Modeling and Performance Improvement of TCP over LTE Handover MATTEO PACIFICO

Transcription

1 Modeling and Performance Improvement of TCP over LTE Handover MATTEO PACIFICO Master s Degree Project Stockholm, Sweden 2009

2

3 CHAOO CHAOO CHAOO con tanto amore mi avete cresciuto, con uno sguardo mi avete insegnato a vivere, con un sorriso mi avete reso felice, con coraggio avete accettato e sostenuto le mie scelte, con tanta pazienza mi avete permesso di diventare quel che sono ora. Siete per sempre nel mio cuore! A Mamma e Papà

4

5 Abstract The Long Term Evolution (LTE) of the UMTS Terrestrial Radio Access and Radio Access Network is a new communication standard aimed for commercial deployment in Goals for LTE include support for improved system capacity and coverage, high peak data rates, low latency, reduced operating costs, multiantenna support, flexible bandwidth operations and seamless integration with existing systems. The aim of this thesis project is to study the impacts on the end-user and system performance when users with high bit rates tcp services are moving through the network. These impacts affect the reduced end-user or system throughput, e.g., due to congestion in the transport network, leading to poor utilization of the transport and radio resources available. To reach such an aim, it has been necessary (1) to create a new simulator with the ns-2 and perform simulation in different network settings and (2) formulate a mathematical model able to capture the principal dynamics of the real system. Possible solutions to mitigate the impacts are investigated by comparing the simulations results of tcp performance in the radio and transport network, with the mathematical model of it. The project has been performed at the Automatic Control Lab at KTH in Stockholm and in collaboration with Ericsson Research Lab. The work is part of a joint effort with Davide Pacifico. Further results are available in the master thesis Analysis and Performance Improvement of tcp during handover of LTE [1]. iii

6

7 Preface LTE, short for Long Term Evolution, is the result of ongoing work by the 3 rd Generation Partnership Project (3GPP), a collaborative group of international standards organizations and mobile-technology companies. 3GPP set out in 1998 to define the key technologies for the third generation of GSM-based mobile networks (3G), and its work has continued to define the ongoing evolution of these networks. Near the end of 2004, discussions on the longer-term evolution of 3G networks began, and a set of high-level requirements for LTE was defined: the networks must transmit data at a reduced cost per bit compared to 3G; they must be able to offer more services at lower transmission cost with better user experience; LTE must have the flexibility to operate in a wide number of frequency bands; it should utilize open interfaces and offer a simplified architecture; and it must have reasonable power demands on mobile terminals. Standardization work on LTE is continuing with some operators projected to deploy the first LTE networks in LTE ( Turbo 3G ), is a wireless broadband technology designed to support roaming Internet access via cell phones and handheld devices. Because LTE offers significant improvements over older cellular communication standards, some refer to it as a 4G (fourth generation) technology along with WiMax. LTE defines new radio connections for mobile networks, and will utilize Orthogonal Frequency Division Multiplexing (OFDM), a widely used modulation v

8 Preface technique that is the basis for Wi-Fi, WiMAX, and the Digital Video Broadcasting (DVB) and Digital Audio Broadcasting (DAB) technologies. The targets for LTE indicate bandwidth increases as high as 100 Mbps on the downlink, and up to 50 Mbps on the uplink. However, this potential increase in bandwidth is just a small part of the overall improvement LTE aims to provide. LTE is optimized for data traffic, and it will not feature a separate, circuit-switched voice network, as in 2G GSM and 3G UMTS networks. LTE is the successor to the current generation of UMTS 3G technology, which is based upon WCDMA (3G), HSDPA, HSUPA, and HSPA. LTE is not a replacement for UMTS in the way that UMTS was a replacement for GSM, but rather an update to the UMTS technology that will enable it to provide significantly faster data rates for both uploading and downloading. With its architecture centered on Internet Protocol (IP), Long Term Evolution promises to have excellent support for browsing Web sites, VoIP and other IPbased services. LTE can theoretically support downloads at 100 Megabits per second (Mbps) or more based on experimental trials. Another important feature of LTE is the amount of flexibility it allows operators in determining the spectrum in which it will be deployed. Not only will LTE have the ability to operate in a number of different frequency bands (meaning operators will be able to deploy it at lower frequencies with better propagation characteristics), but it also features scalable bandwidth. Whereas WCDMA/HSPA uses fixed 5 MHz channels, the amount of bandwidth in an LTE system can be scaled from 1.25 to 20 MHz. This means networks can be launched with a small amount of spectrum, alongside existing services, and adding more spectrum as users switch over. It also allows operators to tailor their network deployment strategies to fit their available spectrum resources, and not have to make their spectrum fit a particular technology. vi

