Research On The Downlink Scheduling Algorithms For Long Term Evolution Communication System

Nowadays, Long Term Evolution (LTE) is becoming the most popular4th generation communication standard all around the world. The commercial deployment of LTE is considered as the milestone of the era of wideband communication systems. Being the most important communication standard that following the trend of Mobile Internet, the3rd Generation Partnership Project (3GPP) has designed the LTE as a pure IP system. For this reason, the LTE has two outstanding changes comparing with other mobile communication standards:firstly, the flat network architecture which makes the evolved NodeB (eNB) as the only node network element in the wireless access networks; secondly, deploying the pure packet switch (PS). The first change reduces the latency in control plane and strengthen the decisive position of eNB in downlink communicate control. The second change unifies the transmission form and exchange method of voice and data services, but meanwhile raising the difficulty in scheduling algorithm design. Therefore, how to design an efficient downlink scheduling algorithm that could guarantee the traffics’quality of service (QoS) is an urgently problem to solve.With the introduction of multiple input multiple output (MIMO) and orthogonal frequency division multiplexing access (OFDMA) technologies in LTE, the concept of scheduling has been expanded from "time domain scheduling" to "jointly time domain and frequency domain scheduling". In other words, within LTE communication system, the scheduling algorithms not only responsible for deciding the priority of services, but also for determining how to allocate the resources and how to configure the relevant parameters of user element (UE). Hence, the LTE downlink scheduling algorithm has the features of multiple information dimensions and high requirement in real time. However,3GPP doesn’t standardize the scheduling algorithm in LTE. This is for the reason that, scheduling algorithm performs differently under different payload. Therefore, the design and optimize of scheduling algorithm that focusing on the specific communication scenario is necessary.In this dissertation, we propose three scheduling algorithms for single user scenario, multi-user scenario and discontinuous reception (DRX) activated transmission scenario, respectively. The performance of all the algorithms are verified by analytical analysis and computer simulations methods. To be specific, the main contribution of this dissertation are listed as follows:1)We concentrates on the problem of how to schedule a great quantity of heterogeneous traffic for one mobile node in LTE. To solve such problem in High Speed Rail (HSR), we present a decoupled time/frequency domain packet scheduler based on the analysis of HSR MIMO channel properties. In time domain, a novel single-user adaptive EXP/PF algorithm is adopted. In frequency domain, a fast exhaustive search-based frequency resource allocation algorithm (FRAA) named SU-OP is proposed, which can utilize the scarce frequency resources with constraint of LTE practical limits. Furthermore, we also introduce an alternative FRAA named SU-Greedy which has lower complexity and could obtain similar performance while the SNR is not high. By our original calculation rule the complexity of both algorithms is reduced further and thus more practical. Based on the system simulation, we find that the proposed scheduler has better delay and throughput performance than the well-known joint time domain and frequency domain proportional fair (TD-FD-PF) scheduler under HSR scenario.2) We propose a traffic-based Queue-aware scheduling (TQS) algorithm for evolved NodeB’s (eNB’s) MAC scheduler in3GPP LTE broadband wireless networks. The proposed TQS can be divided into three sub-algorithms:Firstly, we propose a traffic model construction (TMC) algorithm which can construct a discrete-time Markov-modulated Poisson process (dMMPP) for representing each flow. Secondly, a newly traffic state estimation (TSE) algorithm is designed to obtain the queue’s analytical statistics. Thirdly, based on the derived results of TSE and the channel states, we present a scheduling action decision (SAD) algorithm that can adaptively allocate bandwidth to flows by considering both queue states and spectrum efficiency. The simulation results show that the TMC and TSE algorithm can capture the fluctuation of traffic and queue accurately. Moreover, compared with a widely accepted traffic-based scheduling algorithm, the proposed TQS has better average queue length and overflow probability performance.3) Considering the self-similar property exhibited by Mobile Internet, the truncated-Pareto distributed arrival traffic model has been introduced into the LTE-DRX analytical framework for the first time. On this promise, a DRX analytical model is established based on the Discrete Time Semi-Markov Process (DTSMP) theory, which allowing us to evaluate the performance of DRX operations under certain parameter configuration precisely. To deploy it in practical, we put forward an on-line power saving strategy (OPSS) for the purpose of higher energy efficiency. The OPSS includes two sequential phases, namely, estimation phase and optimization phase. In the first phase, several statistical estimators are derived to capture the fluctuations of traffic conditions and DRX operations. By both mathematical derivation and numerical results, we prove that these estimators could unbiased estimate the statistics within just one second. In the second phase, on the basis of obtained statistics, the DRX parameters’configuration are optimized by considering the trade off between the packet delay and power saving. Solid simulations are conducted to verify the accuracy of the proposed model as well as the efficiency of OPSS. The well-matched results demonstrate the analytical model is correctly derived. Furthermore, it is observed that the OPSS outperforms the conventional LTE DRX mechanism in terms of both energy conservation and packet delay.