In the last decades, mobile communications has evolved from being an expensive technology for a few selected individuals to today��s ubiquitous systems used by a majority of the world��s population. LTE system is the main trend of wireless communication system after 3G. The Long-Term Evolution(LTE) system is the main trend of wireless communication system after 3G. It is often called ��4G��, but many also claim that LTE release 10, also referred to as LTE-Advanced, is the true 4G evolution step, with the . rst release of LTE (release 8) then being labeled as ��3. 9G��. How to take full advantage of bandwidth to improve the throughput and enhance the quality of service becomes more and more important. The main objective of this study is to make a brief introduction of LTE and investigate a comprehensive analysis of physical layer of LTE. Then, comparison of the through-out and fairness between di. erent schedulers for physical layer will be provided to see how they would improve the performance of physical layer. Finally, simulation results based on di. erent scheduling algorithm would be post for us to make trade o.. IntroductionThe term ��Long Term Evolution�� (LTE) represents for the speci. ed technol-ogy on a novel air interface by 3GPP. Some of its targets include reduced latency, higher user data rates, improved system capacity and coverage and reduced cost of operation. In this report, the brief idea of 4G LTE, especially physical layer, will be summarized based on Release notes, books, and papers. 1. 1 DRIVERS FOR LTEThe evolution is driven by the creation and development of new services with advancement of the technology available for mobile systems. A prime driver for 4G LTE is the increasing need for Internet Protocol based services with a . xed broadband connection. The main service-related design parameters for a radio interface supporting a variety of services are:
Our writers will create one from scratch for
Data rate: Higher data rates for web browsing, streaming and . le transfer pushes the peak data rates close to Gbit/s for 4G.
Delay: Very low delay for interactive services.
Capacity: Total data rate on average from each deployed base station site and per hertz of licensed spectrum. Another driver and essential design parameter for 4G LTE is the demand for more spectrum resources to expand systems.1. 1. 1 The 3G Evolution to 4GIn 2004, a workshop was organized on the 3GPP Long-Term Evolution (LTE) radio interface. At the beginning half year, most time was spent on de. ning the requirements, or design targets. These were documented in a 3GPP technical report and approved in June 2005.  Most notable are the requirements on high data rate at the cell edge and the importance of low delay, spectrum . exibility and maximum commonality between FDD and TDD solutions. Work has since then continued on LTE, with new features added in each release, as shown in Figure1. 1. Figure 1. 1: Releases of 3GPP speci. cations for LTE. 1. 1. 2 Performance RequirementsTo achieve its goals, LTE must satisfy the following requirements:
Data rates: Up to 100 Mb/s within a 20 MHz downlink spectrum alloca-tion(2 Ch MIMO) and 50 Mb/s within a 20 MHz uplink(single Ch Tx) or, equivalently, spectral e. ciency values of 5bps/Hz and 2. 5 bps/Hz, respec-tively.
Throughput: The downlink average throughput per MHz: about 3-4 times higher than in the release 6. The uplink average user throughput per MHz: about 2-3 times higher than in the release 6.
Bandwidth: 1. 4 MHz-20 MHz in both paired and unpaired spectrum.
Mobility: Optimized for low terminals speeds(0-15 km/h). Connection maintained for very high UEs speeds(up to 500 km/h).
