在本篇論文中,我們在多用戶正交多頻分工調變(OFDM)通訊系統的環境下探討資源的公平性分配。在以往的研究中藉由最大化最小(max-min)傳輸速率可以使得不同的使用者獲得相似的傳輸速率,但是此種方法並沒有考慮不同使用者的需求和通訊品質(QoS)要求上的差異。此外,系統傳輸速率與公平性分配並不可同時兼得,因此本篇論文探討在考慮使用者需求差異下,如何妥善地將有限的資源(如頻帶)有效率的分配以實現可調式公平資源分配並使系統整體傳輸速率最大化成為本篇論文重要的研究方向。本篇論文在傳統的最大化最小(max-min)最佳化題目中對不同的使用者加入權重係數以及參考傳輸速率以實現可調式公平資源分配,並且提出CSA-MOC 演算法改進Greedy 演算法的缺點。此演算法考慮使用者在通道分配上衝突的情況,並藉由最小化機會成本的決策進一步提升傳輸速率並且得到更為公平的資源分配結果。模擬結果顯示所提出的演算法與傳統資源分配方法比較後發現能得到更佳的傳輸速率以及公平性,而當用戶數增加或可分配資源減少時,CSA-MOC 演算法將更有效率的達到目的。
In this thesis, we consider the fairness issue in the resource allocation scheme for cellular downlink OFDMA systems. The conventional max-min fair approach simply leads to an equal rate result for each user. Consequently, the different rate demands among users are not well taken care of. On the other hand, the fairness and the overall throughput trade off. The fairness index is adjustable and the equal-rate fairness is one of the special case. In this thesis, the max-min weighted surplus rate fairness is presented. In this formulation, the adjustable fairness and different rate demands for individual user are considered. Since the problem is NP-hard, a resource allocation algorithm for general max-min problem is proposed. In this algorithm, we take care of conflicts in subcarrier assignments and design a penalty criteria to explore the opportunity cost, As result, better throughput-fairness region is achieved. The mathematical analysis for the proposed algorithm is devised. The effectiveness of the proposed algorithm has been verified numerically under various channel models with comparison to the conventional Greedy and Vogel's approaches.