentioned MIMO techniques (SM, STC, SU-MIMO, MU-MIMO) are supported in E-UTRA. However, the current specification is a bit ambiguous as to what must be supported and to what may be supported. For sure, MIMO is seen as an integral part of the E-UTRA specification, and if an E-UTRAN network supports MIMO there are very clear rules in the standard as how to implement it in terms of precoding matrices, layer mapping etcetera. Furthermore, it is perfectly clear in the standard that the expected baseline configuration shall be two transmit antennas at the eNB and two receive antennas at the UE (allowing also Tx diversity as described earlier). The apparent ‘ambiguity’ in the standard is certainly not due to uncertainty or lack of knowledge, but rather due to preparedness for the future. The modern concept of MIMO is a relatively young research field where significant ‘jumps’ in knowledge could happen at any time. The goal of the standardization process, regarding MIMO, is therefore to incorporate it as an (expected) possibility rather than as an absolute demand. Care is taken not to standardise the use of MIMO to the smallest detail since that might make it difficult, or impossible, to incorporate future improvements of multi-antenna algorithms in an efficient manner. As already stated above the (expected) baseline antenna configuration in E-UTRA for MIMO (and/or Tx diversity) is two transmit antennas at the eNB and two receive antennas at the UE. Even higher-order downlink MIMO and antenna diversity (four TX and two or four RX antennas) will also be supported in the first (R8) release of LTE. The possible/allowed MIMO modes of operation at the eNB are (at the time of writing): Spatial Multiplexing , for one or more* UEs (SU/MU-MIMO) Beamforming (not ‘true’ MIMO) Single-stream Tx diversity (not ‘true’ MIMO). • • •
*) The Spatial Multiplexing of data streams for different UEs using the same timefrequency resource is, in the standard, denoted as Spatial Division Multiple Access (SDMA) or Multi-User MIMO (MU-MIMO)
The LTE standard allows (semi-static) switching between the SU-MIMO and MU-MIMO modes on a per UE basis. Both SU- and MU-MIMO in LTE uses fixed codebooks with precoding matrices that are known to eNB and UE. The UE reports the desired precoding matrix to use, but there is no requirement for the eNB to actually use this value. As a consequence, the precoding matrix selected by the eNB must be signalled to the UE. The MIMO mode that can be used is, of course, restricted by the UE capability, e.g. the number of UE receive antennas, and is determined taking into account the slow channel variation. The selected MIMO mode is adapted slowly (e.g. set at the beginning of a data session or adapted with a period of several 100ms) in order to reduce the feedback control signalling required to support MIMO mode adaptation. Apis Technical Training AB LTE - MIMO Copyright © Apis Technical Training AB 2007. All rights reserved.
3-7
It should be noted that for lower data rates it is more efficient to transmit using a single stream rather than with spatial multiplexing. It can be shown that for a given low rate and a given total transmit power single stream transmission achieves a lower frame error rate. Therefore, MIMO for LTE will most probably use single stream transmission (perhaps using Tx diversity) for lower data rates and spatial multiplexing for the higher data rates. The “crossover point” at which it becomes more efficient to transmit with spatial multiplexing rather than spatial diversity depends on many factors, the number of receive antennas at the UE being one and the distance between transmitter and receiver being another. In general it can be said that SM is most useful when the distance between the transmitter and the receiver is relatively small.
Apis Technical Training AB LTE - MIMO Copyright © Apis Technical Training AB 2007. All rights reserved.
3-8
3.4
References 25.912 36.211 36.300
Feasibility study for E-UTRA and E-UTRAN E-UTRA; Physical channels and modulation E-UTRA E-UTRAN; overall description; Stage 2
Apis Technical Training AB LTE - MIMO Copyright © Apis Technical Training AB 2007. All rights reserved.
3-9
4
E-UTRA Physical Layer 4.1 INTRODUCTION ...................... .................................. ....................... ....................... ................. ..... 4-2 4.2 RADIO FRAME STRUCTURE ....................... ................................... ..................... ......... 4-3 4.2.1
Frame Type 1...........................................................................4-3
4.2.2
Frame Type 2...........................................................................4-4
4.3 CHANNEL ARCHITECTURE ....................... ................................... ....................... ...........4-5 4-5 4.3.1
Logical Channels......................................................................4-5
4.3.2
Transport Channels..................................................................4-6 Channels..................................................................4-6
4.3.3
Physical Channels....................................................................4-8 Channels....................................................................4-8
4.3.4
Physical Signals Signals .......................................................................4-9
4.4 LAYER 1 PROCESSING CHAIN ...................... ................................. ................. ...... 4-11 4.4.1
Downlink Shared Channel (DL-SCH).....................................4-11 (DL-SCH).....................................4-11
4.4.2
Uplink Shared Channel Channel (UL-SCH) (UL-SCH) .........................................4-15
4.5 RESOURCE MAPPING ...................... .................................. ....................... ................. ...... 4-16 4.5.1
Resource Definitions Definitions ..............................................................4-16
4.5.2
Downlink Subframe Subframe ................................................................4-17
4.5.3
Downlink Cell Search Pattern ................................................4-19
4.5.4
Uplink Subframe: Subframe: PUSCH ......................................................4-19
4.5.5
Uplink Subframe: PUCCH......................................................4-20
4.6 REFERENCES ...................... ................................. ....................... ....................... ................. ...... 4-21
Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-1
4.1
Introduction The E-UTRA physical layer (PHY) offers a highly efficient means of conveying data and control information between the eNodeB and the UE. The E-UTRA PHY employs some advanced technologies that are quite new to cellular applications. These include Orthogonal Frequency Division Multiplexing (OFDM) and Multiple Input Multiple Output (MIMO) data transmission, as described in chapters 2 and 3. On the other hand, the LTE standardisation project aims at reusing legacy solutions wherever possible. A reader who is familiar with the UTRAN channel and protocol architecture will therefore feel quite ‘at home’ with the E-UTRAN channel and protocol architecture. The LTE standardisation project also aims at reducing the overall system complexity, resulting in a simplified layered architecture as compared to UTRAN. The E-UTRA specifications describe both Frequency Division Duplex (FDD) and Time Division Duplex (TDD) to separate UL and DL traffic. The overall channel architecture, layer 1 processing chain and resource mapping is the same for both. Thus, the content in this chapter pertains to both FDD and TDD, unless otherwise stated. (The expected market preferences dictate that the majority of deployed systems will be FDD.) The generic radio frame structure (‘frame Type 1’) and the TDD specific radio frame structure (‘frame Type 2’) is described in section 4.2. The EUTRA channel architecture, focusing on the physical channels and physical signals, is described in section 4.3. The associated layer 2 and layer 3 protocol architecture is dealt with separately in chapter 5. The layer 1 processing chain for the uplink and downlink data channels is described in section 4.4. Section 4.5 deals with the mapping of uplink and downlink data and control channels onto 2-dimensional time-frequency radio resources.
Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-2
4.2
Radio Frame Structure
4.2.1
Frame Type 1 1 Radio Frame = 10ms #0
#1
#2
#3
#4
#5
#6
slot 0
1 slot = 0.5ms
#7
#18
#19
slot 19
1 subframe = 1ms
Figure 4-1: E-UTRA frame Type 1
All bandwidth options have the same basic Transmission Time Interval (TTI) of 1ms. As shown in figure 4-1, the E-UTRA radio frames are 10 ms in duration, divided into 10 sub-frames of 1ms duration. Thus, the subframe length coincides with the TTI. Each subframe is further divided into two slots, each of 0.5ms duration. As mentioned earlier, the downlink transmission scheme is based on conventional OFDM with cyclic prefix and the uplink transmission scheme is based on SC-FDMA with cyclic prefix. Both downlink and uplink use the same cyclic prefix lengths and the same sub-carrier spacing of 15 kHz. In addition there is also a reduced sub-carrier spacing, 7.5 kHz, for MBMS-dedicated cells. In the case of 15 kHz sub-carrier spacing there are two cyclic prefix lengths, corresponding to 7 and 6 OFDM/SC-FDMA symbols per slot respectively: •
Normal cyclic prefix: TCP = 160×Ts (symbol #0) and T CP = 144×Ts (symbol #1 to #6). The slightly longer CP in the first symbol is in order to preserve the 0.5ms slot timing.
•
Extended cyclic prefix: TCP-e = 512×Ts (all symbols). The extended CP is intended for large cells, where larger delay spreads for multipath echoes are to be expected.
The parameter T s above is called the ‘basic time unit’ and is defined as being Ts = 1/ (2048 × Δf) seconds, where Δf is the sub-carrier spacing. The length of Ts corresponds to the 30.72 MHz sample clock for the 2048-point FFT used with the 20 MHz system bandwidth. In case of 7.5 kHz sub-carrier spacing there is only a single cyclic prefix length, TCP-low = 1024×Ts, corresponding to 3 OFDM symbols per slot. The generic frame Type 1 can also be used for TDD operation in unpaired spectrum. DL/UL switching points within the frame are then generated by not transmitting in certain symbols (creating a guard period between uplink and downlink transmissions in different sub-frames). Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-3
4.2.2
Frame Type 2 1 Half-frame = 5 ms 1 Radio Frame #0
#1
DL
UL
#2
#3
#4
#5
#6
Uplink Pilot Timeslot Guard Period
#0
#5
#6
1 Slot = Guard Intervals
1 Subframe
Downlink Pilot Timeslot
Figure 4-2: E-UTRA frame Type 1
Frame structure Type 2 is only applicable to TDD, with the sole purpose of being backwards compatible with the 1.28Mcps TDD option in UMTS. 1.28Mcps TDD is the Chinese 3G standard, also known as Low Chip-rate TDD (LCR-TDD) or Time Division Synchronous Code Division Multiple Access (TD-SCDMA). Each 10ms radio frame consists of two half-frames of length 5ms each. The structure of each half-frame in a radio frame is identical. Each halfframe consists of seven slots and three special fields: the downlink pilot timeslot (DwPTS), the guard period (GP) and the uplink pilot timeslot (UpPTS). A subframe is defined as one slot. This frame structure is identical to the one used for TD-SCDMA. Subframe 0 and DwPTS are always reserved for downlink transmission and UpPTS and subframe 1 are always reserved for uplink transmission. For frame structure Type 2 the CP length in the downlink is T CP = 224×Ts (normal CP) and TCP-e = 512×Ts (extended CP) corresponding to 9 and 8 OFDM symbols per slot respectively. For the uplink the situation is slightly less straightforward when it comes to CP lengths. There are several CP lengths used within each slot, depending on the size of the allocated uplink resource and the index of the SC-FDMA symbol within a slot. The normal CP length is 192, 204, 224, 320, 1024 and 2048 ×Ts, corresponding to 9 SC-FDMA symbols. The extended CP length is 423, 456, 472, 560, 1024 and 2048 ×Ts, corresponding to 8 SC-FDMA symbols.
Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-4
Control Channels Broadcast Control Channel ( BCCH ). Downlink channel for broadcasting BCCH ). system information. BCCH is mapped onto the BCH and DL-SCH transport channels. PCCH ). Paging Control Channel ( PCCH ). Downlink channel that carries paging information. Always mapped onto the PCH transport channel.
