本文針對以多傳送天線多接收天線正交分頻多重接取系統，使用干擾排列技術，提出資源配置演算法讓系統吞吐量最大化。干擾消除被視為一個重要的技術，可以同時消除細胞間的干擾及不同使用者間的干擾，去加強整個系統的吞吐量。在執行干擾排列之前，系統必須選擇使用者組合來排列干擾。干擾排列之後，每一個使用者可以在沒有干擾的情況下接收到想要的訊號。由於設計干擾排列，將會把資源配置的問題變得與以往不同且困難。 我們針對干擾排列的架構，對每一個資源上的使用者的選擇做進一步的研究。首先，我們提出以干擾排列的設計為根基的方法，考慮接收端的束波成型向量，可以使得每個使用者得到比較好的吞吐量。 再者，我們設計兩種疊代的方案，更進一步的提升系統的表現。我們將使用者組合動態的調整以達到最大的系統吞吐量。模擬結果顯示，以干擾排列的設計為根基的方法比以往的演算法有更好的表現。此外，本文提出兩種方法利用疊代的概念可以接近最佳解，而計算複雜度也低於最佳解。且兩種方法的疊代次數是相當小的。 This paper considers resource allocation methods to achieve the maximum system throughput for a MIMO OFDMA system with an interference alignment technique. Interference alignment is considered as an important technique that could eliminate inter-cell interference and inter-user interference, and would enhance the system throughput. Before performing interference alignment, the system has to select paired user equipments that could align the interference signals. Consequently, each user equipment could receive the desired signal without interference. Owing to the design of interference alignment beamforming, the resource allocation problem becomes different and difficult. Based on the structure of interference alignment considered in this paper, the selection of the paired user equipments for each resource block is further investigated. This paper first presents the user equipment selection based on interference alignment. A proposed interference alignment-based selection scheme with a low computational complexity is developed by using the receive beamforming vectors, so that each resource block may be assigned to the paired user equipments that would have better system throughput. In addition, we design Sequential search scheme and Compete-and-compare scheme to further improve the performance. The paired user equipments are adjusted dynamically to achieve the maximum system throughput. Simulation results demonstrate that the proposed interference alignment-based selection scheme outperforms the existing algorithms. The performance of Sequential search scheme and Compete-and-compare scheme is very close to that of the optimal solution with an exhaustive search while the computational complexity is greatly reduced. The number of iterations in Sequential search scheme and Compete-and-compare scheme to obtain a solution is also pretty small.
