Partially adaptive multiuser detection

The problem of reducing the complexity of a blind adaptive multiuser detector is addressed. The original detector is based on constrained optimization where the constraint is constructed from knowledge of the desired user´s code and timing. Since the number of adaptive weights in the constrained optimization detector is proportional to the code length, this fully adaptive algorithm may be too complex for real-time implementation. In order to reduce the number of adaptive degrees of freedom and, hence, the complexity of the detector, multiple constraints constructed from knowledge of the interfering users´ codes may be specified. In scenarios where the information necessary to specify multiple constraints is not available, an alternative technique for rank reduction based on subspace selection is presented. A special implementation for this partially adaptive constrained optimization detector is also proposed. This implementation facilitates the subspace selection technique and has advantages over other proposed implementations. Simulation results that illustrate the detector´s performance are presented in the form of bit error rate curves