Coverage: The above targets should be met for 5 km cells. Some slight degradation allowed in throughput and spectrum e. ciency for 30 km cells. requirements. Figure 1. 2: LTE and its evolutionCHAPTER 2Overview and Channel Structure of LTE2. 1 OVERALL SYSTEM ARCHITECTUREA . atter all-IP, packet-based architecture for Core Network(CN) evolution in the Evolved Packet Core(EPC) and overall system architecture of both the Radio-Access Network (RAN) are the parallel project on the LTE radio-access technology in 3GPP. This work resulted in a . at RAN architecture. 2. 1. 1 Core NetworkThe Evolved Packet Core(EPC) is a radical evolution from the GSM and GPRS core network, which supports access to the packet-switched domain only, with no access to the circuit switched domain. Some di. erent types of nodes in EPC are illustrated in Figure 2. 1In addition to nodes listed in Table 2. 1, the EPC also contains other types of nodes such as Policy and Charging Rules Function (PCRF), and the Home Subscriber Service (HSS) node. 2. 1. 2 Radio-Access NetworkA single type of node-eNodeB, works as a logical node and not a physical implementation, is used as a part of . at architecture in 4G LTE RAN. A common question would be whether a base station is a implementation of eNodeB. TheTable 2. 1: EPC NodesNode Function ResponsibilitiesMobility Management Control-plane node of Connection and release of bearersEntity (MME) the EPC to a terminal, handling of IDLEto ACTIVE transitions, and han-dling of security keys. Serving Gateway (S- User-plane node con- A mobility anchor when terminal-GW) necting the EPC to s move between eNodeBs, and athe RAN mobility anchor for other 3GPPtechnologiesPacket Data Network Connects the EPC to Allocation of the IP address andGateway (PDN Gate- the internet quality of service enforcement forway, P-GW) a speci. c terminal. answer is it is a possible implementation of, but not the same as, an eNodeB.Figure 2. 2: Radio-access-network interfaces2. 2 RADIO PROTOCOL ARCHITECTUREThe RAN protocol architecture is shown in Figure 2. 3. A general overview of protocol architecture for the downlink is shown in Figure 2. 4. Uplink transmissions is quite similar to the downlink structure. Figure 2. 3: Overall RAN protocol architectureThe di. erent protocol entities of the radio-access network are summarized in Table 2. 2.Table 2. 2: The di. erent protocol entities of the radio-access networkProtocol Entity FunctionPacket Data Conver-gence Protocol (PD-CP) IP header compression, ciphering, integri-ty protection of the transmitted data, in-sequence delivery and duplicate removal for handoverRadio-Link Control (RLC) Segmentation and concatenation, retrans-mission handling, duplicate detection, and in-sequence delivery to higher layersMedium-Access Con-trol (MAC) Handles multiplexing of logical channels, hybrid-ARQ retransmissions, and uplink and downlink schedulingPhysical Layer (PHY) Handles coding and decoding, modulation and demodulation, multi-antenna mapping, and other typical physical-layer functions2. 2. 1 SchedulingScheduler, thought to be a part of the MAC layer though sometimes is more suitable to be treated as a separate entity, dynamic scheduling time-frequency resource-block pairs in uplink and downlink for users. However, uplink and down-link scheduling are independent of each other in 4G LTE. Resource blocks corre-spond to a time�Cfrequency unit of 1 ms times 180 kHz. In each 1 ms interval, the eNodeB takes a scheduling decision, and sends scheduling information to the selected set of terminals. The goal of di. erent scheduling algorithm is to take advantage of the channel variations between terminals and preferably schedule transmissions to a terminal on resources with advantageous channel conditions. The detail of scheduling will be discussed in later chapters. CHAPTER 3LTE Physical LayerNot surprisingly, as stated in previous chapter downlink and uplink of LTE Physical layer are quite di. erent. This is the result of the di. erence between eNodeB and UE in capabilities. Therefore, the features of physical layer will be described in the following sections. 3. 1 OVERALL TIME�CFREQUENCY STRUCTUREOFDM is used as the basic transmission scheme for both the downlink and u-plink of LTE physical layer. LTE subcarrier is carefully chosen to be 15 kHz, which provides a good balance between overhead from the cyclic pre. x against sensitivity to Doppler spread and shift and other types of frequency errors and inaccuracies. In the time domain, radio frames and subframes structure are illustrated in in Fig-ure 3. 1. Di. erent time intervals are multiple of Ts = 1/(15000 �� 2048). Di. erent cyclic-pre. x lengths, including seven and six OFDM symbols per slot respectively, may be used for di. erent subframes within a frame.The smallest physical resource in LTE is called a resource element, as illus-trated in Figure 3. 