Reduced-rank advanced RAKE receivers for dispersive CDMA signals

CDMA systems need simultaneous multiple access interference suppression and adaptive interference suppression filters that may span three symbols. Thus, a large number of filter coefficients need to be estimated. By the use of reduced rank filtering, it is possible to lower the number of required filter coefficients with a small decrease in performance. M. Honig (see Proc. IEEE Asilomar Conf. Signals, Syst. Comput., p.1106-10, 1998) made a successful attempt to develop a reduced rank algorithm for non-dispersive CDMA signals based on the multistage Wiener (MSW) filter of J. Goldstein et al. (see IEEE Trans. on Information Theory, vol.44, no.7, p.2943-59, 1998). Motivated by MSW, we propose reduced rank advanced RAKE receivers for dispersive CDMA signals. The proposed receivers are blindly implemented in a lower dimensional space, relative to the full-rank receivers, without the aid of training sequences and channel information. By exploiting the structure of the user signature waveform, the proposed receivers exhibit performance close to that of the reduced rank minimum mean square error (MMSE) receiver implemented with the desired user´s known channel information.