Likelihood-Based Modulation Classification for Multiple-Antenna Receivers

Abstract

Prior to signal demodulation, blind recognition of the modulation scheme of the received signal is an important task for intelligent radios in various commercial and military applications such as spectrum management, surveillance of broadcasting activities and adaptive transmission. Antenna arrays provide spatial diversity and increase channel capacity. This thesis focuses on the algorithms and performance analysis of the blind modulation classification (MC) for a multiple antenna receiver configuration.

For a single-input-multiple-output (SIMO) configuration with unknown channel amplitude, phase, and noise variance, we investigate likelihood-based algorithms for linear digital MC. The existing algorithms are presented and extended to SIMO. Using recently proposed blind estimates of the unknown parameters, a new algorithm is developed. In addition, two upper bounds on the classification performance of MC algorithms are provided. We derive the exact Cramer-Rao Lower Bounds (CRLBs) of joint estimates of the unknown parameters for one- and two-dimensional amplitude modulations. The asymptotic behaviors of the CRLBs are obtained for the high signal-to-noise-ratio (SNR) region. Numerical results demonstrate the accuracy of the CRLB expressions and confirm that the expressions in the literature are special cases of our results. The classification performance of the proposed algorithm is compared with the existing algorithm and upper bounds. It is shown that the proposed algorithm outperforms the existing one significantly with reasonable computational complexity.

The proposed algorithm in this thesis can be used in modern intelligent radios equipped with multiple antenna receivers and the provided performance analysis, i.e., the CRLB expressions, can be employed to design practical systems involving estimation of the unknown parameters and is not limited to MC.