Multistage joint detection with decision feedback for CDMA mobile radio applications

This paper focuses on the performance of a multistage joint detection (MJD) scheme for CDMA mobile radio applications, which combines the relatively low complexity of a linear estimation with low bit error rates. Minimum-mean-square-error linear block estimation (MMSE-LBE) is applied in combination with decision feedback. Due to relatively low computational complexity of the LBE algorithm a large number of transmitted symbols can be reliably and cost effectively estimated simultaneously in a first detection stage. Since in the second stage the number of `unknown´ symbols is significantly reduced, the performance of MMSE-LBE is noticeable increased. Monte-Carlo simulations, which include mobile radio channel models, show that significant SNR gains of more than 25 dB can be achieved due to a second detection stage