Artificial Bias Induction in Fourth-Order Cumulants Based Automatic Modulation Classification Algorithm in AWGN and Multipath Propagation Channel

Abstract

Automatic modulation classification (AMC) represents a wide used technique for modulation format recognition of signals considered to be a priori unknown. Due to the low algorithm and hardware complexity, AMC algorithms based on fourth-order cumulants are still very popular. Presence of bias in standard cumulants estimated values of real signals constellations has positive impact on classification score for distinguishing real from complex signals. Therefore, one new approach in AMC is proposed in this paper, with focus on manipulation with theoretical expected cumulant values of real signals constellations, assuming artificially introduced bias will improve AMC performance. Artificial bias induction is done through modifications of standard cumulants mathematical formula. Performance of modified and standard fourth-order cumulants based AMC algorithms were explored in context of real and complex signals constellations. This was done through Monte Carlo simulations in propagation conditions which included Additive White Gaussian Noise (AWGN) and multipath propagation channel with known and unknown impulse response. Evaluation was done through the probability of correct classifications. Presented numerical results confirmed superiority of algorithm based on artificial bias induction in classification of real and complex signals, in each considered propagation scenarios, especially in a radio environment with lower signal-to- noise ratio (SNR) values. The remarkable AMC performance enhancements are up to 25%.