9 Preface The Transmission Control Protocol (tcp) is one of the core protocols of the Internet Protocol Suite. TCP is so central that the entire suite is often referred to as tcp/ip. tcp has been optimized for wired networks. Any packet loss is considered to be the result of congestion and the congestion window size is reduced dramatically as a precaution. However, wireless links are known to experience sporadic and usually temporary losses due to fading, shadowing, handoff, and other radio effects, that cannot be considered congestion. After the (erroneous) back-off of the congestion window size, due to wireless packet loss, there can be a congestion avoidance phase with a conservative decrease in window size. This causes the radio link to be under utilized. Extensive research has been done on the subject of how to combat these harmful effects. Suggested solutions can be categorized as end-to-end solutions (which require modifications at the client and/or server), link layer solutions, or proxy based solutions, which require some changes in the network without modifying end nodes. In cellular telecommunications, the term handoff refers to the process of transferring an ongoing call or data session from one channel connected to the core network to another. The British English term for transferring a cellular call is handover, which is the terminology standardized by 3GPP within such European originated technologies as GSM and UMTS. The aim of this thesis is to study the performances of the tcp protocol in the handover procedure in LTE. This study proposes a mathematical model of that features a compromise between simplicity and accuracy in the representation of the dynamics of the real system. The handover of an end-user who moves in a train or a bus or walks in a crowded street while is surfing on internet by his terminal could be an abrupt service interruptions. It is therefore important that the operator control properly the handover procedure to avoid these problems. vii

10 Preface For these reasons is essential put the attention to mitigate the problems of TCP during the handover and find a way to keep it transparent to the end-users. To reach this aim, our analytical model is compared with a set of simulations of the real network scenario performed by an original LTE simulation implemented the ns-2 environment. The thesis at a glance This thesis work is organized in five chapters. Chapter 1 is an overview of the LTE system and its main new features, such as the system architecture, the Hybrid Automatic Repeat request and the Evolved RAN. A section is dedicated to LTE handover in which the procedure is explained in detail. In Chapter 2 the tcp protocol is described and its combination with LTE. In particular we focus our attention on the tcp Reno. In Chapter 3 we investigate problems arising during the handover procedure in LTE network and we will show how to solve or mitigate it. We introduce also an ns-2 network simulator and how it is implemented to realize our LTE simulations. In Chapter 5 we develop a model of TCP during handover, which is based on the Joint link model proposed by Möller in 2008 [2]. This model shows the behavior of the TCP congestion window of a mobile that does the LTE handover procedure. Finally in Chapter 6, we discuss the results and propose some developments for future studies. viii

11 Contents Abstract Preface Contents List of Figures List of Tables Abbreviations iii v xi xiv xv xvii 1 Background LTE (Long Term Evolution) Requirements and performance goals for LTE System architecture description Key Features Network Sharing Mobility management LTE Technologies OFDM Multiple antenna techniques Scheduling Handover Summary ix

12 Contents 2 TCP and LTE Importance of TCP TCP key features Flow control TCP Applicability Congestion control Explicit Congestion Notification (ECN) Evaluation of TCP Performance Bottleneck in the core network GPRS Tunneling Protocol (GTP) Summary Management of LTE handover Forwarding procedure Optimal case Window halving case Timeout occurrence case Two improving solutions Advancing the path switch command Predicting the handover occurrence Validation of the solutions Summary LTE simulator: an overview LTE simulator through ns-2 and nsmiracle Network topology Summary x

13 Contents 5 Mathematical model of TCP over LTE Existing models Model adaptation Single flow and multiple flow model Application scenario Handover preparation Handover actuation Handover completion Simulations and Results Summary Conclusion and future work Conclusion Future work Bibliography 81 xi

14

15 List of Figures FDD and TDD operating bands LTE architecture Layer 2 structure for Downlink Example of mapping of logical channels to transport channels Mobility states of the UE in LTE Downlink reference signal structure - normal cyclic prefix, two transmit antennas Radio channel access modes Channel-dependent scheduling Message chart of the LTE handover procedure Sliding window mechanism tcp slow start and congestion avoidance tcp fast retransmit and fast recovery The TOS field in ip header and ECN signaling Bytes 13 and 14 of the tcp header GTP-U Protocol Stack GTPv1 header (default and optional fields) rtt behavior during the LTE handover xiii

16 List of Figures cwnd during handover in case of optimal case, window halving and timeout Difference between optimal case and window halving Difference between optimal case and timeout verification Normalized throughput during the time interval (rtt 1, rtt 2 ) considering optimal case, window halving and timeout for different values of cwnd MAX Different averages of the normalized throughput on varying the cwnd MAX parameter Advancing the path switch command in the handover message chart Modification to the handover message chart to predict the handover cwnd and rtt of the reference ME during the handover Reference ME received sequence number One node of nsmiracle Reference simulation scenario LTE handover scenario Bit rate comparison cwnd comparison Buffer status comparison rtt comparison xiv

17 List of Tables LTE performance goal LTE user throughput and spectrum efficiency requirements Interruption time requirements, LTE-GSM and LTE-WCDMA Logical channel characterized by the transferred information Transport channel characterized by how the data is transferred over radio interface E-UTRA Numerology Simulations parameters about the solution proposed in Section Common parameters of all simulation Parameters for simulation results and mathematical model comparison xv

18

19 Abbreviations 3GPP AIPN AM AMC AQM ARQ CDM CN CQI CWR DL DAE DSCP enb ECE ECN EDGE EPC EPS E-UTRA E-UTRAN FDM FDD FFT GPRS GSM 3 rd Generation Partnership Project All IP Network Acknowledged Mode Adaptive Modulation and Coding Active Queue Management Automatic Repeat Request Code Division Multiplexing Core Network Channel Quality Information Congestion Window Reduced Downlink Differential Algebraic Equation Differentiated Services Codepoint enodeb ECN-echo Explicit Congestion Notification Enhanced Data rates for GSM Evolution Evolved Packet Core Evolved Packet System Evolved UMTS Terrestrial Radio Access Evolved UMTS Terrestrial Radio Access Network Frequency Division Multiplexing Frequency Division Duplex Fast Fourier Transform General Packet Radio Service Global System for Mobile communications xvii