2, 3. 3. They are grouped into blocks that consists of 12 consec-utive subcarriers in the frequency domain and one 0. 5 ms slot in the time domain. Thus, 7 �� 12 = 84 resource elements for a normal cyclic pre. x and 6 �� 12 = 72 resource elements for an extended cyclic pre. x. The Table 3. 1 shows the LTE bandwidth and resource con. guration. Figure 3. 2: The LTE physical time�Cfrequency resourceBandwidth(MHz) 1. 4 3 5 10 15 20Number of RBs 6 15 25 50 75 100Number of occupied subcarriers 72 180 300 600 900 1200IFFT/FFT size 128 256 512 1024 1536 2048Subcarrier spacing(KHz) 15 15 15 15 15 15Table 3. 1: Bandwidth and Resource blocks speci. cations3. 2 DUPLEX SCHEMESLTE provides great . exibility in spectrum. It supports both FDD and TDD-based duplex operation. Time and frequency structures are shown in Figure 3. 4. Figure 3. 4: Uplink/downlink time�Cfrequency structure for FDD and TDD3. 2. 1 Frequency-Division Duplex (FDD)As can be seen in the Figure 3. 4, in the case of FDD, uplink transmission (fUL) and downlink transmission (fDL) is working under two carrier frequencies with ten subframes separately. For full-duplex capability, transmission and reception could occur simultaneously at a terminal, while for half-duplex capability, transmission and reception could not occur simultaneously. Meanwhile, the base station is always in full duplex capability. 3. 2. 2 Time-Division Duplex (TDD)As can be seen in the Figure 3. 4, in the case of TDD, uplink transmission and downlink transmission is working under same carrier frequency, and they are separated in the time domain. With the switch in the special subframe 1 or sub-frame 6 between uplink and downlink, some subframes are used to work for uplink transmission, while the rest subframe are used to work for downlink transmission. As seen in the Figure 3. 5, subframes 0 and 5 are always allocated for downlink transmission while subframe 2 is always allocated for uplink transmissions. The remaining subframes can then be . exibly allocated. Figure 3. 5: Di. erent downlink/uplink con. gurations in the case of TDDCHAPTER 4Scheduling ApproachesBased on the literature, most of the scheduling algorithms are developed for single carrier wireless systems. Further investigations are still required on the performance of these algorithms in multi carrier wireless systems.  We focus on the class of techniques that attempt to balance the desire for high throughput with fairness among the users in the system. The resource allocation is usually formulated as a constrained optimization problem, to either1. minimize the total transmit power with a constraint on the user data rate or2. maximize the total data rate with a constraint on total transmit powerThe . rst objective is appropriate for . xed-rate applications(e. g., voice), while the second is more appropriate for bursty applications like data and other IP appli-cations.  As more relevant to 4G LTE, we will focus on the second constraint, namely the rate adaptive algorithms. 4. 1 MAXIMUM SUM RATE ALGORITHMThe objective of the maximum sum rate (MSR) algorithm is to maximize the sum rate of all users, given a total transmit power constraint. The optimal solution is achieved when the goal is to get as much data as possible. However, the drawback is that the user with excellent channels, for example those who areclose to the base station, will be allocated all the system resources. Let Pk, l denote user k��s transmit power in subcarrier l. The signal-to-interference-plus-noise ratio (SINR) for user k in subcarrier l, denoted as SINRk, l, can be expressed asPk, lh2SINRk, l = ��K k, l j= 1, j. k, l + ��2 B= k Pj, lh2LUsing the Shannon capacity formula as the throughput measure, the MSR algo-rithm maximizes the following quantity: max log(1 + SINRk, l)Pk, l Lk= 1 l= 1KL�� withthetotalpowerconstraint ��BKL�ơ�Pk, l �� Ptotk= 1 l= 1When the total throughput in each subcarrier is maximized, the sum capacity is maximized as well. Thus, the original optimization problem is decoupled in simpler optimization problem in each subcarrier. The sum capacity in subcarrier l, denoted as Cl, can be written as:
KCl = log(1 +k= 1Pk, l
