Dual-decomposition-based Resource Allocation Algorithms In OFDM Systems

This thesis focuses on applying the dual optimization and decomposition theory to the resource allocation in OFDM systems.And some algorithms are developed heuristically.The resource allocation in OFDM systems attracts many interests in the past decade. In this field,many problems are formulated,such as rate adaptive problem,margin adaptive problem and fairness problem,et al.For these problems,some mathematic theories,such as linear programming,convex optimization and game theory,are used to obtain optimal solutions and theoretic understanding.As the WiMAX emerged as a practical system in the recent years,it becomes an important task to develop lowcomplexity allocation algorithms for practice.Though there has been already much effort on this,the available practical algorithms have some common shortages.One is the lack of solid theoretic foundation.Another is that so much simplification was made that the performance was degraded significantly.Hence we draw the conclusion that the conflict between the performance and complexity is still a big problem.This thesis applys the dual decomposition theory to the allocation problem in OFDM systems,and provides a completely novel method for practical algorithm development. Firstly,the quasi-convex analysis is used as the foundation of applying the dual method to a category of allocation problems.Then an optimal solution is obtained.Based on the analysis and understanding of the optimal solution,we make reasonable simplifications and derive some practical algorithms,which are efficient and with low-complexity.For the rate adaptive problem,we propose a jointly subcarrier and power allocation algorithm to guarantee average data rate requirements.Utilizing the time coherence in wireless channel,the algorithm reduces the iterations in the optimal solution to get low-complexity.Also this make it possible to implement the algorithm in a distributed architecture.Based on local information,the users can decide whether to feedback a certain subcarrier to the base station,which reduces the feedback overhead.At the same time,the algorithm's characteristic of joint optimization brings high efficiency.The simulation is used to evaluate the algorithm.The same methodology can be used to solve the resource allocation problem to guarantee proportional fairness.However,there is some difference.The allocation of subcarrier and power are coupled more tightly,which will remarkably slow the convergency progress of the iterations.To solve this problem,we use average data rate instead of the instantaneous date rate.The average window size,and other parameters,will affect the algorithm's performance.This thesis studies these parameters carefully and provides optimal design.The allocation algorithm to guarantee instantaneous data rate requirements is also considered in this thesis.A method different with the above is used to develop the algorithm.We observe the optimal solution and get a new explanation,by which a highly efficient algorithm with low-complexity is presented heuristically.The algorithm provides high throughput as well as low outage probability.The methodology used in this thesis is innovative and important for the development of allocation algorithms in OFDM systems.And the algorithms presented in this thesis are very meaningful for practice.