20 Abbreviations GTP GW HARQ HO HSPA HSS IFFT IP LTE NAS MAC ME MIMO MISO MME MMOG NAS NRT OFDM PDCP PDN PDN GW PHY PLMN QoS RACH RAN RAT RED RLC RNC ROHC RRC RT RTO GPRS Tunneling Protocol Gateway Hybrid ARQ Handover High Speed Packet Access Home Subscriber Server Inverse FFT Internet Protocol Long Term Evolution Non-Access Stratum Medium Access Control Mobile Equipment Multiple Input Multiple Output Multiple Input Single Output Mobility Management Entity Multimedia Online Gaming Non-Access Stratum Non-Real Time Orthogonal Frequency Division Multiplexing Packet Data Convergence Protocol Packet Data Network Packet Data Network Gateway Physical Public Land Mobile Network Quality of Services Random Access Channel Radio Access Network Radio Access Technology Random Early Discard Radio Link Control Radio Network Controller Robust Header Compression Radio Resource Control Real Time Retransmission Timeout xviii

21 Abbreviations RTT RS SAE SDU SeNB SFN SGW SIMO SISMO SNR TA TB TDD TEID TeNB TM TCP TDM TOS TTI UE UL UM UMTS VIP WCDMA WFQ Round Trip Time Reference Symbol System Architecture Evolution Service Data Unit Source enodeb Single frequency network Serving Gateway Single Input Multiple Output Single Input Single Output Signal-to-Noise Ratio Tracking Area Transport Block Time Division Duplex Tunnel Endpoint Identifier Target enodeb Transparent Mode Transport Control Protocol Time Division Multiplexing Type Of Service Transmission Time Interval User Equipment Uplink Unacknowledged Mode Universal Mobile Telecommunications System Voice over IP Wireless Coded Division Multiple Access Weighted Fair Queue xix

22

23 Chapter 1 Background In this chapter we describe the key aspects of LTE. The chapter is then concluded with a detailed description of the handover mechanism of LTE. 1.1 LTE (Long Term Evolution) The recent increase of mobile data usage and the emergence of new applications, such as Multimedia Online Gaming (MMOG), mobile TV, Web 2.0, streaming contents, have motivated the 3 rd Generation Partnership Project (3GPP) to work on the Long Term Evolution (LTE). LTE is the latest standard in the mobile network technology tree, which previously implemented the GSM/EDGE and UMTS/HSxPA network technologies now account for over 85% of all mobile subscribers. LTE will ensure 3GPP s competitive edge over other cellular technologies. LTE, whose radio access is called Evolved UMTS Terrestrial Radio Access Network (E-UTRAN), is expected to substantially improve end-user throughputs and sector capacity also to reduce user plane latency, bringing significantly improved user experience with full mobility. With the emergence of Internet Protocol (ip) as the protocol of choice for carrying all types of traffic, LTE is scheduled to provide support for ip-based traffic with end-to-end Quality of ser- 3

24 1. Background vice (QoS). Voice traffic will be supported mainly as Voice over ip (VIP) enabling better integration with other multimedia services. Initial deployments of LTE are expected by 2010 and commercial availability on a larger scale 1-2 years later. Unlike High Speed Packet Access (HSPA), which was accommodated within the Release 99 UMTS architecture, 3GPP is specifying a new Packet Core, the Evolved Packet Core (EPC) network architecture to support the E-UTRAN through a reduction in the number of network elements, simpler functionality, improved redundancy but most importantly allowing for connections and handover to other fixed line and wireless access technologies, giving the service providers the ability to deliver a seamless mobility experience. LTE has been set aggressive performance requirements that rely on physical layer technologies, such as: Orthogonal Frequency Division Multiplexing (OFDM), Multiple-Input Multiple-Output (MIMO) systems and Smart Antennas to achieve the baseline targets. The main objectives of LTE are to minimize the system and User Equipment (UE) complexities, allow flexible spectrum deployment in existing or new frequency spectrum and to enable co-existence with other 3GPP Radio Access Technologies (RATs) Requirements and performance goals for LTE E-UTRA is expected to support different types of services including web browsing, FTP, video streaming, VIP, online gaming, real time video, push-totalk and push-to-view. Therefore, LTE is being designed to be a high data rate and low latency system as indicated by the key performance criteria shown in Table The bandwidth capability of a UE is expected to be 20MHz for both transmission and reception. The service provider can however deploy cells with any of the bandwidths listed in the table. This gives flexibility to the service providers to tailor their offering dependent on the amount of available spectrum 4

25 LTE (Long Term Evolution) 1.1 Metric Peak data rate Mobility support Control plane latency (Transition time to active state) User plane latency Control plane capacity Coverage (Cell sizes) Spectrum flexibility Requirement DL: 100 Mbps UL: 50 Mbps (for 20 MHz spectrum) Up to 500 km/h but optimized for low speeds from 0 to 15 km/h < 100 ms (for idle to active) < 5 ms > 200 users per cell (for 5 MHz spectrum) km with slight degradation after 30 km 1.25, 2.5, 5, 20 MHz Table LTE performance goal or the ability to start with limited spectrum for lower upfront cost and grow the spectrum for extra capacity. During 2005 the 3GPP activity on 3G evolution was setting the requirement for LTE. These are documented in [8] and were divided into seven categories: capabilities, system performance, deployment-related aspects, architecture and migration, radio resource management, complexity, general aspects. 5