Common Control Channel (CCCH ). ). This is a bi-directional channel for transmitting initial RRC control signalling between the UE and eNodeB. The CCCH logical channel is always mapped onto the UL/DL-SCH transport channels. Dedicated Control Channel DCCH ). Point-to-point bi-directional channel (DCCH ). for sending dedicated RRC control signalling between the UE and the eNodeB. This channel is always mapped onto the UL/DL-SCH transport channels. Multicast Control Channel MCCH ( , optional). A point-to-multipoint downlink only channel used for transmitting Multimedia Broadcast Multicast Service (MBMS) control information from the network to the UE. This channel is only used by UEs that receive MBMS transmissions. The MCCH is mapped to the MCH transport channel in case of an MBMS-dedicated cell or a cell taking part in Single Frequency Network (SFN) transmission. For mixed traffic cells the MCCH is mapped onto the DL-SCH transport channel.
Traffic Channels Dedicated Traffic Channel DTCH ( ). ). Point-to-point channel dedicated to one UE (uplink or downlink or both) for transmission of user data. Always mapped onto the UL/DL-SCH transport channels. Multicast Traffic Channel MTCH ( , optional). A point-to-multipoint downlink only channel for transmission of multimedia traffic (e.g. mobile TV) from the network to the UE. This channel is only used by UEs that receive MBMS transmissions. The MTCH is mapped to the MCH transport channel in case of an MBMS-dedicated cell or a cell taking part in Single Frequency Network (SFN) transmission. For mixed traffic cells the MTCH is mapped onto the DL-SCH transport channel.
4.3.2
Transport Channels Transport channels are offered from PHY to MAC for signalling or data transport. Different transport channels are defined by how and with what characteristics the information is transmitted on the physical layer. Information on transport channels is delivered to/from the physical layer in the form of Transport Blocks (TB).
Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-6
One or two Transport Blocks are delivered per Transmission Time Interval (TTI). The TTI length selected for E-UTRA is 1ms for most transport channels. A Transport Format (TF) is a combination of TB size (in bits), TTI length and layer 1 channel coding and modulation selected for a given transmission.
Downlink Transport Channels Broadcast Channel BCH ). Carries part of the System Information (SI) in (BCH ). a cell. The SI transmitted over BCH is contained in the Master I nformation Block (MIB) that carries information such as system bandwidth, number of eNodeB antennas for MIMO operation and the transmit power used for downlink reference signals (Reference Signal Transmit Power, RSTP). The BCH is mapped onto the PBCH physical channel.
The MIB also carries the scheduling information for Scheduling Unit 1 (SU-1). SU-1 contains the most often repeated non-BCH SI and is mapped onto the DL-SCH. SU-1 contains the PLMN id, Tracking Area Code (TAC) and the cell id. It may also contain scheduling information for additional SI (i.e. scheduling for SU-2 etc). Paging Channel ( PCH ). The PCH carries paging messages from eNodeB PCH ). to the UE (or group of UEs). The PCH is mapped onto the same physical resource as the DL-SCH. Downlink Shared Channel DL-SCH ). This is the main downlink resource (DL-SCH ). in E-UTRA. It carries data (DTCH or MTCH) and signalling (BCCH, CCCH, DCCH and MCCH). The DL-SCH uses hybrid-ARQ (HARQ), channel dependent packet scheduling and adaptive modulation and coding. The DL-SCH is mapped onto the PDSCH physical channel. Multicast Channel ( MCH ). Carries MBMS data and control information in MCH ). case of an MBMS-dedicated cell or a cell taking part in SFN transmission. The MCH is mapped onto the PMCH physical channel.
Uplink Transport Channels Random Access Channel RACH ( ). ). Uplink channel used to carry control information from the UE to the eNodeB. The RACH is used for initial access, when the UE is not known in the eNodeB. It is also used when a known UE wishes to transmit on the PUSCH or PUCCH, but does not have a valid uplink grant (or when the last assigned timing advance value has expired). The corresponding physical channel is the PRACH. Uplink Shared Channel (UL-SCH ). ). This is the main uplink resource in EUTRA. It carries data (DTCH) and signalling (CCCH and DCCH). The UL-SCH uses hybrid-ARQ (HARQ), ( HARQ), channel dependent packet scheduling and adaptive modulation and coding. The UL-SCH is mapped onto the PUSCH physical channel.
Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-7
4.3.3
Physical Channels The E-UTRA specifications define a physical channel as “a set of resource elements carrying information originating from higher layers” (please refer to section 4.5 for an explanation of the ‘resource element’ concept).
Downlink Physical Channels PBCH ). Physical Broadcast Channel ( PBCH ). This channel carries the MIB from the BCH transport channel. The PBCH uses a TTI of 40ms.
Physical Downlink Shared Channel ( PDSCH ). This is the main downlink PDSCH ). radio resource in a cell, carrying data and/or higher layer signalling. The PDSCH is allocated to different UEs on a per-TTI basis (i.e. every 1ms). The channel coding, modulation and sub-carrier allocation is dynamic. Since the PDSCH is a shared resource, and since the Transport Format used is dynamic, all downlink transmissions must be explicitly addressed to the receiving UE. This is done on the PDCCH. Physical Downlink Control Channel PDCCH ). The downlink control (PDCCH ). channel is used for indications of downlink transmission on PDSCH as well as for allocation of uplink resources on PUSCH/PUCCH. The PDCCH signalling is located in the first 1-3 OFDM symbols in each 1ms long sub-frame. It consists of: UE identity, Transport Format, downlink resource allocation and hybrid-ARQ information related to DL-SCH (and PCH). In addition it contains uplink grant, Transport Format and transmit power commands for uplink transmissions on the PUSCH or PUCCH. Transmission of control signalling from these groups is mutually independent, e.g. an uplink grant can be transmitted to a UE regardless of whether the same UE is receiving downlink data or not.
Multiple physical downlink control channels are supported and a UE monitors a set of control channels. Control information for DL-SCH on the PDCCH relates to downlink data transmission in the same subframe. Control information for UL-SCH on the PDCCH relates to uplink data transmission in a future subframe. Exactly what is meant by ‘future subframe’ has not been decided at the time of writing but will probably be in the order of 3-4 sub-frames after reception of the PDCCH. The PDCCH can be transmitted with 4 different formats, which may change on a per sub-frame basis. This necessitates the use of the PCFICH channel. Physical Control Format Indicator Channel PCFICH ). The sole purpose (PCFICH ). of this physical channel is to indicate the format used for the PDCCH transmission in the same sub-frame. Physical HARQ Indicator Channel ( PHICH ). The purpose of this channel PHICH ). is to transmit ACK/NACKs related to uplink transmissions on the PUSCH. Thus, each PHICH is addressed to a single UE at a time. There is an implicit relation between the uplink resource (sub-carriers) used for data transmission and the downlink resource used by the PHICH. Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-8
The exact timing relationship between uplink data transmission on PUSCH and downlink ACK/NACK on PHICH has not been decided at the time of writing but 2-4ms seem to be reasonable. Physical Multicast Channel ( PMCH ). This channel carries MBMS data PMCH ). and control in case of an MBMS-dedicated cell or SFN transmission.
Uplink Physical Channels PUSCH ). Physical Uplink Shared Channel ( PUSCH ). This is the main uplink radio resource in a cell, carrying data and/or higher layer signalling. The PUSCH is allocated to different UEs on a per-TTI basis (i.e. every 1ms). The channel coding, modulation and sub-carrier allocation is dynamic. Since the PUSCH is a shared resource, and since the Transport Format used is dynamic, all uplink transmissions must be explicitly allocated to a given UE. This is done on the PDCCH as mentioned above. (PUCCH ). Physical Uplink Control Channel PUCCH ). The PUCCH conveys uplink control information in the form of channel quality indicator (CQI), uplink scheduling requests and ACK/NACKs for data received on the PDSCH. This channel is never transmitted simultaneously with PUSCH data, meaning that the ‘PUCCH’ control information is instead transmitted on the PUSCH if an uplink grant for data transmission exists in the UE. It can be assumed that the options for CQI reporting will allow both periodic and ‘on-demand’ reporting. The exact content and meaning of the CQI values as well as the timing between downlink data and uplink ACK/NACK has not been decided at the time of writing.
Physical Random Access Channel ( PRACH ). The PRACH carries the PRACH ). random access preambles (see below) during the random access procedure.
4.3.4
Physical Signals The E-UTRA specifications define a physical channel as “a set of resource elements carrying information originating from higher layers”. Similarly, a physical signal is defined as “a set of resource elements not carrying information originating from higher layers”. Hence they constitute “pure” layer 1 information, in the sense that they originate from layer 1 on the transmitting side and are never visible from higher protocol layers on the receiving side.
Downlink Physical Signals There are two types of downlink physical signals: Reference Signals (RS) and Synchronisation Signals (SS). The Reference Signals are used by the UE for channel estimation purposes (to determine the so-called channel impulse response, CIR). A different RS is transmitted from each antenna port to facilitate MIMO and/or Tx diversity operation. The UE must get an accurate CIR from each transmitting antenna. Therefore, when a reference signal is transmitted from one antenna port the other antenna ports in the cell are idle. Reference signals are sent on every sixth sub-carrier and CIR Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-9
estimates for sub-carriers that do not carry reference signals are computed via interpolation. The exact mapping of the RSs can be seen in figure 4-8 in section 4.5.2 below. Reference Signals are generated as the product of an orthogonal sequence and a pseudo-random sequence. These sequences are standardised and hence known to the UE. From system information (the parameter RSTP mentioned earlier) the UE also knows the output power used by the eNodeB for RS transmission. Specified Reference Signals are assigned to each cell within a network. There are two types of Synchronization Signals: the Primary SS and the Secondary SS. The SSs convey network timing information and are used by the UE during the cell search procedure (e.g. after power-on or cell reselection). The Primary SS provides the UE with slot synchronisation and the Secondary SS provides frame synchronisation. The combination of Primary and Secondary SS also act as a cell-specific identifier called the Physical Cell identity. Overall there are 510 unique sequences possible, meaning that the sequences are reused if the system consists of more than 510 cells. Synchronization Signals use the same type of pseudo-random orthogonal sequences as the Reference Signals.
Uplink Physical Signals Three different uplink physical signals are defined: the Demodulation RS, the Sounding RS and the Random Access Preamble . All uplink physical signals discussed are derived from predefined so-called Zadhoff-Chu sequences. The Demodulation RS, as the name suggests, is used by eNodeB for coherent demodulation of uplink transmissions (i.e. the RS is used for channel estimation). The Demodulation RS is always sent by the UE as part of each PUSCH or PUCCH transmission. t ransmission. The Sounding RS is used (when needed) to facilitate frequency dependent scheduling by the eNodeB. By ordering the UE (or group of UEs) to transmit a Sounding RS using the full system bandwidth (or a subset thereof) the eNodeB can estimate which sub-carriers that are best suited, from a radio condition point of view, for the UEs uplink transmissions. The Sounding RS can also be used in cases where the eNodeB does not receive enough uplink data or control signalling to properly update the timing advance and/or transmit power commands for a given UE (or group of UEs). The Random Access Preamble is used for the random access procedure, when the UE wishes to initiate uplink transmission and does not have a valid uplink grant. The UE transmits a random access burst consisting of a long cyclic prefix, the preamble itself and a guard period during which there is no signal transmitted. The burst is sent on blocks of 72 contiguous sub-carriers allocated for random access by the eNodeB. There are 64 possible preamble sequences per cell. Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-10
From the physical layer perspective, the random access procedure encompasses the transmission of Random Access Preambles until a Random Access Response is received (or the maximum allowed number of preambles have been sent without response). The UE will ramp up its output power for each random access burst transmission until it gets a reply, thus providing a simple initial power control scheme. After this, the eNodeB controls the UE output power by means of a-periodic transmit power commands as part of the uplink grants on the PDCCH.
