- 學位論文
依染色體績效進行突變之改良型基因演算法於非同步直序式超寬頻通訊系統多用戶偵測之應用研究
Application of Ranking-Based-Mutation Genetic Algorithm to Multi-user Detection in Asynchronous DS-UWB System
摘要
超寬頻通訊中的多重路徑效應可利用耙式接收器降低其影響，而多用戶所造成之多重存取干擾（Multi-Access Interference, MAI）則可透過最佳多用戶偵測器來獲得最佳的系統效能。然而最佳多用戶偵測器因計算複雜度太高，故無法實現。為了在效能與複雜度之間取得平衡，若干研究在同步直序式超寬頻系統（DS-UWB）環境下提出了次佳之架構於基因演算法的多用戶偵測器（Genetic algorithm based multi-user detector, GA-MUD）。為符實際情況，本研究在非同步DS-UWB環境下使用觀察視窗（Observation Window）將鄰近位元干擾( Neighbor Bit Interference, NBI )擷取加入目標位元，接著應用以CSRAKE為前級之GA-MUD（Constricted Selection Rake GA-MUD, CSRGA-MUD）來探討系統效能。由於CSRGA-MUD 改善BER之幅度有限，因此我們進一步提出依染色體績效進行突變之改良型基因演算法（Ranking-Based Mutation GA, RBM-GA），並結合CSRAKE成為RBM-CSRGA-MUD來降低BER。在UWB通道下的模擬結果顯示RBM-CSRGA-MUD在較複雜之非同步通道環境中有更好的解空間搜尋能力，可改善BER之幅度在K=8時為19.26% ~32.76%，K=16時則為16.53%~23.1%。
並列摘要
Multi-path interference (MPI) and multiple access interference (MAI) are two major factors that degrade performance of multi-user ultra-wide band system (UWB). Generally, the effects due to MPI and MAI can be reduced by using a rake receiver and the optimum multi-user detector (OMU). Though optimum multi-user detector (OMUD) can achieve the best performance, its computational complexity is too high to implement. To overcome this difficulty, many sub-optimal multi-user detectors (MUDs), particularly the genetic algorithm-based multi-user detectors (GA-MUDs), have been proposed recently to compromise system performance and complexity for synchronous direct-sequence UWB system (DS-UWB). In this study an asynchronous DS-UWB system is considered to meet the realistic condition by applying an observation window for neighbor bit interference (NBI) evaluation. To enhance the global search capability of GA, a ranking-based mutation GA (RBM-GA) is presented and combined with constricted selection rake receiver (CSRAKE) to form a RBM-CSRGA-MUD. A series of simulations is conducted over the UWB channels of CM1~CM4. Experimental results show that the proposed RBM-CSRGA-MUD can achieve better performance than the CSRGA-MUD which consists of CSRAKE and conventional GA with BER improvements of 19.26%~32.76% and 16.53%~23.1% for K=8 and K=16, respectively.