Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Coordinated Decentralized Resource Optimization for Cell-Free ISAC Systems

Mehdi Zafari, Rang Liu, A. Lee Swindlehurst

TL;DR

The paper tackles the scalability bottlenecks of centralized ISAC optimization by proposing two coordinated decentralized strategies for beamforming and power allocation in cell-free networks. SplitOpt partitions computation by retaining local beamforming at APs while centralizing power allocation, whereas JointOpt uses a fully decentralized consensus ADMM to jointly optimize BF and PA. Key contributions include LM-RZF and NS-C beamforming designs, an epigraph-based ADMM formulation with inner-approximations, and an open-source Python simulator to evaluate fronthaul overhead and performance. The results demonstrate effective sensing and communication performance with substantially reduced fronthaul communication, enabling scalable deployment of JC&S in distributed ISAC systems.

Abstract

Integrated Sensing and Communication (ISAC) is emerging as a key enabler for 6G wireless networks, allowing the joint use of spectrum and infrastructure for both communication and sensing. While prior ISAC solutions have addressed resource optimization, including power allocation, beamforming, and waveform design, they often rely on centralized architectures with full network knowledge, limiting their scalability in distributed systems. In this paper, we propose two coordinated decentralized optimization algorithms for beamforming and power allocation tailored to cell-free ISAC networks. The first algorithm employs locally designed fixed beamformers at access points (APs), combined with a centralized power allocation scheme computed at a central server (CS). The second algorithm jointly optimizes beamforming and power control through a fully decentralized consensus ADMM framework. Both approaches rely on local information at APs and limited coordination with the CS. Simulation results obtained using our proposed Python-based simulation framework evaluate their fronthaul overhead and system-level performance, demonstrating their practicality for scalable ISAC deployment in decentralized, cell-free architectures.

Coordinated Decentralized Resource Optimization for Cell-Free ISAC Systems

TL;DR

The paper tackles the scalability bottlenecks of centralized ISAC optimization by proposing two coordinated decentralized strategies for beamforming and power allocation in cell-free networks. SplitOpt partitions computation by retaining local beamforming at APs while centralizing power allocation, whereas JointOpt uses a fully decentralized consensus ADMM to jointly optimize BF and PA. Key contributions include LM-RZF and NS-C beamforming designs, an epigraph-based ADMM formulation with inner-approximations, and an open-source Python simulator to evaluate fronthaul overhead and performance. The results demonstrate effective sensing and communication performance with substantially reduced fronthaul communication, enabling scalable deployment of JC&S in distributed ISAC systems.

Abstract

Integrated Sensing and Communication (ISAC) is emerging as a key enabler for 6G wireless networks, allowing the joint use of spectrum and infrastructure for both communication and sensing. While prior ISAC solutions have addressed resource optimization, including power allocation, beamforming, and waveform design, they often rely on centralized architectures with full network knowledge, limiting their scalability in distributed systems. In this paper, we propose two coordinated decentralized optimization algorithms for beamforming and power allocation tailored to cell-free ISAC networks. The first algorithm employs locally designed fixed beamformers at access points (APs), combined with a centralized power allocation scheme computed at a central server (CS). The second algorithm jointly optimizes beamforming and power control through a fully decentralized consensus ADMM framework. Both approaches rely on local information at APs and limited coordination with the CS. Simulation results obtained using our proposed Python-based simulation framework evaluate their fronthaul overhead and system-level performance, demonstrating their practicality for scalable ISAC deployment in decentralized, cell-free architectures.

Paper Structure

This paper contains 11 sections, 24 equations, 2 figures, 1 table, 2 algorithms.

Table of Contents

  1. Introduction
  2. ISAC System Model
  3. Communication Model
  4. Sensing Model
  5. Problem Formulation and Distributed Algorithms
  6. SplitOpt (Distributed BF + Centralized PA)
  7. Distributed BF at APs
  8. Centralized PA at CS
  9. JointOpt (Decentralized ADMM)
  10. Simulation Results
  11. Conclusion

Figures (2)

  • Figure 1: Performance of Algorithm \ref{['alg:splitOpt']}. (Left) SINR and sensing SNR comparison for fixed PSRs (0.2, 0.8) and optimal PSRs (Opt). (Right) SINR and sensing SNR achieved under different SINR constraints $\gamma$ with $N_{ue} \in \{4, 5, 6\}$.
  • Figure 2: Performance of Algorithm \ref{['alg:jointOpt_admm']}. (Left) SINR and sensing SNR vs different number of users for trade-off parameter $\lambda\in\{0.3,0.6,0.9\}$. (Right) SINR and sensing SNR with $N_{ap} = 11$ APs and $M=3$ antennas for both the proposed ADMM-based decentralized solution and the centralized solution.