4.4
Layer 1 Processing Chain The layer 1 processing is different for different transport/physical channels. Most of the physical control channels use convolutional coding with a code rate of 1/3 (meaning 1 bit input to the convolutional coder produces 3 bits output) and only QPSK modulation. The processing of the physical control channels is not treated further in this document. The following sections look closer at the processing of the main transport channels in E-UTRA: DL-SCH and UL-SCH.
4.4.1
Downlink Shared Channel (DL-SCH) DL-SCH
CRC Attachment
Scrambling
Code Block Segmentation
Modulation
Layer Mapping
Turbo Coding R = 1/3
… L1 HARQ Rate Matching
Code Block Concatenation
Precoding
RE Mapper
…
RE Mapper
OFDM Signal Generation
…
OFDM Signal Generation
Figure 4-4: downlink layer 1 processing proces sing chain
Figure 4-4 above shows the processing chain for the DL-SCH transport channel. Data arrives to layer 1 over the DL-SCH transport channel in the form of one or more transport block (MAC PDU) per 1ms TTI. Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-11
CRC Attachment A 24-bit Cyclic Redundancy Check field (CRC) is added for error detection. This information is used in the receiver, after decoding the transport block, to check if the transport block has been correctly decoded or if there are residual bit errors. The receiver transmits a HARQ ACK if the block is successfully decoded or a HARQ NACK if errors are detected.
Code Block Segmentation This step is performed if the number of bits output from the CRC attachment stage is higher than 6144, which is the maximum number of input bits the Turbo coder can handle. The result after segmentation is an integer number of equally sized blocks, possibly with padding bits added.
Channel Coding The error correcting coder selected for DL-SCH is a Parallel Concatenated Convolutional Code (PCCC) with two 8-state constituent encoders and one turbo code internal interleaver (simply called ‘the Turbo coder’ in the following). The coding rate of of the Turbo encoder is 1/3. This is the same Turbo code used as in R6 UMTS, with the exception that the internal interleavers works differently. Turbo codes are error correcting codes with performance coming very close to the Shannon limit, the theoretical limit of maximum information transfer rate over a noisy channel. Thus, Turbo codes make it possible to increase available bandwidth without increasing the power of a transmission, or to decrease the power used to transmit at a certain data rate. The main drawback is the relatively high decoding complexity. The Turbo coder consists of two recursive convolutional coders that each operate (differently) on the input bit sequence. The output from the coder is three sub-blocks of bits: the Systematic bits, which are identical to the input sequence, and the Parity1 bits and Parity2 bits, which are the output sequences from the two internal convolutional coders. The number of input bits divided by the total number of output bits is referred to as the coding rate (R). In general, if the number of Systematic bits is m and the number of Parity1 and Parity2 bits is n/2 respectively, the coding rate becomes m/(m+n). The Turbo coder used in E-UTRA produces an equal number of Systematic, Parity1 and Parity2 bits. Hence, the coding rate becomes R=1/3. Thus, two redundant but different sub-blocks of Parity bits are sent together with the uncoded payload (the Systematic bits). The two sets of Parity bits are used by the Turbo decoder in the receiver to calculate the probability that the payload bits have been decoded correctly. Each of the two convolutional decoders generate a ‘hypothesis’ for the payload. The hypothesis bit-patterns are compared and if they differ the decoders exchange the derived likelihoods they have for each bit in the t he hypotheses.
Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-12
Each decoder incorporates the derived likelihood estimates from the other decoder to generate a new hypothesis for the bits in the payload. Then they compare these new hypotheses. This iterative process continues until the two decoders come up with the same hypothesis for the Systematic bits. The DL-SCH always applies an R=1/3 Turbo code for error correction. However, all bits from the three output sequences (Systematic, Parity1, Parity2) are not always sent. The number of bits from each set that are actually transmitted depends on the applied L1 HARQ rate matching.
L1 HARQ Rate Matching Rate matching (RM) means that bits on a (channel coded) transport channel are repeated or punctured, with the purpose of adapting the number of bits at the output of the channel coder to the total number of bits available on the physical channel. The RM method used in E-UTRA is based on a concept called a circular buffer. The Systematic bits are written into a portion (one third) of a rate matching buffer. The Parity1 and parity2 bits are scrambled and interleaved and then written into i nto the remaining twothirds of the buffer. A subset of all the bits in the buffer are then read out and transmitted. The subset is selected simply by letting an offset define where to start reading consecutive bits in the buffer (e.g. ‘start reading from bit-position 50’). The offset is decided based on the Redundancy Version (RV) selected for the transmission. Thus, the exact set of bits at the output of the HARQ-RM depends on the number of input bits from the Turbo coder, the number of bits to be transmitted and the selected RV. This process makes it easy to select different sets of coded bits from the same Transport Block to be transmitted each time a re-transmission is requested, thus allowing Incremental Redundancy operation - also known as HARQ type-II. A type-II HARQ scheme makes use of the transport blocks that cause retransmission requests (i.e. erroneous transport blocks are not discarded). An erroneous transport block will be stored and later combined with retransmitted version(s) of itself; thereby creating a single combined transport block that is more reliable than any of its constituent parts. The versions of one and the same transport block are produced by selecting different RVs.
Code Block Concatenation If the Code Block Segmentation stage mentioned earlier resulted in more than one code block, they will all have been processed separately up to this point. The Code Block Concatenation stage simply concatenates the Turbo coded and rate matched blocks into a single so-called code word prior to further processing. This stage does not add, remove or alter any bits.
Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-13
Scrambling The bits in the code word are scrambled with a cell specific scrambling sequence prior to modulation.
Modulation Standard QPSK, 16QAM or 64QAM modulation mapping, resulting in complex modulation symbols carrying 2, 4 or 6 coded bits respectively.
Layer Mapping The modulation symbols from one or two (scrambled) code words are mapped onto 1, 2, 3 or 4 antenna ports. Thus, this step is related to MIMO or Tx diversity operation. Basically, a layer corresponds to a spatial multiplexed channel. For E-UTRA the defined configurations are 1x1, 2x2, 3x2 and 4x2 MIMO/diversity. Note that while there are as many as four transmitting antennas (four layers) there are only a maximum of two receivers and thus a maximum of two spatial multiplexed data streams (two code words). For a 1x1 or a 2x2 system there is a simple 1:1 relationship between layers and transmitting antenna ports. However, for a 3x2 and 4x2 system there are still only two spatial multiplexed channels. Therefore, there is redundancy on one or both data streams. The Layer Mapping stage specifies exactly how the extra transmitter antennas are to be employed.
Precoding This step is also related to MIMO or Tx diversity. Precoding is applied to allow the UE to separate the different antenna streams. There are different standardised code books defined for the cases of spatial multiplexing (SUMIMO and MU-MIMO) and Tx diversity. This corresponds to the SpaceTime Coding discussed in chapter 3.
Resource Element Mapping The precoded code words are mapped onto a number of 2-dimensional time-frequency Resource Elements available for the transmission. This step is described in more detail in section 4.5.
OFDM Signal Generation OFDM symbols are created, as described in chapter 2, using the number of sub-carriers allocated for the transmission. A cyclic prefix is appended to each OFDM symbol and the symbols (with CP) are then mapped onto 2 consecutive radio frame slots constituting a subframe. The Resource Element Mapping stage and the OFDM Signal Generation stage takes place separately for each antenna port assigned for the transmission.
Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-14
4.4.2
Uplink Shared Channel (UL-SCH) UL-SCH
CRC Attachment
Code Block Concatenation
Code Block Segmentation
Scrambling
Turbo Coding R = 1/3
Uplink control bits
Transform Precoding
L1 HARQ Rate Matching
Channel Coding
RE Mapper
Data/Control Multiplexing
SC-FDMA Signal Generation
Figure 4-5: uplink layer 1 processing chain
The processing chain for the UL-SCH transport channel is very similar to the one for the DL-SCH. Only differences diff erences are described in the following.
Data/Control Multiplexing Since the PUSCH and the PUCCH physical channels are never transmitted simultaneously, there is instead a possibility to multiplex the ‘PUCCH’ control information with the uplink data transmitted on the PUSCH. The control information is channel coded separately prior to this stage.
Scrambling Scrambling with a UE specific scrambling sequence.
Transform Precoding This is the ‘FFT-spreading’ step as described for the uplink in chapter 2. That is, the modulation symbols are spread over the entire allocated bandwidth, creating a single-carrier signal.
RE Mapping & SC-FDMA Signal Generation The difference to DL-SCH is that there is ever only one antenna port used for the uplink (no need for layer mapping and precoding). Furthermore, the last step will create SC-FDMA symbols instead of OFDM symbols. Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-15
4.5
Resource Mapping Note: the mapping of physical channels to resource elements described in this section assumes the use of frame Type 1 and normal cyclic prefix.
4.5.1
Resource Definitions #0
#1
#2
#3
#4
#5
#18
#19
Control Channel Element Physical Resource Block
s r e i r r a c b u S
Resource Element
OFDM symbols
Figure 4-6: Physical Resource Block (PRB)
The downlink and uplink resources assigned to UEs for the DL-SCH and UL-SCH transmission are referred to as Physical Resource Blocks (PRB). A PRB consists of 12 consecutive sub-carriers in the frequency domain. In the time domain a PRB consists of Nsymb OFDM (or SC-FDMA) symbols, where Nsymb is the number of symbols during a slot (7 in this case). The number of resource blocks, N RB, that may be assigned to the UE can range from N RB-min = 6 to N RB-max = 100. The 2-dimensional time-frequency resource can be represented as a resource grid as depicted in Figure 4-6. Each little box within the grid represents a single sub-carrier for one symbol period and is referred to as a Resource Element (RE). Figure 4-6 shows the resource grid for frame Type 1 using the normal cyclic prefix length, resulting in each PRB containing 84 REs. Note that in MIMO operation there is a resource grid for each transmitting antenna. Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-16
The downlink control channels use a slightly different concept. They are formed by aggregation of Control Channel Elements (CCE). Each CCE is, in turn, an aggregation of 1, 2, 4 or 8 mini-CCEs, where each mini-CCE consists of 4 REs. Thus, a CCE varies in size between 4 and 32 REs. Different code rates (i.e. different levels of robustness) for the PDCCH are realized by aggregating different numbers of CCEs or mini-CCEs. Because multiple CCEs can be combined to reduce the effective coding rate the UE’s PDCCH assignment can be based on the channel quality information reported (CQI), increasing the chance that the PDCCH can be correctly decoded even for UEs experiencing bad channel conditions. 1, 2, 4 and 8 control channel elements can be aggregated to yield approximate code rates of 2/3, 1/3, 1/6 and 1/12 for f or the PDCCH.
4.5.2
Downlink Subframe #0
#1
#2
#3
#4
#5
#18
#19
s r e i r r a c b u S 2 1
Slot #4
Slot #5
RE for PDCCH, PCFICH and PHICH RE for PDSCH RE for antenna RS
Figure 4-7: subframe with PDSCH, PDCCH, PDCCH, PCFICH and PHICH
The resource grid for a downlink subframe is illustrated in figure 4-7 (frame type 1 using normal cyclic prefix length and one antenna port). The grid is shared between all downlink physical channels and signals. Only two PRBs are shown in figure 4-7.
Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-17
Downlink Physical Control Channels The PHICH channel (ACK/NACK for uplink transmissions) is located in st rd the 1 or 3 OFDM symbol of the subframe and always occupies 3 miniCCEs (12 REs). The resources used for the PHICH are configured on a semi-static basis, i.e. the UE knows where to look for it (in terms of REs). The PCFICH channel signals the number of OFDM symbols (1-3) used for PDCCH signaling in each subframe. The PCFICH is transmitted in the first OFDM symbol of the subframe and occupies 4 mini-CCEs (16 REs). The PDCCH is mapped onto the remaining REs in the 1-3 first OFDM symbols in the first slot of each subframe. This enables support for microsleep, i.e. the UE can wake up within one symbol and, seeing no assignment, go back to sleep within one symbol for battery life savings and reduced buffering. It also allows reception of downlink data, if the UE finds an assignment, in the very same subframe, thus reducing latencies.
Downlink Shared Channel, DL-SCH The DL-SCH uses the REs ‘remaining’ after allocation of PCFICH, PDCCH, PHICH and downlink RSs. Thus, all the 84 REs in each Physical Resource Block cannot be used for actual data transmission!
Downlink Reference Signals Figure 4-7 above shows the antenna RSs (black REs) in case only one antenna port is used. Figure 4-8 below shows the case when 2 ports are used. Please note that the two grids are transmitted simultaneously on the two antenna ports. Slot # i
Slot # i+1
R1 s r e i r r a c b u S 2 1
R1
R1
Slot # i
R2
R1
R1
R1 R1 RS for antenna port 1
Slot # i+1
R2
R2
R1
R2
R1
R2
R2
R2
REs that cannot be used
R2
R2 RS for antenna port 2
Figure 4-8: mapping of antenna reference signals (2 antenna ports)
Antenna RSs are transmitted on equally spaced sub carriers within the first and third from-last OFDM symbol of each slot. In order to successfully receive a MIMO transmission the UE must determine the channel impulse response for each transmitting antenna, as already mentioned. Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-18
In E-UTRA the channel impulse responses are determined by sequentially transmitting known reference signals from each transmitting antenna. Note th that the RSs are transmitted on every 6 sub-carrier in a repeated, symmetric, time-frequency pattern. Note also that while one transmitter antenna is sending its Reference Signal, the other antenna is idle.
4.5.3
Downlink Cell Search Pattern Slot #0 36
#0
#1
#2
#3
Slot #1 #4
#5
#6
#0
#1
#2
#3
#4
#5
#6
DC
36
Antenna Antenna RS RS (1 (1 port) port)
Secondary SS
REs REs ffor or P PBCH BCH
Primary SS
Figure 4-9: primary and secondary synchronisation signals and PBCH
During cell search the UE needs to find the Primary and Secondary Synchronisation Signals as well as the Physical Broadcast Channel (PBCH). These are all mapped around the center sub-carrier in the system. This center sub-carrier is called the Direct Current (DC) sub-carrier and never carries any information. The Primary and Secondary SS are transmitted in slot 0 and 10 on 64 subcarriers centered around the DC sub-carrier. The Secondary SS occupies th th the 6 OFDM symbol and the Primary SS occupies the 7 . The PBCH is th transmitted on 72 sub-carriers centered around the DC sub-carrier in the 4 th st nd and 5 OFDM symbol in slot 0 and the 1 and 2 OFDM symbol in slot 1, over 4 consecutive radio frames. Slots 0 and 1 are shown in fig 4-9, including Reference Signals for one antenna port.
4.5.4
Uplink Subframe: PUSCH For the uplink the PRBs are shared between the PUSCH, for actual data transmission, and the uplink Demodulation RSs. The Demodulation RS is transmitted using the same sub-carriers as those assigned for the associated th PUSCH transmission and occupies the 4 SC-FDMA symbol in each slot.
Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-19
Note that figure 4-10 below only shows two PRBs, while the minimum UE allocation is 6 PRBs (i.e. 36 sub-carriers). #0
#1
#2
#3
#4
#5
#18
#19
s r e i r r a c b u S 2 1
Slot #4
Slot #5
RE for PUSCH
RE for Demodulation RS
Figure 4-10: PUSCH subframe with demodulation s ignal
4.5.5
Uplink Subframe: PUCCH #0
#1
#2
#3
#4
#5
#18
#19
s r e i r r a c b u S 2 1
RE for PUCCH Demodulation RS
1 2 S u b c a r r i e r s
Slot #4
Slot #5
Figure 4-11: PUCCH subframe with demodulation signal
The PUCCH resource is defined by a UE specific code and two consecutive PRBs with frequency hopping at slot boundary. Demodulation rd th th RS occupies the 3 , 4 and 5 SC-FDMA symbol in each slot. Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-20
4.6
References 36.211 36.212 36.213
E-UTRA; Physical channels and modulation E-UTRA; Multiplexing and channel coding E-UTRA; Physical layer procedures
Apis Technical Training AB LTE - Physical Layer Copyright © Apis Technical Training AB 2007. All rights reserved.
4-21
5
E-UTRA Layer 2 & 3 5.1 INTRODUCTION ...................... .................................. ....................... ....................... ................. ..... 5-2 5.2 LAYER 2 & 3 PROTOCOLS ....................... ................................... ....................... ...........5-3 5-3 5.2.1
Non Access Stratum Stratum (NAS) (NAS) .....................................................5-3
5.2.2
Radio Resource Resource Control Control (RRC) (RRC) ...............................................5-3
5.2.3
Packet Data Convergence Convergence Protocol Protocol (PDCP) ...........................5-4
5.2.4
Radio Link Control (RLC).........................................................5-6
5.2.5
Medium Access Control (MAC)................................................5-6
5.3 UE STATES (RRC AND NAS) ....................... .................................. ................. ...... 5-8 5.4 PDU FORMATS ....................... ................................... ....................... ....................... ............... ... 5-9 5.5 REFERENCES ...................... ................................. ....................... ....................... ................. ...... 5-11
Apis Technical Training AB LTE - Layer 2 and 3 protocols Copyright © Apis Technical Training AB 2007. All rights reserved.
5-1
5.2
Layer 2 & 3 Protocols
5.2.1
Non Access Stratum (NAS) The Non-access Stratum (NAS) protocols are used between the UE and the Evolved Packet Core (the MME to be more precise) and are therefore, strictly speaking, not E-UTRA protocols. All NAS signalling exchange takes place transparently through the radio access network (i.e. the eNodeB will never interpret these messages). The NAS layer sits ‘on top’ of the RRC layer in the UE (not shown in figure 5-1). All NAS messages are either carried inside, or sent concatenated with, RRC messages when transmitted over the radio interface. There will probably be only two LTE/SAE NAS protocols defined: The EPC Mobility Management protocol (EMM) and the Session Management protocol (SM). The EMM protocol is very similar to the corresponding GSM/UMTS MM and GMM protocols. EMM is responsible for procedures such as: Paging of UE in idle mode. Network Attach, i.e. UE registration in the MME and establishment of basic IP-connectivity. Authentication and NAS key agreement (read more about security in the PDCP section). Tracking Area update (the concept of a ‘Tracking Area’ is fully analogous to the Location/Routing Area concept). SAE-TMSI (S-TMSI) allocation. • •
•
•
•
The SM protocol is responsible for the establishment and management of default and dedicated SAE bearers and is similar to the GPRS/UMTS SM protocol (which is responsible for management of PDP contexts). The SAE bearer and QoS concepts are discussed further in chapter 7.
5.2.2
Radio Resource Control (RRC) The RRC protocol is responsible for all layer 3 control signalling exchange between the UE and the eNodeB. It is expected that the functionality, on the procedure level, will be very similar to the UTRAN RRC protocol. Some RRC procedures are briefly described in the following.
Broadcast of System Information The purpose of System Information (SI) is to provide the UEs with essential parameters needed for communication with E-UTRAN. A few examples of broadcasted information: description of the random access channel, neighbour cell lists, definition of paging groups, periodic update timers, area identities (PLMN, Tracking area, cell). Apis Technical Training AB LTE - Layer 2 and 3 protocols Copyright © Apis Technical Training AB 2007. All rights reserved.
5-3
RRC Connection Establishment This procedure will move the UE from the RRC Idle state to the RRC Connected state (see section 5.3). An RRC Connection is always needed whenever the UE wishes to send/receive control signalling or data to/from the network. The RRC Connection procedure is always co-ordinated with a NAS signalling procedure, such as Attach, resulting in the establishment of basic IP-connectivity for the UE.
Access Stratum (AS) Security Mode Control Procedure used to instruct the UE to begin ciphering and integrity protection of RRC signalling and user plane data. See the PDCP section below for more information about LTE/SAE security.
Management of Radio Bearers Procedures for establishment, reconfiguration and release of user plane radio bearers, including configuration of ARQ (RLC) and HARQ (MAC) operation. It is also possible to configure semi-persistent scheduling rules (e.g. DRX cycles and pre-defined transport formats) for predictable services such as VoIP.
Measurement Control/Reporting Control/Reporting The eNodeB may start, modify or stop a number of measurements in the UE (independently of each other). The measurement reporting can be done periodically or be event triggered. These procedures are used in the RRC Connected state to prepare for handovers.
Handover Control This procedure includes the necessary control signalling to execute hard handovers between eNodeBs or between eNodeB and some other Radio Access Technology (RAT). E-UTRAN will support handover to/from at least GERAN, UTRAN, mobile WiMAX and CDMA2000 systems.
5.2.3
Packet Data Convergence Protocol (PDCP) The PDPC protocol is responsible for header compression (Robust Header Compression, ROHC, as defined in RFC 3095), access stratum security and for in-sequence delivery of user plane packets during handover. Up until recently, there was much debate within the LTE project whether PDCP should be responsible for both User Plane (UP) and Control Plane (CP) security, or only for UP security. It was also debated whether PDCP should reside in the eNodeB or in the SGW. The decision was finally taken to let PDCP reside in the eNodeB and, as a consequence, to be responsible for both UP and CP security.
Apis Technical Training AB LTE - Layer 2 and 3 protocols Copyright © Apis Technical Training AB 2007. All rights reserved.
5-4
The LTE/SAE system uses a security key hierarchy (figure 5-2) with multiple levels. The base keys on the top level (Ciphering Key, CK, and Integrity Key, IK) are only visible to the UE and the home network domain databases (HSS/AuC). On the next level there is a so-called Access Security Management Entity (ASME) key, which is only visible to the UE and the visited MME (the ‘ASME’ node in figure 5-2 is the MME in case of the Evolved Packet Core). The ASME key is derived from the base CK/IK pair and passed from HSS to the MME. The ASME key is, in turn, used for derivation of the ciphering and integrity keys needed to protect NAS signalling messages (i.e. signalling between UE and MME). USIM / AuC
K
CK, IK
UE / HS HSS S KASME
UE / ASME
KNAS enc
KNAS int
KeNB
UE / MME KeNB-UP-enc
KeNB-RRCeNB-RRC- int
KeNB-RRC-enc
UE / eNB
Figure 5-2: LTE/SAE security key hierarchy (from TR 33.821)
The ASME key is also used for derivation of an eNodeB key. The eNodeB key, in turn, is used for derivation of keys for ciphering and integrity protection of RRC signalling messages and a key for the ciphering of user data over the radio interface (i.e. between UE and eNodeB). This hierarchy allows the keys in the Home domain, the (visited) EPC domain and the Access domain to be cryptographically separate, while still being produced by the same set of Home domain controlled base keys.
