A quadratically constrained decision feedback equalizer for DS-CDMA communication systems

Among multiuser detection techniques for DS-CDMA communication systems, decision feedback equalization (DFE) offers close to optimal steady state performance, but requires a training sequence and suffers from catastrophic error propagation. Minimum output energy (MOE) detection structurally allows blind adaptation but can be severely degraded from the optimal performance under mismatch situations. We propose to combine these two techniques so that close to optimal performance can be achieved without the use of a training sequence while mitigating the effect of error propagation. To achieve this goal, a quadratic inequality constraint is also incorporated. A recursive least squares (RLS) implementation of the proposed robust decision feedback equalizer is derived. Simulation results that illustrate the performance of the proposed detector as compared with standard MOE and DFE detectors are presented.