Digital modulation recognition based on constellation diagram

Abstract

A new classification algorithm based on constellation diagram is proposed to overcome the weakness of traditional algorithms by witch few digital modulation types can be recognized. The blind equalization is first used to mitigate effects of multipath and system synchronization errors. Then subtraction-clustering clusters signal spatial distribution and the resulting centers are extracted to match standard constellation patterns. Thus, MASK, MPSK and MQAM signals can be classified. The computer simulations indicate that the constellation diagram is a stable and robust signature.