Apis Technical Training AB LTE - Layer 2 and 3 protocols Copyright © Apis Technical Training AB 2007. All rights reserved.
5-5
5.2.4
Radio Link Control (RLC) The Radio Link Control (RLC) protocol offers link control over the radio interface for user data and control signalling. One UE may use multiple RLC entities simultaneously, with each entity configured to operate in one of three modes: Acknowledged Mode (AM), Unacknowledged Mode (UM) or Transparent Mode (TM). In AM mode the RLC layer guarantees that all PDUs delivered to the higher layer are received without errors. This guarantee is ensured by means of RLC-level acknowledgements (ACK) and retransmission requests (NACK) as well as through interaction with the MAC level HARQ functionality. ACK/NACK is indicated with Status Reports (RLC Control PDUs) sent from receiver to transmitter. As the term suggests, there are no acknowledgements or retransmissions in the UM mode. Hence delivery of data cannot be guaranteed. Any erroneous RLC PDUs received are either discarded or delivered to higher layers with an indication that they contain errors. The TM mode, finally, is used for minimum protocol overhead (the TM mode PDU has no header). The service the RLC layer provides to the User Plane (PDCP protocol) is called a Radio Bearer (RB). The RLC mode to use is configured with the Radio Bearer Establishment procedure. The service the RLC layer provides to the Control Plane (RRC protocol) is called a Signalling Radio Bearer (SRB). The SRBs are configured during the RRC Connection Establishment procedure. It is expected that only two types of SRBs will be used in E-UTRA: one low priority SRB and one high priority SRB.
5.2.5
Medium Access Control (MAC) The main functions of the MAC protocol are to perform multiplexing of one or several upper layer PDUs onto transport blocks, to perform UL and DL resource allocation through dynamic scheduling and to handle error correction through HARQ operation. The scheduling function manages DL-SCH and UL-SCH radio resources between HARQ entities (i.e. between UEs). The scheduler determines which UE (or group of UEs) to be serviced each 1ms TTI. The exact scheduling algorithm used is implementation dependent but should preferably take into account: Availability of radio resources Data stream priority levels (for each UE) The current channel conditions (for each UE) The amount of data awaiting transmission (for each UE) How long time since UE X was last served Users that have retransmissions pending. •
•
•
•
•
•
Apis Technical Training AB LTE - Layer 2 and 3 protocols Copyright © Apis Technical Training AB 2007. All rights reserved.
5-6
The scheduler selects a proper Modulation and Coding Scheme (MCS) and Redundancy Version (RV) for each scheduled MAC PDU based on each scheduled UEs current channel condition, the retransmission status and, possibly, on the UE capabilities. The RV is used as input to the HARQ layer 1 rate matching function discussed in chapter 4. One HARQ Entity within MAC handles the HARQ functionality for one user. The HARQ protocol selected for E-UTRA is of the ‘Stop-and-Wait’ type (SAW). This means that it is not allowed to transmit a PDU with sequence number ‘N’ until the PDU with sequence number ‘N-1’ is positively acknowledged. acknowledged. Remember that the TTI used in E-UTRA is only 1ms. Each time the UE receives data in a 1ms TTI it must, according to the SAW protocol, send back either an ACK (‘everything OK, please send next PDU’) or a NACK (‘please retransmit the PDU’). The creation and sending of an ACK/NACK takes a certain amount of time. So does the processing of the ACK/NACK in the NodeB. And so does the scheduling of a new re/transmission to this UE. All this is simply impossible to execute before the start of the next 1ms TTI. The consequence is then that it becomes impossible to schedule transmissions in consecutive 1ms TTIs to the same UE, resulting in waste of resources- or at least waste of time. (The same logic holds, of course, for uplink transmissions). The solution is to allow each HARQ Entity to work with several processes simultaneously. When one HARQ process is awaiting ACK/NACK for a transmitted MAC PDU, the scheduler can order transmission of the next MAC PDU from the next HARQ process, that then stops and awaits ACK/NACK, and so on. It is expected that 8 HARQ processes will be sufficient to allow continuous transmission to/from a given UE. Thus, the shortest HARQ round-trip time is expected to be 8ms.
Apis Technical Training AB LTE - Layer 2 and 3 protocols Copyright © Apis Technical Training AB 2007. All rights reserved.
5-7
5.3
UE States (RRC and NAS) L TE IDLE
RRC IDLE
LTE ACTIVE
RRC CONNECTED
LTE DETACHED
Figure 5-3: UE states and state transitions
From a radio resource point of view there are two operational states for the UE: RRC Idle State and RRC Connected State. In the RRC Idle state the UE is unknown in E-UTRAN and will remain so until it requests the establishment of an RRC Connection. Such a request can be triggered by higher protocol layers in the UE (i.e. ( i.e. mobile originating service request) or by the paging procedure (initiated from the EPC). In RRC Idle state the UE moves around in the network and change from one cell to another through the process of cell reselection. It continuously monitors the broadcasted system information and the paging channel. No data/signalling transmission or reception, except paging and system information, is possible in the RRC Idle state. The RRC Connected state allows data or signalling to be sent or received. The UE enters the Connected state through the establishment of an RRC connection. The UE is always allocated a cell specific identifier, the Cell Radio Network Temporary Identity (C-RNTI) when in Connected state. The C-RNTI is, among other things, used for addressing the UE on the downlink resource assignment channel, the PDCCH. UE mobility is network controlled through handovers. The UE may have a DRX cycle configured in order to allow ‘sleep periods’ in-between monitoring the PDCCH. RRC Connection Release brings the UE back to RRC Idle state. The NAS states (EPC related states) are aligned with the RRC states. A UE in RRC Idle state is, from the MMEs point of view, in the NAS state LTE Idle. In this state the UE is registered in the MME and has an IPaddress allocated. Whenever the UE detects a change of Tracking Area it performs a Tracking Area update towards the MME. Apis Technical Training AB LTE - Layer 2 and 3 protocols Copyright © Apis Technical Training AB 2007. All rights reserved.
5-8
Paging or a request from higher layers to transmit uplink data or signalling will cause a transition from LTE Idle to the LTE Active state. In LTE Active state the UE has at least one SAE bearer allocated, allowing uplink or downlink data/signalling transfer to take place. The S-TMSI is used to identify/ address the UE in NAS signalling messages. The UE can never be in LTE Active state without also being in RRC Connected state. Transition from LTE Active to LTE Idle can, for example, be triggered by user inactivity. In the LTE Detached state there is no information known about the UE in the eNodeB or the MME. No data or signalling transfer is possible. This state is left/entered through t hrough the Attach/Detach procedures.
5.4
PDU Formats RRC Message or IP Packet
PDCP PDU
RLC PDU
Seq. No
MAC PDU
LCID1
Seq. No
PDCP SDU
E Length Ind. E
L1
E1
MAC-I
.....
RLC SDU 1
MAC SDU 1
.....
Pad
Figure 5-4: layer 2 and layer 3 PDU formats
Figure 5-4 shows the PDU formats for (from top to bottom) the PDCP, RLC and MAC protocols. The payload of a given protocol is referred to as a Service Data Unit (SDU). PDCP PDUs only carry one SDU while RLC and MAC PDUs may carry multiple SDUs. The PDCP protocol takes as input either an RRC message (CP) or an IP packet (UP). RRC messages are encrypted and integrity protected. The integrity protection results in a Message Authentication Code for Integrity (MAC-I) field being added at the end of the PDCP PDU. User plane packets are encrypted and compressed but never integrity protected. The PDCP protocol also adds a one or two byte long sequence number, unless configured for transparent operation where no sequence number is present. Apis Technical Training AB LTE - Layer 2 and 3 protocols Copyright © Apis Technical Training AB 2007. All rights reserved.
5-9
The RLC protocol takes as input PDCP PDUs. Several PDCP PDUs may be concatenated into one and the same RLC PDU. The RLC protocol may also perform segmentation, meaning that only part of a given PDCP PDU is fitted within one RLC PDU. An RLC sequence number is added for ARQ operation, sequence control and SDU reassembly purposes. One or more length indicator is added to indicate the presence of multiple SDUs, or SDU segments. The presence of the length indicators themselves is indicated with an extension bit (E) following the sequence number and each present length indicator. Thus, the E-bit following the last length indicator will indicate ‘no more length indicator fields present’. The MAC protocol takes as input RLC PDUs. Several RLC PDUs may be concatenated into one and the same MAC PDU. One MAC PDU is the same as one Transport Block. Thus, one and the same Transport Block may carry information from more than one logical channel (figure 5-1). The identity number of the logical channel where a given MAC SDU originated is indicated with the Logical Channel Identity field (LCID). The length (in bits) of each MAC SDU is indicated with the Length field (L). There is one LCID/L pair for each MAC SDU in the payload field. The presence of yet another LCID/L pair is indicated with the extension bit (E) following the previous pair. Thus, the E-bit following the last LCID/L pair will indicate ‘no more LCID/L fields’. Padding may be added if the total length of the LCID/L/E fields and the associated MAC SDUs do not exactly match the number of bits to be transmitted on the assigned physical resource. One or two Transport Blocks per 1ms TTI are delivered to the physical layer for further processing, as described in chapter 4.
Apis Technical Training AB LTE - Layer 2 and 3 protocols Copyright © Apis Technical Training AB 2007. All rights reserved.
5-10
5.5
References 24.801 33.821 36.321 36.322 36.323 36.331
3GPP System Architecture Evolution: CT WG1 aspects (NAS) Rationale and track of security decisions in LTE/SAE E-UTRA; Medium Access Control (MAC) protocol E-UTRA; Radio Link Control (RLC) protocol E-UTRA; Packet Data Convergence Protocol (PDCP) E-UTRA; Radio Resource Control (RRC) protocol
Apis Technical Training AB LTE - Layer 2 and 3 protocols Copyright © Apis Technical Training AB 2007. All rights reserved.
5-11
6
X2 and S1-interface 6.1 INTRODUCTION ...................... .................................. ....................... ....................... ................. ..... 6-2 6.2 THE X2-INTERFACE ....................... .................................. ....................... ..................... ......... 6-3 6.2.1
X2-interface Protocols..............................................................6-3
6.3 THE S1-INTERFACE ....................... .................................. ....................... ..................... ......... 6-5 6.3.1
S1-interface Protocols..............................................................6-5
6.4 SELF-ORGANIZING NETWORKS ...................... ................................. ................. ...... 6-7 6.4.1
Self-configuration Self-configuration .....................................................................6-7
6.4.2
Self-optimization.......................................................................6-8
6.5 HOME ENB ...................... ................................. ....................... ....................... ...................... ...........6-8 6-8 6.5.1
Node Configuratio Configuration n ..................................................................6-9
6.5.2
Access Restriction....................................................................6-9
6.5.3
Mobility ...................................................................................6-10
6.5.4
Security ..................................................................................6-11
6.5.5
QoS and Interference.............................................................6-11
6.6 REFERENCES ...................... ................................. ....................... ....................... ................. ...... 6-12
Apis Technical Training AB LTE - X2 and S1-interface Copyright © Apis Technical Training AB 2007. All rights reserved.
