TY - JOUR
T1 - Multichannel quasi-/constant modulus signal detection without spatial interference prior
AU - Wang, Tao
AU - Liang, Junli
AU - Liu, Weijian
AU - So, Hing Cheung
AU - Hao, Chengpeng
AU - Li, Yitong
AU - Gu, Ming
N1 - Publisher Copyright:
Copyright © 2026. Published by Elsevier Inc.
PY - 2026/11/15
Y1 - 2026/11/15
N2 - Constant modulus (CM) signals are widely adopted in radar and communication systems, making their reliable detection in the presence of spatial interference a fundamental problem. However, most existing multichannel signal detectors require either prior knowledge of interference directions (IDs) and powers or secondary data to compute the interference-plus-noise covariance matrix. In this paper, we devise a CM signal detector without prior knowledge of spatial interference, and formulate the problem within the generalized likelihood ratio test framework. Inspired by the ideal scenario with known interference information, we construct alternative likelihood functions with unknown and parametric noise variance. These likelihood functions incorporate spatial interference suppression characteristics under both the signal-absence and signal-presence hypotheses. Moreover, under the adopted alternating optimization (AO) variable partition, the weight-vector update takes a mixed logarithmic and fractional (LF) form. To efficiently solve this subproblem, we introduce an alternative scale variable (SV) transformation, which decouples the LF structure and reduces the optimization to a nonlinear univariate unconstrained subproblem with respect to the SV. Furthermore, we extend our framework to quasi-CM signals arising from hardware nonidealities. The associated fractional optimization problem with double-sided amplitude constraints is efficiently addressed. Numerical results demonstrate that the proposed solutions consistently outperform existing methods, achieving detection probabilities comparable to those obtained with perfectly known interference prior information.
AB - Constant modulus (CM) signals are widely adopted in radar and communication systems, making their reliable detection in the presence of spatial interference a fundamental problem. However, most existing multichannel signal detectors require either prior knowledge of interference directions (IDs) and powers or secondary data to compute the interference-plus-noise covariance matrix. In this paper, we devise a CM signal detector without prior knowledge of spatial interference, and formulate the problem within the generalized likelihood ratio test framework. Inspired by the ideal scenario with known interference information, we construct alternative likelihood functions with unknown and parametric noise variance. These likelihood functions incorporate spatial interference suppression characteristics under both the signal-absence and signal-presence hypotheses. Moreover, under the adopted alternating optimization (AO) variable partition, the weight-vector update takes a mixed logarithmic and fractional (LF) form. To efficiently solve this subproblem, we introduce an alternative scale variable (SV) transformation, which decouples the LF structure and reduces the optimization to a nonlinear univariate unconstrained subproblem with respect to the SV. Furthermore, we extend our framework to quasi-CM signals arising from hardware nonidealities. The associated fractional optimization problem with double-sided amplitude constraints is efficiently addressed. Numerical results demonstrate that the proposed solutions consistently outperform existing methods, achieving detection probabilities comparable to those obtained with perfectly known interference prior information.
KW - Constant modulus (CM) signal
KW - Generalized likelihood ratio test
KW - Interference prior
KW - Maximum likelihood estimation
KW - Quasi-constant modulus (QCM) signal
UR - https://www.scopus.com/pages/publications/105051209148
U2 - 10.1016/j.dsp.2026.106502
DO - 10.1016/j.dsp.2026.106502
M3 - 文章
AN - SCOPUS:105051209148
SN - 1051-2004
VL - 184
JO - Digital Signal Processing: A Review Journal
JF - Digital Signal Processing: A Review Journal
M1 - 106502
ER -