2tBPtot, l . Pk, l + h2k, lLwhere Ptot, l . Pk, l denotes other users�� interference to user k in subcarrier l. When the single user with the largest channel gain obtains all available power Ptot, l in subcarrier l, Cl is also maximized as a result of ��greedy�� optimization, which agrees with the intuition that give each channel to the user with the best gain in that channel.4. 2 MAXIMUM FAIRNESS ALGORITHMMaximum fairness algorithm is to maximize the minimum data rate, which is a Max-Min problem. In order to solve the problem that some users will beextremely underserved by maximum sum rate algorithm, even though the totalthroughput is maximized, the maximum fairness algorithm aims to allocate the subcarriers and power such that the minimum user��s data rate is maximized. This algorithm equalizing the data rates of all users, hence the name ��Maximum Fairness��. However, Maximum Fairness algorithm is not perfect. An obvious weakness is that the rate distribution among users is not . exible, and the total throughput is largely limited by the user with the worst SINR, as most of the resources are allocated to that user. 4. 3 PROPORTIONAL RATE CONSTRAINTS ALGORITH-MA more general form of the Maximum Fairness algorithm is the Proportional Rate Constraints (PRC) algorithm. Its goal is to maximize the sum throughput, with the additional constraint that each user��s data rate is proportional to a set of pre-determined system parameters (. k)K Mathematically, the proportionalk= 1. data rate��s constraint can be expressed as: R1 R2 RK
== … =
. 1 . 2 . K where each user��s achieved data rate Rk isL
��k, lBPk, lh2Rk = log2(1 + k, l )��2 BLl= 1 L and ��k, l can only be the value of either 1 or 0, indicating whether subcarrier l is used by user k or not. Clearly, this is the same setup as the Maximum Fairness algorithm if . k = 1 for each user. By varying the . k, any arbitrary data rates can be achieved. 4. 4 PROPORTIONAL FAIRNESS SCHEDULINGThe three algorithms discussed above make an e. ort to achieve a goal like the total sum throughput (Maximum Sum Rate Algorithm), equal data rates among all users (Maximum Fairness Algorithm), or pre-set proportional rates for each user (Proportional Rate Constraints Algorithm). Latency, thought to be the third element, enters the tradeo. in addition to fairness and throughput. Since even seconds of latencies are generally unacceptable, scheduling algorithms should balance latency and throughput and achieve some degree of fairness at the same time. The most popular framework for this type of scheduling is Proportional Fairness (PF) scheduling. The average throughput Tk(t) for all users is then updated according to: Tk(t +1) =
. .. ..
(1 . 1 )Tk(t)+ 1 Rk(t) k = k. tc tc(4. 1)= k.(1 . t1 c )Tk(t) k . where Rk(t), the instantaneous data rate that user k can achieve at time t; Tk(t), the average throughput for user k up to time slot t; tc, controls the latency of the system. If tc is large, then the latency increases, with the bene. t of higher sum throughput. If tc is small, the latency decreases since the average throughput values change more quickly, at the expense of some throughput. k., the user picked by the Proportional Fairness scheduler with the highest Rk(t) for transmission. InTk(t)the long-term, this means to selecting the user with the highest instantaneous rate relative to its mean rate. Since the user with the largest instantaneous data rate relative to its average throughput is selected by the PF algorithm, ��bad�� channels for each user are unlikely to be selected. On the other side, considering fairness, the user who has been underserved will receive higher priority in scheduling algorithm. Let Rk(t, n) be the support able data rate for user k in subcarrier n, at timeRk(t)slot t. Then for each subcarrier, the user with the largest is selected forTk(t) transmission. Let . k(t) denote the set of subcarriers in which user k is scheduledfor transmission at time slot t, then the average user throughput is updated as
11 Tk(t +1) = (1 . )Tk(t)+ Rk(t, n)tc tcn��. k(t)for k = 1, 2,… , K. 4. 5 PERFORMANCE COMPARISONIn this section, we brie. y compare the performance of the various schedul-ing algorithms that we have discussed, in order to gain intuition on their relative performance and merits. Table 4. 1 compares the four resource allocation algo-rithms for OFDMA systems. In summary, the Maximum Sum Rate allocation is the best in terms of total throughput, achieves a low computational complexity, but has a terribly unfair distribution of data rates, Hence, the MSR algorithm is viable only when all users have nearly identical channel conditions and a relatively large degree of latency is tolerable. The Maximum Fairness algorithm achieves complete fairness while sacri. cing signi. cant throughput, and so is appropriate only for . xed, equal rate applications. The Proportional Rate Constraints (PRC) algorithm allows a . exible tradeo. between these two extremes, but it may not always be possible to aptly set the desired rate constraints in real time. We also described the popular Proportional Fairness algorithm, which is fairly simple to implement and also achieves a practical balance between throughput and fairness.Algorithm Sum Capacity Fairness Complexity Simple? Maximum Sum Rate (MSR) Best Poor and in. exible Low Very simpleMaximum Fairness (MF) Poor Best but in. exible Medium see Proportional Con-straints (PRO) Good Most . exible High see Proportional Fair-ness (PF) Good Flexible Low see Table 4. 