Joint modulation classification and multi-user detection using multi-modulation sphere decoding

In this paper, we propose a simple yet effective modulation classification method for ML-based multi-user detection. We show that the proposed method, which is a simplification of Bayesian likelihood ratio test, can be implemented by modifying the sphere decoding algorithm. As a result, the modulation classification as well as the multi-user detection can be jointly achieved with minimal classification overhead. Simulation results in multi-user detection of high-speed downlink packet access (HSDPA) system show that the proposed method offers considerable performance gain over conventional RAKE and MMSE algorithms.