Resource Allocation Algorithm In LTE Transmission Systems

Recent years have witnessed a growing demand for ubiquitous wireless multimedia service, such as VoIP, web browsing, video telephony, and video streaming. However, the time-varying nature of wireless channel makes it challenging to design an efficient cellular network to meet the requirements on delay and bandwidth of these applications. In order to solve these problems, Long Term Evolution(LTE) was devised by 3GPP, which represents an important milestone towards the so called 4G cellular networks. A key feature of LTE is the adoption of resource scheduling at the Base Station. That is to allocate time and frequency resources among multi-users by jointly considering time-varying channel condition and diversity Quality of Service(QoS) requirements.3GPP does not standardize the scheduling algorithm in LTE. Because the scheduling algorithm performs differently under different payload. It is necessary to focus on the specific communication scenario when designing the scheduling algorithm. In this dissertation, we will focus on the resource optimization problems in LTE system. The main research work can be summarized as follows:1. In LTE uplink resource allocation process, the buffer state of User Equipments(UE) is indicated by the index of Buffer Status Report(BSR). However, BSR is a lag-ging and inaccurate indicator, which may incur a waste of the radio resources. In order to improve the system utilization, we propose an algorithm to improve the accuracy of BSR. The buffer size with a large index is further divided into 64 levels which constitute a new table. One of the reserved bits in the head of Media Access Control(MAC) unit is used to indicate whether or not to look up the new table. Obviously, this allows for a further fine description of UE’s buffer state. Experimental results show that the proposed algorithm reduces the BSR errors without additional cost, meanwhile it improves the efficiency and utilization of radio resources.2. In LTE downlink transmission systems, in order to meet different QoS require-ments of users, we propose a buffer-aware adaptive resource allocation scheme by jointly exploiting the priorities of user queues and Resource Blocks(RBs) capac-ity. Different user queue has different priority which is calculated by remaining life time or overflow probability. By applying the large deviation principle, we can derive an overflow probability estimation model which does not require any prior knowledge of the network dynamics. This benefit is achieved by using the sliding window value and historical observation. Afterwards, the RBs are dy-namically allocated based on an online measurement algorithm in order to avoid queue overflow and provide QoS guarantee. Experimental results show that the proposed algorithm is able to improve system throughput while guaranteeing cer-tain fairness among users and reducing the average bit loss rate.3. In LTE multicast systems, Scalable Video Coding(SVC), combined with Modu-lation and Coding Scheme(MCS), can provide good video quality to users with good channel conditions while maintaining basic video quality for users with bad channel conditions. We formulate a joint optimization problem which considers selection of MCS, transmission power and RBs allocation to maximize the total video quality of all users. Firstly, we propose a dynamic programming to assign MCS and power for each layer in a single-session. We relax the constraints of original problem to create a dual problem by using Lagrange multiplier which is updated with a subgradient method. Secondly, in multisession we propose a RBs allocation algorithm based on dynamic programming. Simulation results demon-strate the effectiveness of the proposed schemes.4. In LTE scalable video unicast systems, based on proportional fairness, we pro-pose an algorithm to maximize the system throughput. We attempt to combine SVC with proportional fairness. We give priority guarantee on base layer for each session, then the enhancement layers scheduling is considered by using the pro-portional fairness algorithm. Since the original problem is a double-convex func-tion, in order to reduce the complexity, we divide the objective function into two sub-models:power scheduling model and RBs allocation model. By using the block coordinate descent algorithm, we can converge the result to a near optimal solution. Experimental results show that the proposed algorithm can maximize the system throughput while guaranteeing the fairness.