26 1. Background Performance measure Downlink target Uplink target relative to baseline relative to baseline Average user throughput 3x-4x 2x-3x (per MHz) Cell-edge user throughput 2x-3x 2x-3x (per MHz, 5th percentile) Spectrum efficiency 3x-4x 2x-3x (bit/s/hz/cell) Table LTE user throughput and spectrum efficiency requirements Capabilities The targets are 100 Mbit/s for downlink and 50 Mbit/s for uplink when operating in 20 MHz spectrum allocation. Thus, the requirements can be expressed as 5 bit/s/hz for the downlink and 2.5 bit/s/hz for uplink. There are two kind of latency requirements: control-plane requirements and user-plane requirements. The control-plane latency requirements address the delay for transiting from different non-active terminal states to an active state. The user-plane latency requirement is expressed as the time it takes to transmit a small ip packet from the terminal to the RAN edge node or vice versa measured on the ip layer. System performance The LTE system performance design targets user throughput, spectrum efficiency, mobility and coverage. The first two are summarized in Table The mobility requirements focus on the mobile terminals speed. For speeds up to 15 km/h there are the best performances; for speeds up to 120 km/h there should be high achievement and for speeds above 120 km/h the system should be able to keep the connection. The maximum speed allowed in the LTE system is set to 350 km/h or 500 km/h depending on the frequency band. 6

27 LTE (Long Term Evolution) 1.1 Non-real-time (ms) Real-time (ms) relative to baseline relative to baseline LTE to WCDMA LTE to GSM Table Interruption time requirements, LTE-GSM and LTE-WCDMA. The coverage requirements focus on the cell range. The cells up to 5 km of radius allow non-interference-limited scenarios; for cells range up to 30 km, a slight performances degradation are tolerated and cell ranges up to 100 km are not precluded but no performance requirements are stated yet. Deployment-related aspects The requirement on the deployment scenario includes both the case when the LTE system is deployed as a stand-alone system and the case when it is deployed together with WCDMA/HSPA and/or GSM. For mobile terminals supporting those technologies, Table lists the interruption requirements, that is, longest acceptable interruption in the radio link when moving between the different radio access technologies, for both real-time and non-real-time services. The basis for the requirements on spectrum flexibility is the requirement for LTE to be deployed in existing IMT-2000 frequency bands (coexistence with the systems that are already deployed in those bands), that is LTE should support both Frequency Division Duplex (FDD), and Time Division Duplex (TDD) (Figure 1.1.1). Architecture and migration LTE RAN architecture should be packet based (and also support real-time class traffic), simplify and minimize the interface, support an end-to-end QoS and designed to minimize the jitter (for example, for tcp/ip traffic type) [8]. 7

28 1. Background TDD-Based radio access FDD-Based radio access Uplink Downlink Frequency (MHz) Figure FDD and TDD operating bands. Radio resource management The radio resource management requirements are divided into enhanced support for end-to-end QoS, efficient support for transmission of higher layers, and support of load sharing and policy management across different radio access technologies. Complexity and general aspects LTE complexity requirements imply that the number of options should be minimized with no redundant required features. This has an impact also to the cost and service related aspects that, specific to the cost, it is desirable to minimize it maintaining the desired performance System architecture description The architecture consists of the following functional elements: Evolved Radio Access Network (RAN) The evolved RAN for LTE is composed of a single node, i.e., the enodeb (enb) that interfaces with the UE. The enb hosts the PHYsical (PHY), Medium Access Control (MAC), Radio Link Control (RLC), and Packet Data Control Protocol (PDCP) layers that include the functionality of user-plane header-compression 8

29 LTE (Long Term Evolution) 1.1 and encryption. It also offers Radio Resource Control (RRC) functionality corresponding to the control plane. It performs many functions including radio resource management, admission control, scheduling, enforcement of negotiated UL QoS, cell information broadcast, ciphering/deciphering of user and control plane data, and compression/decompression of DL/UL user plane packet headers. App User Plane TCP IP PDCP RLC/MAC PDCP RLC/MAC GTP UDP/IP IP GTP UDP/IP PHY PHY L1 L1 S-GW ME enodeb X2 S1-UP S1-CP IP transport network MME Control Plane NAS RRC PDCP RLC/MAC PHY RRC PDCP RLC/MAC PHY S1-AP SCTP IP L1 NAS S1-AP SCTP IP L1 Figure LTE architecture. Serving Gateway (SGW) The SGW routes and forwards user data packets, while also acting as the mobility anchor for the user plane during inter-enb handovers and as the anchor for mobility between LTE and other 3GPP technologies (terminating S4 interface [7] and relaying the traffic between 2G/3G systems and Packet Data Network Gateway, PDN GW). For idle state UEs, the SGW terminates the DL data path 9

