Research Of New Subspace-based Method Of Blind Multiuser Detection And Channel Estimation In Dispersive CDMA Channels

Subspace梑ased high梤esolution methods play an important role in sensor array processing, spectrum analysis, and general parameter estimation. With the improve of chip梥peed in high? rate CDMA system, the 151 together MAT become the inherent problem and constituted the major impediment to the overall system performance. In this paper, firstly, to the case of short 151, It is discussed that modified algorithms are used to estimate the response of channel or signature blindly, which includes a signal subspace method or a noise subspace method, the affection of the nonorthogonality of signature in the estimation of response of multi梡ath channel. Secondly ,an algorithm of subspace training method to signature waveform estimation in multipath CDMA channel with colored noise is proposed. Finally, to do with the case of long JSI ,we cast the dispersive CDMA channel into a MJMO signal model framework. By applying the theory of blind MIMO channel identification and the desired user抯 signature waveform, we propose a subspace? based method of estimation of response of channel or signal detection. It is seen that we can perform signal detection blindly at any case of CDMA system.