Subblock Equalization and Code Averaging for DS-CDMA Receivers

As third-generation (3G) cellular communication systems evolve to higher data rates, channel equalization becomes an important tool in maintaining receiver performance in dispersive channels. Because these systems employ direct-sequence code-division multiple-access (DS-CDMA), they pose unique challenges in equalizer design. In this paper, linear equalization (LE) and decision feedback equalization (DFE) designs are developed based on subblock equalization, in which a limited form of maximum-likelihood (ML) joint detection is used to detect symbols within a subblock. A pseudo-ML formulation is used, treating interference from symbols outside the subblock of interest as noise. This formulation leads to a feedforward filter (FFF) whose coefficients depend on the spreading codes of the symbols. These spreading codes change every symbol period, due to long-code scrambling, requiring constant recomputation of the FFF coefficients. To reduce complexity, code averaging is used, resulting in a code-independent periodically varying transversal filter. Another design is obtained by constraining the transversal filter to be time invariant. Simulation results for the high-speed packet-access uplink show that code averaging introduces a small loss in subblock equalization performance and that interesting gains over conventional LE can be obtained.