30 1. Background and triggers paging when DL data arrives for the UE. It manages and stores UE contexts, e.g. parameters of the ip bearer service, network internal routing information. It also performs replication of the user traffic in case of lawful interception. Mobility Management Entity (MME) The MME is the key control-node for the LTE access network. It is responsible for idle mode UE tracking and paging procedure including retransmissions. It is involved in the bearer activation/deactivation process and is also responsible for choosing the SGW for a UE at the initial attach and at the time of intra- LTE handover involving Core Network (CN) node relocation. It is responsible for authenticating the user (by interacting with the Home Subscriber Server, HSS). The Non-Access Stratum (NAS) signaling terminates at the MME and it is also responsible for generation and allocation of temporary identities to UEs. It checks the authorization of the UE to camp on the service provider s Public Land Mobile Network (PLMN) and enforces UE roaming restrictions. The MME is the termination point in the network for ciphering/integrity protection for NAS signaling and handles the security key management. Lawful interception of signaling is also supported by the MME. The MME also provides the control plane function for mobility between LTE and 2G/3G access networks with the S3 interface terminating at the MME from the SGSN. The MME also terminates the S6a interface towards the home HSS for roaming UEs Key Features EPS to EPC A key feature of the EPS is the separation of the network entity that performs control-plane functionality (MME) from the network entity that performs bearer- 10

31 LTE (Long Term Evolution) 1.1 plane functionality (SGW) with a well defined open interface between them (S11). Since E-UTRAN will provide higher bandwidth to enable new services as well as to improve existing ones, separation of MME from SGW implies that SGW can be based on a platform optimized for high bandwidth packet processing, where as the MME is based on a platform optimized for signaling transactions. This enables selection of more cost-effective platforms for, as well as independent scaling, of each of these two elements. Service providers can also choose optimized topological locations of SGWs within the network independent of the locations of MMEs in order to optimize bandwidth, reduce latencies and avoid concentrated points of failure. S1-flex Mechanism The S1-flex concept provides support for network redundancy and load sharing of traffic across network elements in the CN, the MME and the SGW, by creating pools of MMEs and SGWs and allowing each enb to be connected to multiple MMEs and SGWs in a pool Network Sharing The LTE architecture enables service providers to reduce the cost of owning and operating the network by allowing the service providers to have separate CN (MME, SGW, PDN GW) while the E-UTRAN (enbs) is jointly shared by them. This is enabled by the S1-flex mechanism by enabling each enb to be connected to multiple CN entities. When a UE attaches to the network, it is connected to the appropriate CN entities based on the identity of the service provider sent by the UE. In this section, we describe the functions of the different protocol layers and their location in the LTE architecture. In Figure 1.1.2, the NAS protocol, which 11

32 1. Background runs between the MME and the UE, is used for control-purposes such as network attach, authentication, setting up of bearers, and mobility management. All NAS messages are ciphered and integrity protected by the MME and UE. The RRC layer in the enb makes handover decisions based on neighbor cell measurements sent by the UE, pages for the UEs over the air, broadcasts system information, controls UE measurement reporting such as the periodicity of Channel Quality Information (CQI) reports and allocates cell-level temporary identifiers to active UEs. It also executes transfer of UE context from the Source enb to the Target enb during handover, and does integrity protection of RRC messages. The RRC layer is responsible for the setting up and maintenance of radio bearers. In the user-plane, the PDCP layer is responsible for compression/decompression the headers of user plane ip packets using Robust Header Compression (ROHC) to enable efficient use of air interface bandwidth. This layer also performs ciphering of both user plane and control plane data. Because the NAS messages are carried in RRC, they are effectively double ciphered and integrity protected, once at the MME and again at the enb. The RLC layer is used to format and transport traffic between the UE and the enb. RLC provides three different reliability modes for data transport: Acknowledged Mode (AM), Unacknowledged Mode (UM), or Transparent Mode (TM). The UM mode is suitable for transport of Real Time (RT) services because such services are delay sensitive and cannot wait for retransmissions. The AM mode, on the other hand, is appropriate for non-rt (NRT) services such as file downloads. The TM mode is used when the PDU sizes are known a priori such as for broadcasting system information. The RLC layer also provides in-sequence delivery of Service Data Units (SDUs) to the upper layers and eliminates duplicate SDUs from being delivered to the upper layers. It may also segment the SDUs depending on the radio conditions. 12

33 LTE (Long Term Evolution) 1.1 Furthermore, there are two levels of re-transmissions for providing reliability, namely, the Hybrid Automatic Repeat request (HARQ) at the MAC layer and outer ARQ at the RLC layer. The outer ARQ is required to handle residual errors that are not corrected by HARQ, that is kept simple by the use of a single bit error-feedback mechanism. An N-process stop-and-wait HARQ is employed that has asynchronous re-transmissions in the DL and synchronous re-transmissions in the UL. Synchronous HARQ means that the re-transmissions of HARQ blocks occur at pre-defined periodic intervals, hence, no explicit signaling is required to indicate to the receiver the retransmission schedule. Asynchronous HARQ offers the flexibility of scheduling re-transmissions based on air interface conditions. The Figure show the structure of layer 2 for DL. The PDCP, RLC and MAC layers together constitute layer 2. Radio Bearers PDCP ROCH ROCH ROCH ROCH Security Security Security Security RLC Segm. ARQ Segm. ARQ Segm. ARQ Segm. ARQ BCCH PCCH Logical Channels Scheduling / Priority Handling MAC Multiplexing UE1 Multiplexing UEn HARQ HARQ Transport Channels Figure Layer 2 structure for Downlink. In LTE, there is significant effort to simplify the number and mappings of logical and transport channels. The different logical and transport channels in LTE are illustrated in Table and Table 1.1.5, respectively. The transport channels are distinguished by the characteristics (e.g. adap- 13

