Multi Cell OFDMA System Resource Allocation And Channel Estimation

With the rapid development of wireless communications, the contradiction between thelimited wireless resources and the increasing requirements for Quality of Service (QoS)becomes more and more apparent. To solve this problem various technologies at differentlayers of wireless systems should be involved. In the physical layer, Orthogonal FrequencyDivision Multiplexing (OFDM) technology has proved to be very suitable for widebandwireless systems. In the data link layer, resource allocation and scheduling can be used toimprove the frequency efficiency and throughput of communication systems. Moreover,OFDMA provides flexibility for resource allocation due to its unique characteristic. Thus,resource allocation in OFDMA systems has attracted great attention in recent years.Orthogonal Frequency Division Multiple Access (OFDMA) is a multiple accesstechnology combining OFDM with Frequency Division Multiple Access (FDMA) and hasbeen considered as one of the key technologies in4G mobile communication systems. Basedon the OFDM modulation，OFDMA achieves the multi-user access by allocating subcarriersto every user independently. In the scenario of multi-user accessing simultaneously, OFDMAsystems choose resource allocation strategy according to the condition of wireless channel,QoS, etc, in order to realize optimal adaptive resource allocation. Then the system can notonly achieve higher spectrum utilization, but also better satisfy QoS requirements. Therefore,adaptive resource allocation is a very important research topic and is also one of the mainproblems to be solved in OFDMA systems.In this thesis，we researched on adaptive resource allocation in multi-cell OFDMA andproposed a resource allocation algorithm based on the penalty function and a low-complexityresource allocation algorithm. We also studied channel noise estimation and came up with anon-data-aided algorithm for joint measurement of interference and noise power based on themaximum likelihood criterion. Our contributions can be summarized in the following threeaspects:(1) In the multi-cell OFDMA systems, based on the centralized resource management,taking the interference between cells into consideration, the subcarrier and power allocationfor each cell is adjusted so as to minimize the total power. Discretize power to establish a multi-cell OFDMA model. Then the problem is simplified by penalty function and solved byan improved simulated annealing algorithm. In this thesis, a penalty function SA(Penalty-Simulation Annealing) resource allocation algorithm has been proposed to simplifythe model and reduce the complexity. Theoretical analysis and simulation results show thatthe selection of the number of discrete power is random within certain range. The proposedalgorithm can achieve better throughput with reduced complexity and the same overallperformance; compared with the multi-allocation algorithm, the unit power throughput issignificantly improved.(2) A low complexity resource allocation algorithm is proposed, which minimizestransmission power and in the mean time satisfies great transmission demand by solving a setof linear equations. The resource allocation process is carried out in two steps: sub-carrierassignment and power allocation. In the sub-carrier assignment, it is assumed that fixing thefrequency efficiency simplifies allocation. With fixed frequency efficiency, power allocationproblem can be solved through a set of linear equations. When the solution is negative, correctdistribution results are obtained by adjusting the algorithm. Simulation results show that thealgorithm achieves a compromise between the optimal algorithm and the multi-allocationalgorithm in throughput, power efficiency, complexity, etc.(3) A flat channel estimation algorithm based on Maximum Likelihood criterion isproposed. It is a non-data-aided algorithm, which carries out joint measurement ofinterference and noise power. Targeting at finite state discrete signals interfered by unknownwhite Gaussian noise, the algorithm first assumes that under certain conditions, anapproximate closed form solution can be obtained. Then with this solution as the initial value,using an iterative approach, a complete estimation result is obtained based on the maximumlikelihood criterion. Simulation results show that this method has the advantages of fastconvergence, high accuracy, etc.