6-1
6.1
Introduction Evolved UTRAN
Evolved Packet Core
S1-MME
eNB
X2-C
MME
X2-U
S11
eNB
SGW S1-U
Figure 6-1: the X2-interface and the S1-interface
The X2-interface connects the eNBs within E-UTRAN together. The X2interface is an IP-based interface and can therefore be seen as a point to multi-point interface (the eNB may communicate with every other eNB). The Control Plane (CP) instance of the X2-interface (X2-C) uses the X2 Application Protocol (X2AP) for control signalling purposes between eNBs. The User Plane (UP) instance of the X2-interface (X2-U) uses the GPRS Tunnelling Protocol- User plane (GTP-U) for user data tunnelling between eNBs. The X2-interface is described in section 6.2. The S1-interface connects the Evolved UTRAN with the Evolved Packet Core (EPC). The termination point for the S1-interface on the E-UTRAN side is the eNB, and on the EPC side the Mobility Management Entity (MME) and the Serving Gateway (SGW). The S1-interface is, like the X2interface, an IP-based point to multi-point interface. The CP instance of the S1-interface (S1-MME) uses the S1 Application Protocol (S1AP) for control signalling purposes between eNB and MME. The UP instance of the S1-interface (S1-U) uses GTP-U for user data tunnelling between eNB and SGW. The S1-interface is described in section 6.3. Section 6.4 deals with network self-organization issues and section 6.5 contains an overview on the t he so-called home eNB. At the time of writing (Oct-07) all specifications pertaining to the X2- and S1-interfaces are all immature drafts. The reader is therefore strongly advised to regularly check for updated versions of the standardisation documents listed at the end of this chapter. Apis Technical Training AB LTE - X2 and S1-interface Copyright © Apis Technical Training AB 2007. All rights reserved.
6-2
6.2
The X2-interface
6.2.1
X2-interface Protocols
eNB eNB
e eNB NB
X2AP
X2AP
SCTP
SCTP
IP
IP
Data Link Layer
Data Link Layer
Physical Layer
Physical Layer X2-C
User Data
User Data
GTP-U
GTP-U
UDP
UDP
IP
IP
Data Link Layer
Data Link Layer
Physical Layer
Physical Layer X2-U
Figure 6-2: X2-interface protocols for CP (top) and UP (bottom)
X2 Application Protocol (X2AP) The X2AP protocol is used for control signalling exchange between eNBs. It supports Common and Dedicated procedures. Common procedures are signalling procedures that do not relate to a specific UE. An example of a Common procedure is Load Indication. Dedicated procedures are signalling procedures that do relate to a specific UE. An example of this is Handover. Currently specified X2AP procedures include: •
Handover. Handover. Initiated by the source eNB by sending a Handover Request message to the target eNB. The target eNB reserves the necessary radio resources and sends a Handover Request Acknowledge message back to the source eNB. The ACK message contains a complete radio interface Handover Command message, to be sent to the UE, and the preferred Target eNB IP-address for data forwarding over X2 during the handover execution phase.
Apis Technical Training AB LTE - X2 and S1-interface Copyright © Apis Technical Training AB 2007. All rights reserved.
6-3
•
•
Release resource. resource. After a successful handover, the Target eNB initiates this procedure to inform the Source eNB that it can now stop data forwarding over X2 and release all resources for this UE. Load Indication. Indication. The purpose of the Load Indication procedure is to transfer an uplink ‘Interference Overload Indication’ between intra-frequency neighboring eNBs for interference coordination purposes. The Overload Indication is sent when the eNB experiences too high interference level on some resource blocks. This procedure is linked to the E-UTRAN Inter-Cell Interference Cancellation (ICIC) function, that allows eNBs to ‘agree’ on what set of OFDM sub-carriers to use at overlapping cell borders.
Stream Control Transmission Protocol (SCTP) The Stream Control Transmission Protocol (SCTP) is used to support the exchange of X2 Application Protocol (X2AP) signalling messages between two eNBs. SCTP is a session-oriented protocol providing connection-oriented, error-free, flow-controlled, in-sequence transfer of signalling messages over IP. It is in many respects similar to TCP, but there are some differences. One such difference is that SCTP is message oriented while TCP is byte oriented . Another difference is that the insequence delivery is optional for SCTP (i.e. it can be ‘turned off’) while it is always mandatory for TCP. SCTP makes use of so-called Stream Identifiers to identify a logical signalling connection (‘stream’) between two network nodes. A single SCTP association per X2-C interface instance is used, with different pairs of Stream Identifiers for f or X2-C common and dedicated procedures. A UE specific Context Identity is assigned by the Source eNB (i.e. the one sending a message) and the Target eNB (the one receiving a message) for signalling related to dedicated X2-C procedures. The purpose of the UE Context Identity is to distinguish UE specific X2-C signalling transport bearers from each other. The UE Context Identity is conveyed in all X2AP messages pertaining to the specific UE.
GPRS Tunnelling Protocol- User Plane (GTP-U) This is the same protocol as used in GPRS and UMTS (release 8). The main task of the GTP-U protocol is encapsulation and tunnelling of user data packets between network nodes. It is used on the X2-interface for forwarding of packets during handover execution, on the S1-interface (between eNB and SGW) and on a number of additional EPC interfaces. Each user data IP-packet is encapsulated by adding a GTP header. The header contains, among other things, a Tunnel Endpoint Identifier (TEID). The TEID is a locally l ocally allocated reference number that uniquely identifies a GTP tunnel in the node that allocated it. Thus, a GTP tunnel has two TEIDs associated with it (one in each ‘end’). Apis Technical Training AB LTE - X2 and S1-interface Copyright © Apis Technical Training AB 2007. All rights reserved.
6-4
Transport Layer Protocols The transport layer uses standard Internet Engineering Task Force (IETF) defined protocols, i.e. UDP/IP running over the selected data link and physical layer protocols. These transport layer protocols are not discussed further in this document.
6.3
The S1-interface
6.3.1
S1-interface Protocols
eNB
MME
S1AP
S1AP
SCTP
SCTP
IP
IP
Data Link Layer
Data Link Layer
Physical Layer
Physical Layer S1-MME
eNB
SGW
User Data
User Data
GTP-U
GTP-U
UDP
UDP
IP
IP
Data Link Layer
Data Link Layer
Physical Layer
Physical Layer S1-U
Figure 6-3: S1-interface protocols for CP (top) and UP (bottom)
As seen by comparing figures 6-2 and 6-3 the protocols for the X2- and S1-interfaces are close to identical, with the Application Protocol in the Control Plane being the only difference. Hence only the S1AP protocol is described here.
Apis Technical Training AB LTE - X2 and S1-interface Copyright © Apis Technical Training AB 2007. All rights reserved.
6-5
S1 Application Protocol (S1AP) The S1AP protocol is used for control signalling exchange between eNB and MME. All S1 procedures relate to a specific UE and are in that sense ‘dedicated’, even though the S1 specification does not use the term. Currently specified S1AP procedures include: •
•
•
•
•
Paging. Paging. Enables the MME to page the UE in a specific eNB. The MME initiates the paging procedure by sending a Paging message to each eNB with cells belonging to the Tracking Area(s) in which the UE is registered. The paging response back to the MME is initiated on the NAS layer and is forwarded to MME by the eNB as part of the NAS Signalling Transport procedure. NAS Signalling Transport. This procedure provides means to transport NAS messages to/from a given UE over the S1-interface. (This procedure is in all respects the same as the UTRAN Direct Transfer procedure). Initial Context Setup. Setup. This procedure supports the establishment of the necessary overall initial UE Context in the eNB to enable fast Idle-to-Active transition. The UE Context includes: SAE Bearer context, security context, roaming restriction, UE capability info, “subscriber type” info etc. The procedure is always initiated from the MME, typically in combination with network registration (Attach or Tracking area update). SAE Bearer Management. Management . The SAE Bearer management function is responsible for establishing, modifying and releasing E-UTRAN resources for user data transport with a given QoS (once an initial UE context is available in the eNB). The procedure is always initiated from the MME, with the exception of SAE Bearer Release that may be initiated from the eNB. Handover. Handover. Handover preparation and execution signalling over the S1-interface is only needed during inter-RAT handover or when there is no X2-interface present between the Source eNB and the Target eNB. For a normal X2-interface initiated inter-eNB handover a S1AP Handover Notification message is sent from the Target eNB to the MME after the UE has been successfully transferred to the new cell.
Apis Technical Training AB LTE - X2 and S1-interface Copyright © Apis Technical Training AB 2007. All rights reserved.
6-6
6.4
Self-organizing Networks With the term Self-organizing Networks (SON) is meant the functionality in network elements for self-configuration and self-optimization without (or with minimal) manual intervention.
6.4.1
Self-configuration Self-configuration is defined as the process where newly deployed network nodes (i.e. eNBs) are configured by automatic installation procedures in order to get the necessary basic configuration for system operation. This process works in the pre-operational state. Pre-operational state is understood as the state from when the eNB is powered up and has backbone connectivity connectivity until the RF transmitter is switched on.
After power-up the eNB needs to make its presence know to the MME, or MMEs, in the network. This requires that the eNB knows the transport IPaddress of the MME(s). An initial remote IP endpoint to be used for SCTP initialisation is provided to the eNB for each MME in the pre-operational state (the exact mechanism for this is not yet standardised). For each MME the eNB tries to initialize a so-called SCTP association (RFC 2960), using the known initial remote IP endpoint, until SCTP connectivity is established. Once SCTP connectivity has been established the eNB and MME are in a position to exchange application level configuration data needed for the two nodes to interwork correctly. During this process the eNB provides relevant information to the MME (e.g. eNB ID, list of supported Tracking Area(s) etc). The MME similarly provides relevant information to the eNB (e.g. MME ID, PLMN ID etc). When the application layer initialization is successfully concluded, and has been mutually acknowledged by the two peer nodes, the dynamic configuration procedure is completed and the S1-MME interface is operational. It is expected that some form of mutual node authentication procedure is needed prior to initiating this process (i.e. to detect fake or ‘impersonated’ ‘i mpersonated’ nodes). The eNB can then download additional configuration software, either from/via the MME or from some network management system. This may include configuration parameters such as: cell-id, neighbour cells (cell-ids, IP addresses), sub-carrier allocation, reference signal mapping, reference signal power, antenna tilt etc.
Apis Technical Training AB LTE - X2 and S1-interface Copyright © Apis Technical Training AB 2007. All rights reserved.
6-7
6.4.2
Self-optimization Self-optimization is defined as the process through which UE and eNB measurements are used to auto-tune the network. This process works in the operational state. Operational state is understood as the state where the RF interface is switched on (i.e. the eNB is being used for real traffic).
The current draft specification clearly states that the UE shall (‘shall’ is the same as ‘must’ in 3GPP language) support measurements and procedures that can be used for self-configuration and self-optimisation of the EUTRAN system. It should also be possible to associate the measurements for self-optimisation purposes with location information (e.g. the UE may provide GPS coordinates to the eNB). Such UE-assisted measurements can be used to, for example, optimize neighbour cell lists. The active RRC connections and their accompanying measurements can be used to gather needed information about neighbours. Known neighbours can be checked if they are really appropriate concerning radio conditions and new ones can be included based on information about detected cells received from the t he UEs. The radio measurements of eNB and UEs together with call events like call drops, failed or ‘ping-pong’ handovers etc may also influence the handover algorithm used. For example, if certain (average) measurement values fall below a certain threshold a (pre-configured) modified handover algorithm may be used until the problem disappears. Furthermore, through the use of OFDM the opportunity exists to distribute radio interface resources in a dynamic way to optimise the traffic situation or interference situation based on statistical measurements of power and interference level for single sub-carriers or groups of sub-carriers. This may be performed as an intra eNB process, but may also be linked to the X2-interface Load Indication (ICIC) procedure.
