在OFDMA 系统中，根据发端获得的信道状态信息（CSI）为各个不同用户合理分配共享的子载波和发送功率资源，对于优化系统性能至关重要。然而，在实际工作情况下，由于CSI 的获取存在噪声、检测误差、传输延迟和反馈传输速率受限等因素的影响，发端所获得的CSI 是不完全的部分CSI。针对部分CSI 下的OFDMA 系统资源分配问题，既往研究多依靠减小CSI误差或采用基于门限值的量化和反馈策略从各自关注的方面对系统性能作次优意义上的改善，尚不能形成统一的科学理论以指导实践。本项目根据真实的CSI 相对于发端所获得的部分CSI 的条件概率分布关系，提出OFDMA系统在多种重要应用场景下的资源分配最优策略和与最优性能接近的低复杂度技术方案。在统计平均意义下，该方案可以反映在给定CSI 限制情况下系统所能达到的最佳性能，因而可为实际OFDMA 系统最佳资源分配方案的选用提供理论和关键技术上的指导。

Since the OFDMA systems divide the entire channel into multiple orthogonal narrowband flat fading subbands (subcarriers), how to adaptively allocate OFDMA resources to enable efficient use of limited spectrum, becomes a critical issue for the mobile broadband systems. In OFDMA systems, if the transmitter has perfect channel state information (CSI), it can exploit the CSI to optimize the system performance. However, due to CSI estimation error, delay, and limited feedback rate, the transmitter may only obtain imperfect CSI. In this case, if the transmitter can obtain the probability density function (pdf) of actual CSI conditioned on the obtained imperfect CSI, it can still optimize the system performance in the average sense. In this project, we first derive the pdf of actual CSI conditioned on the obtained imperfect CSI. Based on the derived conditional pdf, we study resource allocation schemes for OFDMA systems in several important scenarios with imperfect CSI. The proposed resource allocation schemes enable the OFDMA system to be robust against the CSI estimation error, delay, and limited feedback rate. The proposed schemes may provide theoretical and technical guidance to the practical OFDMA resource allocation issues.