Complexity reduction of signal detection by exploiting correlation characteristics of spreading sequences

In this paper, we propose an approach to reduce the multi-user detection (MUD) complexity based on user grouping and signal replica classificatin by exploiting the correlation characteristics of spreading sequences in multipath fading channels. The spreading sequences are constructed from inter-group complementary codes with a sparse and regular correlation matrix and inherit its attractive auto/cross-correlation properties. Users are first partitioned into independent user groups according to whether or not there is interference among them, and then the replicas of user signals from the same user group are further classified into independent replica classes. The MUD is carried out within each low-dimensional user group or replica class, respectively, reducing the MUD complexity substantially. This approach can be applied to most of the existing MUD algorithms for complexity reduction, and in this paper optimal MUD and multi-stage MUD are exemplified. The analytical and simulated results demonstrate that this approach can reduce the MUD complexity significantly under any load conditions without performance loss.