6.5
Home eNB The ‘home base station’ is not really a new concept since many people today have wireless LAN access points in their homes in conjunction with their broadband access. The standardisation work in 3GPP regarding home base stations belongs to Release 8 and incorporates both UTRAN NodeBs and E-UTRAN eNBs. The following description focuses on the E-UTRAN case but many of the questions and considerations raised apply equally well to the UTRAN case. It is expected that the home eNB will connect to the MME/SGW using the standard S1-interface and to other eNBs using the standard X2-interface. There are many fundamental issues that must be solved before home eNBs can be fully and securely included in the LTE/SAE ‘macro’ architecture, some of which are touched upon in the following.
Apis Technical Training AB LTE - X2 and S1-interface Copyright © Apis Technical Training AB 2007. All rights reserved.
6-8
6.5.1
Node Configuration Installation of the home eNB should require a minimum amount of manual intervention, both from the user and the operator. The existence of functions for Self-organizing Networks (SON) is expected because the: number of home eNBs may become very large subscribers may switch on and off the home eNB frequently operator may not be able to access the home eNB physically. •
•
•
The home eNB shall therefore be able to download the latest firmware and software to be used, as part of an initial or periodic activation procedure. procedure. A possible solution is that the eNB downloads the initial configuration from a known configuration server prior to powering up the radio interface. Thus, initialisation of the home eNB should be automated and require no manual configuration by either the user or network operator. The initialisation process should include also configuration of neighbour relationships, which requires the presence of the X2-interface. It must therefore be possible for the home eNB to trigger establishment and release of the X2-interface between the home eNB and Macro eNBs, as well as accept incoming X2-interface connection requests from Macro eNBs. This is also applicable for the X2 connection to another home eNB. It should be possible for both the owner and the network operator to cause the home eNB to download and install the latest software updates or configuration files. There may also be a function present that allows the operator to switch off the home eNB remotely. r emotely.
6.5.2
Access Restriction Naturally, the eNB should only allow access for a single subscriber (or group of subscribers) while all other subscribers must be barred from using it. The cell served by the home eNB is referred to as a Closed Subscriber Group (CSG) cell. As the term suggests, only a UE from a specific user group should be allowed access to that cell. This access restriction is needed because some backhaul links for this type of deployment are not considered to provide adequate QoS to support a large numbers of UEs. There may also be regulatory issues with sharing the backhaul link/eNB access in that location. Finally, the backhaul link may be owned by or paid for by the subscriber and he/she may not be too happy to share the link with others! The user group associated with a specific home eNB needs to be updated, under the supervision of the network operator, by the subscriber which is registered as the owner of the home eNB. When a subscriber is added to the user group by the registered owner the UE of the subscriber should be able to (almost) immediately camp on the cell of the home eNB and then acquire service through the home eNB. This is especially important in the
Apis Technical Training AB LTE - X2 and S1-interface Copyright © Apis Technical Training AB 2007. All rights reserved.
6-9
deployment scenario where this subscriber has no other means to access the network, i.e. there is no macro-layer coverage available. A UE should not camp on or access a CSG cell if it is not part of the user group that is allowed to access that CSG cell. The exact mechanisms for this is currently still under investigation.
6.5.3
Mobility The home eNB/CSG cells are part of the network of the operator, and therefore the design needs to support mobility of UEs between the macrolayer network and the home eNB/CSG cells. This is true for both Idle state behaviour (cell re-selection) and Connected state behaviour (handover). The home eNBs will be deployed in order to improve network coverage, to improve network capacity and to offer differentiated billing models. As the user billing could be dependent on whether the UE is using the home eNB or not it is important that the UE, when it is in range, automatically camps on the home eNB. This can be done by setting broadcasted re-selection parameters in such a way that the UE will always prioritize the CSG cell. It is also important that UEs camped on the home eNB do not cause excessive signalling load or processing load if/when the UEs moves frequently between the macro-layer network and the home eNB (e.g. excessive Tracking Area update signalling should be avoided). A possible solution to this is to, during automated initialization, make sure the home eNB belongs to the same Tracking Area as the surrounding macro eNBs. As discussed above the home eNBs will have an associated user group defining which UEs can access the home eNB. The handover procedure needs to take the user group of the Target home eNB into account when deciding whether to handover a UE to a specific home eNB. As the number of home eNBs in the network will become large the proportion of measurements made by a UE which could be wasted may become large, to the point where it affects the mobility performance of the UE/system as well as draining the battery of the UE. It is therefore necessary for the UE to, somehow, be able to avoid unnecessary measurements of home eNBs where it does not belong to the user group. It should be noted that, due to the expected high number of home eNBs and the nature of their deployment, it would not be practical to change the configuration for the mobility procedures (measurements, handover etc) in the macro layer nodes whenever a home eNB is deployed/dismissed.
Apis Technical Training AB LTE - X2 and S1-interface Copyright © Apis Technical Training AB 2007. All rights reserved.
6-10
6.5.4
Security The operator’s network must be protected from cases where the user (or someone else) is ‘tampering’ with the home eNB in an un-desired manner. Thus, the implementation of the home eNB must offer appropriate security protecting home eNB users and the connected network from security threats arising from accessing the backhaul link or internal interfaces (or configuration data) within the home eNB. To protect both the operator and the eNB owner it is desirable that mutual authentication, between home eNB and network, and establishment of a secure connection with a Security Gateway (SeGW) is part of the home eNB initialization process. The exact security mechanism to be used and the location of the SeGW function is not yet decided. Furthermore, since the home eNB will be a small, easily portable, device it is desirable for the operator that the home eNB recognises when it is operated in a different country to the HPLMN and, as a result, deactivates itself. Such a function can be important for charging reasons.
6.5.5
QoS and Interference The user should expect the same (or higher) QoS level when connected through the home eNB as when using an ‘ordinary’ eNB. It therefore makes sense for the UE to be able to display whether it is attached to a home eNB or to a macro eNB, regardless if it is in Idle or Connected state. The home eNB should not be allowed to cause excessive interference in the local radio environment. The home eNB must therefore continually or periodically perform measurements, and/or request measurements from the UE, to modify the current site-specific configuration. This is needed in order to dynamically adapt the radio r adio resource and parameter settings to the local environment (within the constraints imposed by the operator). Further, the operator can (remotely) define the maximum output power that a UE can use on the home eNB. Related to both QoS and interference issues it will be possible for the network operator to query the home eNB to send a report of the node status. The report should contain: information gathered from eNB selftesting activities, average data throughput, radio configuration (frequency and power), dropped calls and mobility flows (e.g. handovers between the home and macro layer).
Apis Technical Training AB LTE - X2 and S1-interface Copyright © Apis Technical Training AB 2007. All rights reserved.
6-11
6.6
References 25.820 32.816 32.821 36.410 36.413 36.420 36.423
3G home NodeB study item technical report Telecommunication mgmt; study on mgmt in LTE and SAE Telecommunication mgmt; study on SON for home NodeB E-UTRAN; S1 general aspects and principles E-UTRAN; S1 Application Protocol (S1AP) E-UTRAN; X2 general aspects and principles E-UTRAN; X2 Application Protocol (X2AP)
Apis Technical Training AB LTE - X2 and S1-interface Copyright © Apis Technical Training AB 2007. All rights reserved.
6-12
7
Evolved Packet Core 7.1 EPC NETWORK ARCHITECTURE ...................... ................................. ............... .... 7-2 7.1.1
Baseline EPC EPC Network Network Architecture Architecture ........................................7-2
7.1.2
EPC Architecture for MBMS.....................................................7-5
7.2 EPC QUALITY OF SERVICE ....................... ................................... ..................... ......... 7-7 7.2.1
EPC Bearer Concepts..............................................................7-7
7.2.2
QoS Parameters Parameters ......................................................................7-8
7.3 REFERENCES ....................... .................................. ...................... ....................... ................... ....... 7-9
Apis Technical Training AB LTE - Evolved Packet Core Copyright © Apis Technical Training AB 2007. All rights reserved.
7-1
7.1
EPC Network Architecture Section 7.1.1 presents the network architecture as it is most often displayed in the current specifications. However, the standardised network architecture for the Evolved Packet Core (EPC) is very flexible and allows various network configurations to be realised. Rather than specifying monolithic network nodes , the standard identifies functional entities that may be physically co-located or distributed according to product development and deployment scenarios. For instance, the MME may be co-located with the SGW that, in turn, may be co-located with the PGW that, in turn may be co-located with the PCRF etc. The nodes/functions and interfaces needed for MBMS transmission are described separately in section 7.1.2.
7.1.1
Baseline EPC Network Architecture UTRAN Iu
GERAN
SGSN
Gb Gr
HSS
S3
S4
S12
S6a
S1-MME
MME
S11
PCRF
Rx+
S7
S10
E-UTRAN
SGW
S1-U
3GPP IP-access Trusted
Non-3GPP IP-access
Non-trusted
Non-3GPP IP-access
S5 (S8a/b)
S2c
IMS / Internet /…
PGW
SGi
S2b
S2a
Wn* W n*
ePDG
Wm*
AAA
To HSS Wx* W x*
Figure 7-1: baseline EPC network architecture
Figure 7-1 shows the EPC network architecture for the non-roaming case. That is, the access network and the core network both belong to the same operator. The roaming case primarily affects the S5 and S7-interfaces. The legacy interfaces (Gr, Iu and Gb in fig 7-1) are not described here. For several of the EPC interfaces it has not yet been decided what application signalling protocol to use. In the following, the mentioning of a specific signalling protocol is simply omitted in such cases. Mobility Management Entity (MME). The MME manages and stores contexts (i.e. information lists) relating to UEs in both LTE Idle and LTE Active state. The UE context contains parameters such as: IMSI, S-TMSI, current Tracking Area, security keys, UE capabilities and currently assigned EPC bearer QoS. Apis Technical Training AB LTE - Evolved Packet Core Copyright © Apis Technical Training AB 2007. All rights reserved.
7-2
The MME is responsible for handling mobility management procedures such as Paging, Attach/Detach and Tracking Area updates. It handles security procedures such as Authentication and allocation of temporary identities (S-TMSI). It is also responsible for the transfer of UE contexts (in a backwards compatible format) to the SGSN in inter-RAT mobility scenarios. MME to SGSN signalling takes place over the S3-interface using the GPRS Tunnelling Protocol- Control plane (GTP-C). The MME connects to the E-UTRAN (eNB) with the S1-MME interface. This interface uses the S1 Application Protocol (S1AP), as described in chapter 6. When needed, the MME updates the HSS with UE location information and retrieves subscription and authentication data. This is done over the S6a-interface using either the Mobile Application Part (MAP) protocol or the DIAMETER protocol. For inter-MME user mobility, the S10-interface is used for UE context transfer between MMEs. The S11-interface connects the MME to the SGW. This interface is used for paging initiation (SGW to MME), handover/re-routing indications and establishment of EPC bearers (MME to SGW). Home Subscriber Server (HSS). (HSS) . The HSS holds subscription subscription profiles and security related information for each registered subscriber. It is an evolved version of the 2G/3G Home Location Register (HLR) that also includes the functionality of the Authentication Centre (AuC). Serving Gateway (SGW). (SGW) . The SGW terminates the downlink data path for UEs in LTE Idle state and initiates paging (to MME) when downlink data arrive for the UE. It manages and stores UE contexts (user IP-address, EPC bearer QoS, eNB/PGW IP-addresses and TEIDs). The SGW connects to E-UTRAN (eNB) via the S1-U interface using the GTP-U protocol. During, and after, a handover to GERAN/UTRAN the SGW acts as User Plane anchor, forwarding downlink user IP-packets to the SGSN via the S4-interface (GTP-U). In case the UTRAN system has a ‘Direct Tunnel’ architecture (SGSN eliminated from the User Plane) this forwarding takes place over the S12-interface (GTP-U). In the non-roaming case the S5-interface connects the SGW with the PGW for uplink and downlink user IP-packet transfer. In the roaming case the SGW is located in the visited network and the PGW in the home network, connected via the S8a/b-interface. The S5 and S8a-interfaces use GTP-U while S8b uses Proxy Mobile Ipv6, PMIP (RFC 3775).
