Kalman filter based MMSE receiver for the DS-CDMA system in a Rayleigh fading channel

Abstract

MMSE based detectors are studied as receivers for DS-CDMA systems for their ease of implementation and their ability to be implemented with adaptive algorithms. The adaptive algorithms helps the receiver in adjusting to the changing statistical characteristics of the transmission. LMS (NLMS) and RLS are the two widely used adaptive algorithms. They are optimum in minimizing the mean square error in the non-stationary environments. Furthermore, when there is a relative motion between the transmitter and the receiver, the channel is modeled as a fading channel. Fading affects the transmitted signal by changing its amplitude and phase. This in turn plays havoc with the detector. The receiver not only makes wrong decisions about the transmitted bits (because of the induced phase) but also loses track of the desired user in a multiuser system. The first problem as mentioned above can be solved by using differential encoding and decoding. In addition to the differential encoding/decoding, two other methods are used to overcome the fading related issues in the MMSE receivers. These include an explicit estimation of the induced phase by using prediction filter and the use of modified adaptive algorithms based on differential detection technique. These techniques also help in preventing the tracking failure to a certain extent. The aim of this thesis is to develop a Kalman filter (KF) based MMSE detector which would be immune to the problems faced by receivers in the flat Rayleigh fading channel. The purport is that we may do away with the extraneous phase estimator block and the differential detection technique. In previous works, KF had been implemented either as a multiuser MMSE detector in non fading channels or for the estimation of fading channel itself. It is shown here that the Kalman alter performance is much superior to both the unmodified NLMS and RLS algorithms for a DS-CDMA system in Rayleigh fading channel. KF also locks the receiver to the desired user signal in the multiuser and near-far environment.