Study On MUD Algorithms In DS-CDMA

MUD is one of crucial techniques in CDMA system. The main work is as following: Firstly, a deep investigation on LMS and the way of improving LMS algorithm is made. Optimizing time-varying step factor is proposed, which is effective in promoting LMS-like algorithms. The modified step-varying decorrelation LMS is near to RLS in algorithm performance, and excellent BER performance is obtained when it is used to linear MUD.Secondly, a blind optimal MUD based on MMSE criterion (MMSE blind MUD) is proposed, and two implementations are introduced. The new MUD is different from Verdu's MUD due to the new MUD is blind optimal MUD. It is also different from MMSE MUD in other literature, which is based on weight vector iteration. Simulations verify the correctness of the new MUD.Thirdly, proposing two new algorithms. Genetic Engineering Optimal Algorithm and RSD optimizing algorithm can greatly decrease the computational complexity of the optimal MUD. And the detect accuracy is almost equal to that of totally searching algorithm. But the detect speed of the new methods is as more than 20 times high as that of the latter when there are only six users in the system. It is obvious that the effect will be more outstanding when the number of users is increased.Fourthly, proposing a new method of estimating DOA, based on train sequence. The new method of estimating DOA is fast to estimate, and high in accuracy. And then developing a new MUD based on spatial separation of signals. The new MUD can effectively separate the mixing signals of users at different places by using of T-shape antenna array without spatial ambiguity.Fifthly, proposing to simulate stochastic time-varying channel by using channel model stack. A new method of blind estimating channel based on MMSE criterion is developed. Simulations show the method is correct in theory. And pointing out the further work is investigating the coefficients of real channel.