上行正交分頻多重存取(Orthogonal Frequency Division Multiple Access, OFDMA)系統中的多用戶載波頻率偏移(Carrier Frequency Offsets, CFOs)是造成系統效能惡化的重要因素。快速進化規劃法(Fast Evolutionary Programming, FEP)是一種可用來估計CFOs的最佳化演算法,但因缺乏權重(Inertia Weight)概念,故在高維度解空間下其效能有待改進。禁忌搜尋 (Tabu Search, TS)演算法有較佳的區域搜尋能力,可經由不斷重複搜尋鄰近解,尋得最佳解,但缺點是容易陷入區域最佳解。因此,本研究整合FEP與TS,利用FEP所擁有良好的全域搜尋能力,為TS提供合適的初值,以增進搜尋效果,此一新的演算法稱為FEP-TS (Fast Evolutionary Programming-Tabu Search)。 此一混合型演算法經測試函數驗證,其效果優於FEP演算法,接著我們進一步以此演算法估計CFOs。模擬結果顯示,FEP-TS可以獲得優於FEP的效能。
Carrier frequency offsets (CFOs) caused by multiple users is a major factor that degrades the performance of orthogonal frequency division multiple access (OFDMA) system. The fast evolutionary programming (FEP) is a global search algorithm applicable for CFOs estimation. However, it performance is limited in high dimensional solution space because it lacks the help of inertia weight. The Tabu Search (TS) technique is a powerful local search algorithm which finds the optimum solution by iteratively searching the neighbor solutions. However, TS can be easily trapped in the local optimum. Therefore, this research combines the advantages of FEP and TS in global search and local search, respectively, to develop a new optimization approach called FEP-TS. With this approach, FEP provides suitable initial solution for TS to exploit better result. The proposed FEP-TS has been verified by using benchmark functions and the result shows its superiority as compared to the FEP. Then, the FEP-TS is employed to estimate the CFOs of OFDMA system and the simulation result illustrates that FEP-TS attains better performance than that of the FEP.