Robust Beamforming for CoMP ISAC Networks With GLRT-Based Target Detection

Abstract

Coordinated multi-point (CoMP) integrated sensing and communications (ISAC) architecture enhances cell-edge coverage, yet its sensing performance is severely limited when target statistics (e.g., radar cross-section and noise variance) are unknown a priori. Treating the unknown parameters as deterministic, we propose a generalized likelihood ratio test (GLRT) based invariant detector that maintains constant false alarm rate (CFAR). Leveraging the monotonic relationship between the detection probability and the non-centrality parameter to model the intractable detection probability constraint, we formulate a joint transmit beamforming problem that maximizes the communication sum-rate under radar detection and minimum user signal-to-interference-plus-noise ratio (SINR) constraints, robustly accommodating both perfect and imperfect CSI. To solve this highly non-convex optimization problem, a two-stage algorithm is developed: Stage I extracts a feasible initial point, while Stage II employs fractional programming for iterative sum-rate maximization. Numerical results show that the proposed joint design significantly enlarges the feasible region and achieves substantial gains over non-CoMP baselines.