Node Optimization for Multi-Node Robust Integrated Sensing and Communications Networks

Abstract

This paper investigates the node optimization problem for robust localization and communication in multi-node integrated sensing and communications networks, where the communication users and the target are randomly distributed in a two-dimensional area according to Poisson distribution and uniform distribution, respectively. By defining the communication robustness as the average sum-rate over random user distributions and the localization robustness as the worst-case localization error over random target positions, we analyze the impact of node deployment on both the communication and localization robustness. By discretizing the area into uniform grids, we derive the closed-form expressions for the average sum-rate and the worst-case localization error, thereby providing theoretical metrics for the quantitative analysis of network robustness. Based on these metrics, we propose a weighted optimization framework that can trade off between the communication and localization robustness. Within this framework, we develop a hybrid genetic algorithm and multi-start pattern search-based node deployment (HGMPS-ND) scheme to jointly design node positions and array orientations, aiming to maximize the weighted robustness of both communication and localization. Furthermore, for the special case of fixed deployments, we develop a marginal-contribution-repaired genetic node selection (MCR-GNS) scheme for robust node selection. Extensive simulations validate the effectiveness of the HGMPS-ND and MCR-GNS schemes.