34 1. Background Channel type Channel Name Carried information Broadcast Control DL channel for broadcasting system control info Channel (BCCH) Paging Control DL channel for transferring paging Channel (PCCH) Common Control UL channel for transmitting control info and Channel (CCCH) used Control channels (carry control plane info) Traffic channels (carry user plane info) by UE without RRC connection Multicast Control DL point-to-multipoint Channel (MCCH) channel for transmitting MBMS control info Dedicated Control DL point-to-point bidirectional channel for Channel (DCCH) exchanging control information and used by UE with RRC connection Dedicated Traffic Bi-directional channel Channel (DTCH) dedicated to a single UE Multicast Traffic DL point-to-multipoint Channel (MCCH) channel for transmission of MBMS data Table Logical channel characterized by the transferred information. 14

35 LTE (Long Term Evolution) 1.1 Channel type Channel Name Carried information Broadcast Channel fixed transport format (BCH) Downlink Shared HARQ, dynamic link Channel (DL-SCH) adaptation, support for UE DRX, dynamic Downlink channels and semi-static resource allocation Paging Channel required to be broadcast (PCH) Multicast Channel Support for SFN combining (MCCH) and semi-static re- source allocation Uplink Shared HARQ, dynamic link Channel (UL- adaptation, support SCH) for UE DRX, dynamic Uplink channels and semi-static resource allocation Random Access limited control information, Channel (RACH) collision risk Table Transport channel characterized by how the data is transferred over radio interface. tive modulation and coding) with which the data are transmitted over the radio interface. The MAC layer performs the mapping between the logical channels and transport channels, schedules the different UEs and their services in both UL and DL depending on their relative priorities, and selects the most appropriate transport format. The logical channels are characterized by the information carried by them. The mapping of the logical channels to the transport channels is shown in Figure The physical layer at the enb is responsible for protecting data against chan- 15

36 1. Background Downlink or uplink Downlink only DTCH DCCH BCCH PCCH MTCH MCCH Logical Channels Transport Channels DL-SCH UL-SCH BCH PCH MCH Figure Example of mapping of logical channels to transport channels. nel errors using adaptive modulation and coding (AMC) schemes based on channel conditions. It also maintains frequency and time synchronization and performs RF processing including modulation and demodulation. In addition, it processes measurement reports from the UE such as CQI and provides indications to the upper layers. The minimum unit of scheduling is a time-frequency block corresponding to one sub-frame (1 ms) and 12 sub-carriers. The scheduling is not done at a subcarrier granularity in order to limit the control signaling. QPSK, 16QAM and 64QAM will be the DL and UL modulation schemes in E-UTRA. For UL, 64-QAM is optional at the UE. Each radio frame is 10ms long containing 10 sub-frames with each sub-frame capable of carrying 14 OFDM symbols. For more details on these access schemes, refer to [4]. Multiple antennas at the UE are supported with the 2 receive and 1 transmit antenna configuration being mandatory. MIMO (multiple input multiple output) is also supported at the enb with two transmit antennas being the baseline configuration. Orthogonal Frequency Division Multiple Access (OFDMA) with a sub-carrier spacing of 15 khz and Single Carrier Frequency Division Multiple Access (SC-FDMA) have been chosen as the transmission schemes for the DL and UL, respectively. 16

37 Mobility management Mobility management Power-Up LTE_DETACHED LTE_ACTIVE LTE_IDLE No IP address Position not known IP address assigned Connected to known cell OUT_OF_SINK IN_SINK IP address assigned Position partially known DL DRX period DL reception possible No UL transmission DL reception possible UL transmission possible Figure Mobility states of the UE in LTE. Mobility management can be classified based on the radio technologies of the source and the target cells, and the mobility-state of the UE. From a mobility perspective, the UE can be in one of the three states: LTE DETACHED, LTE IDLE, and LTE ACTIVE as shown in Figure LTE DETACHED state is typically a transitory state in which the UE is powered-on but is in the process of searching and registering with the network. In the LTE ACTIVE state, the UE is registered with the network and has an RRC connection with the enb. In LTE ACTIVE state, the network knows the cell to which UE belongs and can transmit/ receive data from the UE. The LTE IDLE state is a power-conservation state for the UE, where typically the UE is not transmitting or receiving packets. In LTE IDLE state, no context about the UE is stored in the enb. In this state, the location of the UE is only known at the MME and only at the granularity of a tracking area (TA) that consists of multiple enbs. The MME knows the TA in which the UE last registered and paging is necessary to locate the UE to a cell. 17

