以序的最佳化理論為基礎的多重使用者正交分頻多工系統的適應性次載波指定與位元分配方法

Abstract

下一世代的無線通信系統被期望能夠在無須考慮使用者移動情況與使用者所在位置的情況下，提供高的資料傳輸率的能力，以符合提供數位影音廣播與無線網際網路等資料存取服務的需求。要達到此目的會面臨的主要挑戰有幾項: (1)惡劣的無線通道環境，(2) 服務品質，例如資料錯誤率與使用者的資料需求量，(3)通信系統中資源的配置，例如功率消耗與頻帶資源的管理問題，(4)因為使用者的可移動特性所造成的通信系統狀態變化。最近，正交分頻多工系統受到熱切的關注且被認為適合被用於下一世代無線通信系統之中，因為它能夠有效率地傳輸資料，同時能夠抵抗各種傳播通道破壞並克服符間干擾問題。這個需求趨使我們去發展聰明且有效的資源管理方法，使得在滿足多重使用者的服務品質需求下，能夠即時的調配整個無線通信系統的功率與頻譜配置，以達到通信系統整體最有效率的運用。因此在本論文中，我們將提出兩個以序的最佳化理論為基礎的多重使用者正交分頻多工系統的適應性次載波指定與位元分配方法。 第一個方法採用序的最佳化理論為基礎的四層次演算法來求解一個足夠好的解。在前三個層次中，我們採用較粗略的模型來從候選的解空間中篩選出一些近似足夠好的可行解，直到選出l (=3)個足夠好的解為止。之後的第四個層次中，採用精確的模型來從這l個足夠好的解之中篩選出最佳的那一個當做我們所要尋找的足夠好的解。這個四層次演算法保證能夠獲得一個足夠好的解，但是其中的第一層次耗費很多的計算時間於求解連續性問題的最佳解，因此我們提出可以採用硬體電路為輔助的方法與架構，利用深次微米的技術優勢來快速計算出連續性問題最佳解。 在大維度多重使用者正交分頻多工系統問題之中，採用硬體電路來實現第一層次時會有晶片面積太大的問題，因此在我們提出的第二個以序的最佳化理論為基礎的演算法中，我們採用一個近似目標函數值的計算模型，配合遺傳演算法來從可能的解空間中，快速地搜尋出I (=200)個足夠好的解。 經由數值模擬以及與其他現行的方法做比較的結果可知，我們所提出的兩個方法在解的優質性與計算效率上都很成功，除了有效改善通信系統的功率效益之外，其中的一個方法可藉由硬體電路的輔助來達到即時應用之需求，而第二個方法很適合用於大維度多重使用者正交分頻多工系統問題之中。

The next generation wireless communication systems are expected to provide high rate transmission in the applications of digital audio broadcast, digital video broadcast and wireless internet access but regardless to the users’ mobility and location. The major challenges we are confronted with include the harsh channel conditions, QoS (Quality of Services) requirements such as BER (Bit Error Rate) and users’ data rate request, scarce resources such as power and spectrum, and the knowledge of the most updated state of the mobile users or devices. Orthogonal frequency-division multiplexing (OFDM) technology is recently recognized as one of the leading candidates for supporting the next generation wireless communication systems due to its ability to combat inter-symbol-interference (ISI) over harsh channel conditions. This stimulates the development of both intelligent and efficient resource management algorithms to achieve efficient utilization of power and spectrum while providing QoS requirements in the multiuser OFDM communication system. Therefore in this dissertation, we will present two computationally efficient methods to solve the Adaptive Subcarrier Assignment and Bit Allocation (ASABA) problem of multiuser OFDM system using Ordinal Optimization (OO) approach. Our first method consists of four OO stages to find a good enough solution to the ASABA problem. In the first three stages, we use surrogate models to select a subset of estimated good enough feasible solutions from the candidate solution set so as to reduce the search space of subcarrier assignment until l (=3) good enough subcarrier assignment patterns are obtained. Then in the fourth stage, we use the exact objective function to evaluate the l subcarrier assignment patterns, and the best one associated with the corresponding optimal bit allocation is the good enough solution that we seek. The four-stage OO approach ensures the quality of the obtained solution, however at the cost of solving a continuous version of the considered problem in the first stage. To resolve this computational complexity problem, we propose a hardware implementable Dual Projected Gradient (DPG) method to exploit deep submicron technology so as to obtain the optimal continuous solution extremely fast. Due to the large dimension of the ASABA problem, implementing the first stage in hardware is almost impossible for area concern. Therefore in the first stage of our second method, we develop an approximate objective function to evaluate the performance of a subcarrier assignment pattern and use a genetic algorithm to efficiently search through the huge solution space to find I (=200) good solutions. Numerical results and comparisons with various existing algorithms are provided to demonstrate the potential of the proposed techniques. It is shown that the proposed resource allocation methods substantially improve the system’s power efficiencies and are more computationally efficient. Moreover, the first method can meet the real-time application requirement and the second method is suitable for large dimensional ASABA problems.