Resource Allocation In Multi-user MIMO-OFDM Systems With Dual Joint-optimized Objectives

With the increasing demand for data rate in modern communications,it is important to improve system throughput by allocating radio resources(time,frequency,power).The allocable resources dimensions have doubled due to the introduction of SDMA in MIMO-OFDM system,which makes the problem become more complex and difficult to be optimized.Current studies limit the free degree of resources in order to reduce complexity.Some studies stipulate that only one user can be accommodated on each subcarrier,giving up the spectral gain generated by SDMA.Others limit the power to be equally distributed over all sub-channels,which means abandoning the power gain acquired from different channels.On the other hand,modern society advocates green communications,it is also important that minimize the power consumption under a certain system throughput.The optimization of throughput and power are contradictory on the target.The existing studies focuse on optimizing one of them.Either guarantee the system throughput to minimize the power,or ensure the power to maximize the system throughput.As far as we know,no research focuse on the power-data rate double objective optimization.There are many reasons: Such as the existence of non-linear constraints cause the problem non-convex and difficult to be solved.The traditional greedy algorithm can not optimize power and throughput simultaneously.As a result,the only solution to the problem is exhaustive method.But the huge computation is not feasible owning to the big free degree of resources in MIMO-OFDM system.To address the problem above,this thesis proposes a power-data rate double objective optimization algorithm for the downlink of multi-user MIMO-OFDM system,which can minimize the total power consumption of the system while maximizing the system throughput.This algorithm assimilates the user selection of sub-carrier into the optimization of power and data rate.Thus,the complexity,which caused by big free degree of resources,can be reduced.Firstly,this thesis establishes a pattern of power-data rate double objective optimization.Due to the existence of non-linear constraints,this thesis designs a dual problem to solve the non-convexity of the original problem,transforming the original problem into its convex dual problem.Meanwhile,the duality gap between the original problem and its dual problem is proved to be zero,which means the two problems share the same optimal solution.Secondly,to solve the dual problem,this thesis proposes a decomposition and iteration method in order to reduce the complexity in a certain degree.Lastly,by controlling the accuracy of iteration,the algorithm can minimize the power consumption and maximize the system throughput as much as possible.Based on these steps,the algorithm can improve the efficiency of power and spectrum simultaneously.The simulation results show that,compared with the traditional single-objective optimization algorithm,the proposed simultaneous optimization algorithm performs better in terms of total power consumption and system throughput.