Joint Fronthaul Compression and Power Allocation Optimization for Networked ISAC

• Published in: IEEE transactions on vehicular technology Vol. 74; no. 5; pp. 8279 - 8284
• Main Authors: Zhang, Kexin, Li, Lei, Chang, Tsung-Hui, Shen, Chao, He, Ruisi
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.05.2025; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Approximation; Big Data; Closed-form solutions; Communication; Constraints; Convexity; fronthaul compression; Integrated sensing and communication; Interference; Location awareness; Lower bounds; Microprocessors; Networked integrated sensing and communication; Noise; Noise level; Noise levels; Optimization; power allocation; Power management; Resource management; Simulation; Wireless sensor networks
• ISSN: 0018-9545; 1939-9359
• DOI: 10.1109/TVT.2024.3523927

This article studies the joint design of fronthaul compression and power allocation (FCPA) in a networked integrated sensing and communication (ISAC) system. In this system, a central processor (CP) is connected to multiple base stations (BSs) via capacity-limited fronthaul links to perform cooperative communication and sensing. To meet the capacity limitation, we employ a compressed forward relaying strategy and formulate a joint FCPA (J-FCPA) optimization problem to minimize the transmit power subject to the fronthaul capacity limit, the signal-to-interference-compression-noise-ratio (SICNR) requirements of communication, and the Cramér-Rao lower bound (CRLB) constraint for target sensing. Due to the non-convexity and the large number of constraints, the problem is challenging to solve. To tackle it, we first analyze the optimal compression noise levels and derive their closed-form expressions. Based on these findings, we can significantly reduce the number of constraints, achieving a simplified problem. By demonstrating that the CRLB constraint admits a difference-of-convex (DC) form, we further attain a convex approximation of the simplified one. Finally, following the idea of successive convex approximation (SCA), we develop an efficient J-FCPA algorithm by solving the approximate problem successively. Simulation results validate that our proposed design can effectively save power compared to the state-of-the-art benchmarks.