38 1. Background Transmission BW (MHz) Sub-frame duration 1.0 ms Sub-carrier spacing 15KHz Sampling frequency (MHz) FFT size Number of occupied sub-carriers CP length (µs) Normal , Extended 16.6 Table E-UTRA Numerology 1.3 LTE Technologies OFDM In the downlink, OFDM is selected to meet efficiently E-UTRA performance requirements. With OFDM, it is straightforward to exploit frequency selectivity of the multi-path channel with low complexity receivers. This allows frequency selective in addition to frequency diverse scheduling and one cell reuse of available bandwidth. Furthermore, due to its frequency domain nature, OFDM enables flexible bandwidth operation with low complexity. Smart antenna technologies are also easier to support with OFDM, since each sub-carrier becomes flat faded and the antenna weights can be optimized on a per sub-carrier (or block of sub-carriers) basis. In addition, OFDM enables broadcast services on a synchronized Single Frequency Network (SFN) with appropriate cyclic prefix design. This allows broadcast signals from different cells to combine over the air, thus significantly increasing the received signal power and supportable data rates for broadcast services. To provide great operational flexibility, E-UTRA physical layer specifications are bandwidth agnostic and designed to accommodate up to 20 MHz system bandwidth. Table provides the downlink sub-frame numerology for different spectrum allocations. Sub-frames with one of two cyclic prefix (CP) durations 18

39 LTE Technologies 1.3 may be time-domain multiplexed, with the shorter designed for unicast transmission and the longer designed for larger cells or broadcast SFN transmission. The useful symbol duration is constant across all bandwidths. The 15 khz subcarrier spacing is large enough to avoid degradation from phase noise and Doppler (250km/h at 2.6 GHz) with 64QAM modulation. The downlink reference signal structure for channel estimation, CQI measurement, and cell search/acquisition is shown in Figure Reference symbols (RS) are located in the 1st OFDM symbol (1 st RS) and 3 rd to last OFDM symbol (2 nd RS) of every slot. For FDD, it may be possible to reduce overhead by not transmitting the 2nd RS for at least low to medium speeds, since adjacent subframes can often be used to improve channel estimation performance. This dual TDM (or TDM) structure has similar performance to a scattered structure in 0.5 ms sub-frames, and an advantage in that low complexity channel estimation (interpolation) is supported as well as other excellent performance low-complexity techniques, such as IFFT-based channel estimators. To provide orthogonal signals for multi-antenna implementation, FDM is used for different TX antennas of the same cell, and CDM is used for different cells Multiple antenna techniques Central to LTE is the concept of multiple antenna techniques - often loosely referred to as MIMO - which take advantage of spatial diversity in the radio channel. Multiple antenna techniques are of three main types: diversity, MIMO, and beamforming. These techniques are used to improve signal robustness and to increase system capacity and single-user data rates. Each technique has its own performance benefits and costs. Figure illustrates the range of possible antenna techniques from the simplest to the most complex, indicating how the radio channel is accessed by 19

40 1. Background R0 R0 R1 R1 Frequency R0 R0 Frequency R1 R1 R0 R0 R1 R1 R0 l=0 l=6 l=0 l=6 R0 R1 l=0 l=6 l=0 l=6 R1 Slot Slot Slot Slot (a) Antenna port 0. (b) Antenna port 1. Figure Downlink reference signal structure - normal cyclic prefix, two transmit antennas the system s transmitters and receivers. Transmit antennas The radio channel Receive antennas Transmit antennas The radio channel Receive antennas (a) SISO system. (b) SIMO system. (c) MISO system. Figure (d) MIMO system. Radio channel access modes SISO The most basic radio channel access mode is single input single output (SISO) in which, only one transmit antenna and one receive antenna are used. This is the form of communication that has been the default since radio began and is the baseline to which all the multiple antenna techniques are compared. 20

41 LTE Technologies 1.3 MISO Slightly more complex than SISO is multiple input single output (MISO) mode, which uses two or more transmitters and one receiver. (Figure 1.2(c) shows only two transmitters and one receiver for simplicity.) MISO is more commonly referred to as transmit diversity. The same data is sent on both transmitting antennas but coded such that the receiver can identify each transmitter. Transmit diversity increases the robustness of the signal to fading and can increase performance in low signal-to-noise ratio (SNR) conditions; however, it does not increase data rates as such, but rather supports the same data rates using less power. Transmit diversity can be enhanced with closed loop feedback from the receiver to indicate the balance of phase and power used for each antenna. SIMO Figure 1.2(b) is single input multiple output (SIMO), which - in contrast to MISO - uses one transmitter and two or more receivers. SIMO is often referred to as receive diversity. Similar to transmit diversity, it is particularly well suited for low SNR conditions in which a theoretical gain of 3dB is possible when two receivers are used. As with transmit diversity, there is no change in the data rate since only one data stream is transmitted, but coverage at the cell edge is improved due to the lowering of the usable SNR. MIMO Finally the Figure 1.2(d) shows full MIMO, which requires two or more transmitters and two or more receivers. This mode is not just a superposition of SIMO and MISO since multiple data streams are now transmitted simultaneously in the same frequency and time, taking full advantage of the different paths in the radio channel. For a system to be described as MIMO, it must have at least as many 21