Apis Technical Training AB LTE - Evolved Packet Core Copyright © Apis Technical Training AB 2007. All rights reserved.
7-3
Packet Data Network Gateway (PGW). (PGW). The PGW (also called PDN GW) is the connection between the EPC and external packet data networks, over the SGi-interface. It is responsible for allocation of user IP-addresses. The PGW also acts as the User Plane anchor for mobility to/from IP-access networks other than UTRAN/GERAN. These ‘other IP-access networks’ are divided into three groups: 3GPP IPaccess, trusted non-3GPP IP-access and non-trusted non-3GPP IP-access. Examples of non-GERAN/UTRAN access networks are: WLAN, xDSL, CDMA2000 and WiMAX. The 3GPP IP-access is the Interworking WLAN access (I-WLAN), first defined in Release 6, connected to PGW with the S2c-interface using the Dual Stack Mobile Ipv6 protocol (DS-MIPv6). Trusted non-3GPP IP-access connects to the PGW with the S2a-interface using PMIP or Mobile Ipv4, MIPv4 (RFC 3344). Non-trusted non-3GPP IP-access needs an ePDG in the connection path. The S2b-interface (PMIP) connects the PGW with the ePDG. Evolved Packet Data Gateway (ePDG). The ePDG performs access authentication when the UE tries to connect to the home domain. If needed, it performs QoS authorization and generates charging information for the packet data session. It may also perform packet filtering/policing functions. The ePDG connects to the non-trusted access network with the Wn*interface. The ePDG may require interaction with an AAA-server, using the Wm*-interface. Authentication, Authorization and Accounting server (AAA). This function either executes the AAA-functions or, alternatively, provides necessary data for the ePDG to do so. The AAA-server may need to download subscription information from the HSS via the Wx*-interface. Policy and Charging Rules Function (PCRF). The PCRF provides operator specific (or UE specific) QoS policies and charging rules to the PGW when an EPC bearer is to be established. The S7-interface uses the DIAMETER protocol. In a roaming scenario there may be interaction between a local (visited) PCRF and the home domain PCRF. This inter-PCRF interface is called the S9-interface (DIAMETER). The PCRF retrieves the necessary policy and charging parameters from the proper IMS Application Function (AF) over the Rx+-interface.
Apis Technical Training AB LTE - Evolved Packet Core Copyright © Apis Technical Training AB 2007. All rights reserved.
7-4
7.1.2
EPC Architecture for MBMS The MBMS Service The Multimedia Broadcast Multicast service (MBMS) allows multimedia content (messages, audio, video etc) to be efficiently transmitted to all users, or a well-defined group of users, in a given cell (or cells). The service was first introduced in 3GPP Release 6, supporting MBMS transmissions over the GERAN and UTRAN radio access networks. MBMS is really two services in one: •
•
The The broadcast service (or mode) where the transmitted content can be received by all terminals in a given area without restriction (provided the terminal supports MBMS of course). This service does not require any subscription support and no charging will be incurred. The multicast service (or mode) is transmitted solely to terminals that have actively joined a particular multicast group. This service may require subscription support and may also be charged for by the operator.
The MBMS service requires its own infrastructure (i.e. network nodes) and its own set of logical, transport and physical channels. A brief overview of the network architecture for MBMS within the EPC and the required functionality in E-UTRAN/E-UTRA is given in the following sections.
MBMS Architecture MBMS Gateway M3
MCE
MBMS1 Sm
M2
SGmb
M1
eNB
MBMS2
eBM-SC
Content Provider
Figure 7-2: EPC network architecture for MBMS
Evolved Broadcast Multicast Service Centre (eBM-SC) The eBM-SC provides functions for general MBMS service provisioning and delivery. It serves as an entry point for the content provider and authorises, schedules and delivers MBMS transmissions to the EPC. The eBM-SC connects to the MBMS GW via the SGmb-interface.
Apis Technical Training AB LTE - Evolved Packet Core Copyright © Apis Technical Training AB 2007. All rights reserved.
7-5
MBMS Gateway (MBMS GW) The MBMS GW is a logical entity whose function is the synchronised sending/broadcasting sending/broadcasting of MBMS data packets over the M1-interface to each eNB transmitting the service. The MBMS GW also performs MBMS session control signalling (e.g. session start/stop time) towards the eNBs (or MCE) over the M3-interface. The MBMS GW may be functionality split into MBMS1 into MBMS1 (Control Plane) and MBMS2 and MBMS2 (User Plane) functions. If so, they are connected through the Sm-interface. Multicell/multicast Coordination Entity (MCE) The MCE is a logical entity that may, or may not, be physically co-located with eNB. If it is co-located with the eNB, the M2-interface does not exist. Its task is to allocate the radio resource (time/frequency resource and the modulation and coding scheme) to be used by all eNBs taking part in MBSFN transmission (see 7.2.3). Note: if the MCE exists as a stand-alone function, it should logically belong to E-UTRAN and not the EPC.
E-UTRAN Support for MBMS For the MBMS service in E-UTRAN, a group of cells can be ‘combined’ to achieve an MBMS an MBMS Single Frequency Network (MBSFN) transmission. An MBSFN transmission is a simulcast transmission technique realised by the transmission of identical waveforms at the same time on the same frequency from multiple cells. A fundamental requirement for multi-cell MBSFN deployment is inter-site synchronization for which the cells should be synchronized to within a few microseconds An MBSFN transmission from multiple cells is seen as a single cell transmission by the UE. Signals from all MBSFN cells arrive at the UE and are dealt with in the same manner as multipath delayed signals. In this manner the UE can combine the energy from multiple transmitters with no additional receiver complexity. If the UE is close to one MBSFN eNB and relatively distant from a second MBSFN eNB, the amount of delay between the two signals can be substantial. For this reason, the MBSFN transmissions in E-UTRAN are supported using a 7.5 kHz sub-carrier spacing, as opposed to 15kHz in the ‘normal’ case, and a much longer cyclic prefix. MBSFN cells also use a common MBSFN-specific reference signal from all participating eNBs to facilitate channel estimation. As a consequence of the MBSFN transmission scheme, the UE can roam between cells with no handover required. Signals from various cells will vary in strength and in relative delay, but in aggregate the received signal is still dealt with in the same way as conventional single-cell OFDM transmissions. t ransmissions. Furthermore, a cell in E-UTRAN is either an MBMS-dedicted cell or an MBMS/unicast mixed cell. mixed cell. In an MBMS-dedicated cell the Multicast Traffic Channel (MTCH) and Multicast Control Channel (MCCH) are mapped on the Multicast Channel (MCH) transport channel. Such a cell has no uplink resources. Apis Technical Training AB LTE - Evolved Packet Core Copyright © Apis Technical Training AB 2007. All rights reserved.
7-6
In an MBMS/unicast mixed cell the MTCH and MCCH are mapped on the Downlink Shared Channel (DL-SCH) or, in case of MBSFN transmission, the MCH. The MBMS packets and the ‘regular’ unicast packets are timemultiplexed on the DL-SCH through MAC-layer packet scheduling. Uplink resources exist as normal, but the use of HARQ operation (and hence uplink ACK/NACKs) ACK/NACKs) for MBMS transmission is optional. Both MBMS-dedicated and MBMS/unicast mixed cells can partake in MBSFN transmissions.
7.2
EPC Quality of Service
7.2.1
EPC Bearer Concepts An EPC An EPC Bearer (sometimes called Evolved Packet System, EPS, Bearer) is a combination of one or more Service Data Flows (SDFs). An SDF can most easily be seen as a uni-directional flow of data relating to a specific service (thus a VoIP call requires both an uplink and a downlink SDF). The SDFs are defined between the UE and the PGW. Quality of Service (QoS) is defined per EPC Bearer, meaning that all SDFs mapped to the same EPC Bearer will have the same QoS. Each EPC Bearer is associated with an uplink Traffic Flow Template (TFT) in the UE and a downlink TFT in the PGW. A TFT is a set of rules on how to perform IP packet filtering. An EPC Bearer is either a default bearer or a dedicated bearer. The default bearer provides the UE with always-on IP-connectivity and is established when the UE attaches to the network. Any EPC Bearer that is established after this point is called a dedicated bearer. The initial QoS parameter values for the default bearer are assigned by the network based on subscription data (the MME sets those initial values based on subscription data retrieved from HSS). The PGW may change those values based on interaction with the PCRF or based on local policy configurations. A dedicated EPC Bearer is either a GBR bearer or a non-GBR bearer, depending on if network resources relating to a Guaranteed Bit Rate (GBR) have been associated with the EPC Bearer or not. A default bearer is always a non-GBR bearer. A Radio Bearer transports the packets of an EPC Bearer between the UE and the eNB. There is one-to-one mapping between an EPC Bearer and a Radio Bearer.
Apis Technical Training AB LTE - Evolved Packet Core Copyright © Apis Technical Training AB 2007. All rights reserved.
7-7
7.2.2
QoS Parameters Each EPC Bearer (GBR or non-GBR) is associated with an Allocation an Allocation and Retention Priority (ARP) value, a Maximum Bit Rate (MBR) and a Label . The main purpose of the ARP is to decide whether a bearer establishment/modification request can be accepted or needs to be rejected in case of resource limitations. In addition, the ARP can be used by the eNB to decide which bearer(s) to drop during exceptional resource limitations (e.g. at handover). It should be noted that the ARP is not intended to be used as input to the eNB packet-scheduling algorithm. A Label is an operator-defined value (1, 2, 3, …) that is used as an internal reference to eNB specific parameters that control Radio Bearer packet treatment (e.g. HARQ operation, scheduling weights, packet queue management, admission thresholds etc). There are also 3GPP standardised Label characteristics, relating to specific combinations of Bearer of Bearer Type (GBR or non-GBR), non-GBR), Layer 2 Packet Delay Budget (L2PDB) and Layer and Layer 2 Packet Loss Rate (L2PLR). The purpose of these parameters is to support the t he configuration of MAC packet scheduling and layer 2 ARQ and HARQ functions (e.g. the setting of scheduling priority weights and the number of HARQ retransmissions).
Apis Technical Training AB LTE - Evolved Packet Core Copyright © Apis Technical Training AB 2007. All rights reserved.
7-8
7.3
References 23.234 23.246 23.401 23.402 23.882 36.300 36.938
3GPP system to WLAN interworking: system description MBMS; architecture and functional description GPRS enhancements enhancements for E-UTRAN access Architecture enhancements enhancements for non-3GPP accesses 3GPP SAE: report on technical options and conclusions E-UTRA/E-UTRAN; overall description Stage 2 Improved network controlled mobility between LTE and 3GPP2/WiMAX radio technologies
Apis Technical Training AB LTE - Evolved Packet Core Copyright © Apis Technical Training AB 2007. All rights reserved.
7-9
