Classification of BPSK and QPSK Signals Using an Antenna Array

Abstract

We propose in this paper the use of a uniform linear array to improve the classification accuracy of two phase-shift keying (PSK) signals at different multiplicities, in particular, binary PSK (BPSK) and quadrature PSK (QPSK) signals. The proposed modulation classifier is based on the likelihood ratio test, where an additive white Gaussian noise with direct line-of-sight transmission is assumed. The likelihood functions of the two PSK signals depend on an unknown phase shift due to the signal arrival angle and the spatial separation of the antenna elements. A computationally efficient two-step processing is proposed to estimate the phase shifts to be used in the likelihood ratio test. The proposed phase shift estimation method reduces the computation significantly compared to the maximum likelihood solution, which requires numerical exhaustive search, and has an estimation accuracy reaching the Cram\'er-Rao lower bound at moderate signal-to-noise ratios. The proposed classifier improves the performance considerably compared to a single element receiver. Furthermore, the proposed classifier demodulates the transmitted symbol sequence in conjunction with the classification process, and no separate demodulation for the data symbol is needed after classification. Simulation results for BPSK and QPSK classification are provided.