42 1. Background receivers as there are transmit streams. The number of transmit streams should not be confused with the number of transmit antennas. Consider the Tx diversity (MISO) case in which two transmitters are present but only one data stream. Adding receive diversity (SIMO) does not turn this into MIMO, even though there are now two Tx and two Rx antennas involved. SIMO + MISO MIMO. It is always possible to have more transmitters than data streams but not the other way around. If N data streams are transmitted from less than N antennas, the data cannot be fully descrambled by any number of receivers since overlapping streams without the addition of spatial diversity just creates interference. However, by spatially separating N streams across at least N antennas, N receivers will be able to fully reconstruct the original data streams provided the crosstalk and noise in the radio channel are low enough. One other crucial factor for MIMO operation is that the transmissions from each antenna must be uniquely identifiable so that each receiver can determine what combination of transmissions has been received. This identification is usually done with pilot signals, which use orthogonal patterns for each antenna. The spatial diversity of the radio channel means that MIMO has the potential to increase the data rate. The most basic form of MIMO assigns one data stream to each antenna. In this form, one data stream is uniquely assigned to one antenna. The channel then mixes up the two transmissions such that at the receivers, each antenna sees a combination of each stream. Decoding the received signals is a clever process in which the receivers, by analyzing the patterns that uniquely identify each transmitter, determine what combination of each transmit stream is present. The application of an inverse filter and summing of the received streams recreates the original data. The theoretical gains from MIMO are a function of the number of transmit and 22

43 LTE Technologies 1.3 receive antennas, the radio propagation conditions, the ability of the transmitter to adapt to the changing conditions, and the SNR. The ideal case is one in which the paths in the radio channel are completely uncorrelated, almost as if separate, physically cabled connections with no crosstalk existed between the transmitters and receivers. Such conditions are almost impossible to achieve in free space, and with the potential for so many variables, it is neither helpful nor possible to quote MIMO gains without stating the conditions. The upper limit of MIMO gain in ideal conditions is more easily defined, and for a 2x2 system with two simultaneous data streams a doubling of capacity and data rate is possible. MIMO works best in high SNR conditions with minimal line of sight. Beamforming Beamforming uses the same signal processing and antenna techniques as MIMO but rather than exploit de-correlation in the radio path, beamforming aims to exploit correlation so that the radiation pattern from the transmitter is directed towards the receiver. This is done by applying small time delays to a calibrated phase array of antennas. The effectiveness of beamforming varies with the number of antennas. With just two antennas little gain is seen, but with four antennas the gains are more useful. Obtaining the initial antenna timing calibration and maintaining it in the field are challenge. Combining beamforming and MIMO The most advanced form of multiple antenna techniques is probably the combination of beamforming with MIMO. In this mode MIMO techniques could be used on sets of antennas, each of which comprises a beamforming array. Given that beamforming with only two antennas has limited gains, the advantage of combining beamforming and MIMO will not be realized unless there are many 23

44 1. Background antennas. This limits the practical use of the technique on cost grounds Scheduling Scheduling controls the allocation of the shared resources among the users and often is jointed to link adaptation; if this happens we have a channel-dependent scheduling (Figure 1.3.3). In LTE a part of the scheduler is the rate adaptation, so it determines the data rate to be used for each link. Effective channel variations seen by the base station User #1 User #2 Effective channel variations seen by the base station User #1 User #2 Channel quality Channel quality #1 #2 #1 #2 Time (a) Time domain scheduling. #1 #2 #1 #2 #1 #2 Frequency (b) Frequency domain scheduling. Figure Channel-dependent scheduling. LTE has also, in addition to the time domain, access to the frequency domain, due to the use of OFDM. In other words, scheduling in LTE can take channel variations into account not only in the time domain, as HSPA, but also in the frequency domain and this meant that for LTE, scheduling decisions can be taken as often as once every 1 ms and the granularity in the frequency domain is 180 khz. 1.4 Handover In case of a handover the protocol endpoints that are located in the enodeb will need to be moved from the Source enodeb to the Target enodeb. Then, it is an option whether the full protocol status of the Source enodeb is transferred to the Target enodeb or the protocols are reinitialized after the handover. 24

45 Handover 1.4 This raises the question for example, whether the HARQ and ARQ window state is discarded and reset or transferred during the handover. Since, it would be overly complex and not always feasible to transfer the whole protocol state, it is an assumption in LTE that the RLC/MAC protocols are reset after a handover. The message sequence diagram of the LTE handover procedure is shown in Figure The figure shows both the control plane messages (black and blue solid arrows) and the flow of the user plane packets (purple dashed arrows). See also [4] for a description of the procedure. The UE sends measurement reports to the Source enodeb, which may decide on the execution of a handover. The Source enodeb requests the preparation of a handover at the Target enodeb. The Target enodeb can perform admission control to check whether the established QoS bearers of the UE can be accommodated in the target cell. Next, the Source enodeb sends the Handover Command to the UE, which includes all necessary information for the UE to access the target cell. At the same time the Source enodeb suspends the RLC/MAC protocols and it may start to forward the SDUs that have not yet been successfully sent to the UE toward the Target enodeb. That is, partially transmitted SDUs at the HARQ/ARQ layers will be forwarded along with the buffered and not yet transmitted SDUs, also including all the incoming SDUs from the GW. Whether SDU forwarding is employed at all by the enodeb is left as a vendor specific implementation detail. At the same time, the UE starts to execute the handover to reconnect at the Target enodeb. The UE needs to perform a random access procedure on the RACH (Random Access Channel) in the target cell. The UE also needs to get uplink time alignment assigned, which means that the enodeb has to measure on the uplink transmission of the UE (on the RACH) and determine the timing advance that the UE has to use for its uplink transmissions in order for 25