1: Comparison of Di. erent Scheduling AlgorithmCHAPTER 5Simulating the LTE Physical LayerResearch and development of signal processing algorithms for UMTS Long Term Evolution (LTE) requires a realistic, . exible, and standard-compliant simu-lation environment. We use a MATLAB-based downlink physical-layer simulator for LTE from Vienna University of Technology.  We will brie. y introduce the structure of this simulator in the next section. Realistic performance evaluations of LTE require standard compliant simu-lators. For that reason, commercially available simulators have been developed, for example   . The simulator currently implements a standard compli-ant LTE downlink with its main features being Adaptive Modulation and Coding (AMC), MIMO transmission, multiple users, and scheduling. Most parts of the LTE simulator are written in plain Matlab-code. 5. 1 SIMULATOR STRUCTURE5. 1. 1 Overall Simulator StructureAs mentioned in the above chapters, the LTE link level simulator should consist of these functional parts: eNodeB, transmitter, receiver, UEs, and channel model with adjustable delay. The elements of the simulator are shown in Figure 5. 1. The structure of the transmitter is illustrated in Figure 5. 2 Scheduler assign a set of Resource Blocks to UEs based on CQI from UEs with a subframe interval of 1 ms. The receiver structure is shown in Figure 5. 3. Each UE willFigure 5. 1: Overall simulator structure. Figure 5. 2: Structure of the LTE transmitter. 5. 2 SIMULATION RESULTS OF SCHEDULING ALGO-RITHMIn this section we present simulation results obtained with the above standard compliant LTE link level simulator implemented in MATLAB. Figure 5. 4 shows a sample aggregate UE results for proportional fair, as well as some cell-related statistics. For other algorithm, we could get the similar . gure. For the UE-related results, the UEs from which the results are obtained are the ones pertaining to any of the selected cells. Deactivated UEs (i. e. NaN values) are ignored. Scatterplot showing for each UE in the set the mapping between the wideband SINR and the throughput/spectral e. ciency. Since many points could be overlapping, there is the option of showing a binned (over wideband SINR) mean throughput mapping (in red). The resulting plots are shown in Figure 5. 5 5. 6, and depict a comparison between round robin, proportional fair, best CQI, max min, max throughput, and resource fair scheduling algorithms for a 2 �� 2 CLSM LTE setup. The mean value is marked in the CDF as a black dot. From the simulation result, we could . nd max min has best fairness perfor-mance and mean throughput. However, a weakness of the Maximum Fairness algorithm is that the rate distribution among users is not . exible. The total throughput is largely limited by the user with the worst SINR as stated previ-ously. Consider its weakness, we may think proportional fair scheduler has better performance with good fairness and large mean throughput. Note that best CQI scheduler has the worst fairness performance as predicted since it always assign the resources to the best CQI channel. CHAPTER 6Summary and ConclusionsEvolution of wireless communication has been expanded in recent years by in-troducing the 3rd generation of standards. This evolution is based on demands for higher data rate, lower latency, and better coverage resulting in user satisfaction. In this report, I make a brief introduction of the history of 4G LTE , in-cluding its overview, how it comes and the feature of standard. According to the , 4G LTE should satisfy the requirements on speci. c areas. This is shown in Chapter 1. After that, the overall and channel structure of LTE are discussed in Chapter 2. It provides an overview of the whole picture of LTE. In next Chapter, I focus on the Physical Layer. Speci. cally, I illustrate the frequency structure, duplex schemes, transport channel and downlink reference signals. In chapter 4, I summarize some most famous scheduling algorithm, which schedules the Resource Blocks in the physical layer. Good scheduling algorithm increases the throughput of physical layer while guarantees the fairness between UEs. In last chapter, I use the simulator to simulate the di. erent algorithm and compare the throughput and fairness. In reality, we should make trade-o. between suitable scheduling algo-rithms. Due to the evolution track toward 4G wireless communication standards, physical layer throughput can be further improved since new technologies.