| Summary: | 碩士 === 長庚大學 === 電機工程研究所 === 90 === The maximum likelihood detection of superimposed signal in direct sequence code division multiple access (DS-CDMA) has a computational complexity, increased in exponential with numbers of active users, and its implementation is prohibitive even for a moderate number of users. Some research has applied expectation maximization (EM) algorithm to decompose the multi-user detection problem into a series of single-user problems, and presents an iterative computational efficient algorithm for detection of signals in synchronous CDMA system. However the EM-based detector estimates each user’s signal to interference ratio (SIR) only by the property of cross-correlation of different users, but it does not reflect the actual SIR of each user. So，we propose a new algorithm called AISI algorithm ( Approximate ISR estimator by Side-lobe Information algorithm ) for improving the EM-based detector , The AISI algorithm can estimate signal to interference ratio (SIR) of different users directly by each user’s despreading filter output without increasing additional hardware circuit,. Within estimated SIR, new detector changes its parameter dynamically to increase the performance of detection. Our new detector also adds the novel idea of improved PIC that has different weight in each stage. By combining these two opinions，New detector changes its parameter dynamically in each stage to increase performance .At the end of this paper; we compare several kinds of multi-user detector. Simulation result also shows that when near-far effect occurs or not, the new detector still can have better performance than others.
|
|---|
