Optimal Joint Fronthaul Compression and Beamforming Design for Networked ISAC Systems

Kexin Zhang, Yanqing Xu, Ruisi He, Chao Shen, Tsung‐Hui Chang

Abstract: This study investigates a networked integrated sensing and communication (ISAC) system, where multiple base stations (BSs), connected to a central processor (CP) via capacity-limited fronthaul links, cooperatively serve communication users while simultaneously sensing a target. The primary objective is to minimize the total transmit power while meeting the signal-to-interference-plus-noise ratio (SINR) requirements for communication and sensing under fronthaul capacity constraints. We demonstrate that the optimal fronthaul compression variables can be determined in closed form alongside the beamformers. Leveraging this insight, we show that the remaining beamforming design problem can be solved globally using the semidefinite relaxation (SDR) technique. Furthermore, we exploit a novel Uplink-Downlink (UL-DL) duality within the ISAC framework to develop an efficient primal-dual (PD)-based algorithm. Numerical results confirm the global optimality and efficiency of the proposed PD-based algorithm, achieving significant computational speedups.

The Challenge of Networked ISAC

Integrated Sensing and Communication (ISAC) is a flagship technology for 6G, allowing the same radio waves to transmit data to users and "see" objects like a radar. Networked ISAC takes this further by coordinating multiple base stations (BSs) through a Central Processor (CP) to improve coverage and accuracy.

However, a critical bottleneck exists: the fronthaul links (the data pipes connecting the CP to the BSs) have limited capacity. You cannot simply send raw, uncompressed signals across these links without overwhelming them. To make this work, signals must be compressed, which introduces "quantization noise." This creates a "chicken and egg" problem: the optimal radio beamforming depends on how much compression noise is present, but the best compression strategy depends on the beamforming.

Networked ISAC Architecture

System model of a networked ISAC system with capacity-limited fronthaul links connecting the Central Processor to transmitters and receivers.

As illustrated in the system model above, the network consists of multiple transmitters (TXs) and a receiver (RX) coordinated by a CP. The CP manages the data flow for $K$ communication users and a sensing target (e.g., a vehicle). The challenge lies in the Joint Fronthaul Compression and Beamforming Design (J-FCBD), where the system must minimize total transmit power while satisfying strict SINR constraints for both communication and sensing tasks.

Methods and Mathematical Insights

The researchers formulated a mathematical optimization problem to minimize transmit power subject to performance requirements and fronthaul capacity limits. Their key technical contributions include:

  1. Closed-Form Solutions: They proved that the optimal compression variables can be determined analytically once the beamformers are known. This simplifies the problem from a massive search to a more targeted one.
  2. Semidefinite Relaxation (SDR) Tightness: They demonstrated that a standard relaxation technique (SDR) yields the globally optimal solution (zero duality gap), proving the problem has "hidden convexity."
  3. Low-Complexity PD Algorithm: Because SDR is computationally expensive for large networks, they developed a Primal-Dual (PD)-based algorithm. This exploits a novel extension of Uplink-Downlink (UL-DL) duality to the ISAC framework.

The Primal-Dual Algorithm

The proposed PD-based algorithm (Algorithm 1) which utilizes fixed-point iterations and bisection to find the global optimum efficiently.

The PD-based algorithm alternates between solving beamforming with a fixed dual variable via fixed-point iteration and updating the dual variable via bisection. This approach ensures global optimality while remaining computationally inexpensive compared to standard convex optimization toolboxes.

Key Results and Performance

  • Global Optimality: The proposed PD-based algorithm achieves the exact same power-saving performance as the heavy-duty SDR method, confirming it doesn't sacrifice quality for speed.
  • Massive Speedup: For systems with 40 or more antennas per base station, the PD algorithm reduced computation time by more than 80% compared to the SDR-based approach.
  • Superiority over Baselines: The joint design significantly outperformed "suboptimal" methods that design sensing and communication separately, proving that treating them as a single problem is essential for power efficiency.
  • Fronthaul Sensitivity: The results quantified the "bottleneck effect"—increasing fronthaul capacity from 20 to 60 Mbps drastically reduces required transmit power, but gains diminish after 60 Mbps as the system becomes limited by radio interference rather than data links.

Conclusion

This research provides a robust framework for managing the trade-offs in networked ISAC systems. By discovering the hidden mathematical structure of the J-FCBD problem and leveraging UL-DL duality, the authors have created a practical, high-efficiency solution for 6G network management. Future work will look to extend these models to multi-target scenarios and account for imperfect channel state information (CSI).

Summary updated

Abstract

This study investigates a networked integrated sensing and communication (ISAC) system, where multiple base stations (BSs), connected to a central processor (CP) via capacity-limited fronthaul links, cooperatively serve communication users while simultaneously sensing a target. The primary objective is to minimize the total transmit power while meeting the signal-to-interference-plus-noise ratio (SINR) requirements for communication and sensing under fronthaul capacity constraints, resulting in a joint fronthaul compression and beamforming design (J-FCBD) problem. We demonstrate that the optimal fronthaul compression variables can be determined in closed form alongside the beamformers, a novel finding in this field. Leveraging this insight, we show that the remaining beamforming design problem can be solved globally using the semidefinite relaxation (SDR) technique, albeit with considerable complexity. Furthermore, the tightness of its SDR reveals zero duality gap between the considered problem and its Lagrangian dual. Building on this duality result, we exploit the novel UL-DL duality within the ISAC framework to develop an efficient primal-dual (PD)-based algorithm. The algorithm alternates between solving beamforming with a fixed dual variable via fixed-point iteration and updating dual variable via bisection, ensuring global optimality and achieving high efficiency due to the computationally inexpensive iterations. Numerical results confirm the global optimality, effectiveness, and efficiency of the proposed PD-based algorithm.
BibTeX
@article{zhang2025optimal,
  author = {Kexin Zhang and Yanqing Xu and Ruisi He and Chao Shen and Tsung‐Hui Chang},
  title = {{Optimal Joint Fronthaul Compression and Beamforming Design for Networked ISAC Systems}},
  journal = {IEEE Trans. Signal Process.},
  year = {2025},
  doi = {10.1109/tsp.2025.3617253},
}
Figure