本論文針對多載波分碼多工存取(Multi-Carrier Code Division Multiple Access, MC-CDMA)系統,應用基因演算法(Genetic Algorithm, GA)和模擬退火法(Simulated Annealing, SA)同時進行通道估計和多用戶偵測(Joint Channel Estimation and Multi-user Detection, JCEMUD)。聯合最佳化(Joint Optimization)的問題使用最大概似法(Maximum Likelihood, ML)做竭盡式搜尋(Exhaustive Search)固然可獲得最佳的效能,但因複雜度過高而難以實行,故一般文獻都採用次佳演算法來處理,諸如GA、SA、粒子群演算法(Particle Swarm Optimization, PSO)等。為降低運算複雜度,本研究應用GA同時執行通道估計和多用戶偵測,GA可以平行處理待解向量中的所有解,故可快速獲得最佳解。但傳統的GA易因早熟陷入區域最佳解,因此,我們在GA中結合SA之機率函數來執行GA之突變運算,以跳脫出區域最佳解,並提升GA的區域搜尋能力,增加其尋得全域最佳解之機率。實驗結果顯示,結合SA之GA,可有效改善傳統GA於JCEMUD的效能。
In this thesis, a hybrid approach that employs genetic algorithm (GA) and simulated annealing (SA) for joint channel estimation and multi-user detection (JCEMUD) is proposed for MC-CDMA systems. Using exhaustive search, the maximum likelihood (ML) approach can achieve optimal result in the joint estimation and detection problem, but its computational complexity is too high to be implemented. Therefore, suboptimal algorithms with less computational complexity, such as GA, SA and particle swarm optimization (PSO) are commonly adopted. In this study, we apply GA to deal with channel estimation and multi-user detection simultaneously. However, conventional GA usually converges prematurely to fall into the local optimal. In order to improve local searching ability of the conventional GA for better solution, a hybrid algorithm combining GA and probability function of SA for mutation operation is suggested. Experimental results show that the proposed approach is able to escape from the local optimum and thus achieves better performance than that of conventional GA scheme.