Medium Access for Multi-Cell ISAC through Scheduling of Radar and Communication Tasks

TL;DR

This work tackles the problem of coordinating communication and radar tasks in multi-cell ISAC networks under QoS requirements. It introduces a frame-based, time-multiplexed framework that jointly optimizes radar scan patterns and data throughput using an interference-aware beamforming/assignment strategy, demonstrated to meet target detection and tracking with high reliability while reducing resource usage. The main contributions are (i) a three-step scheduling algorithm prioritizing tracking, then communication, then search; (ii) an assignment-problem formulation for tracking and search beamformer selection, solvable by the Hungarian algorithm; and (iii) a numerical demonstration showing superior performance over baselines in terms of detection reliability, tracking efficiency, and overall resource utilization. The approach offers a practical pathway to scalable, coordinated ISAC deployments in dense multi-cell networks, with clear implications for efficient spectrum sharing and coordinated sensing and communication operations.

Abstract

This paper focuses on communication, radar search, and tracking task scheduling in multi-cell integrated sensing and communication (ISAC) networks under quality of service (QoS) constraints. We propose a medium access control framework multiplexing the tasks while optimizing radar scan patterns through an interference-aware assignment formulation. Simulations show that our solution guarantees target QoS with improved resource efficiency over baseline schemes, highlighting the benefits of coordinated scheduling in multi-cell ISAC.

Figures (3)

• Figure 1: Multi-cell sensing and communication network and proposed frame for task coexistence. During the tracking and search subframes, transmit radar pulses with the beamforming vectors $\bm{\psi}_j\in\Psi_i$. During the communication subframe, employ combining to receive ' data.
• Figure 2: CDF of the detection probability, $P_{ik}^\text{d}(\bm{\phi})$, given by the search scan pattern and radar , $\Gamma_{ik}(\bm{\phi)}$, given by the tracking scan pattern. The proposed search pattern when $N_\text{l}=12$ requires $D_\text{s}=13$, so each needs to be turned off for one dwell interval. For all the other cases, $D_\text{s}=N_\text{l}$.
• Figure 3: (a) Number of tracking dwell intervals required by the scan patterns vs. the number of tracked targets per cell. (b) Tracking subframe duration vs. the tracking update rate and number of tracked targets per cell. (c) Average communication throughput vs. the search rate and number of tracked targets per cell; the proposed patterns for search and tracking are adopted. All results are calculated over $10^4$ realizations.

Theorems & Definitions (1)

• Definition 1: Virtual Scatterers