Cooperative ISAC Networks: Performance Analysis, Scaling Laws, and Optimization

Integrated sensing and communication (ISAC) networks are investigated with the objective of effectively balancing the sensing and communication (S&C) performance at the network level. Through the simultaneous utilization of multi-point (CoMP) coordinated joint transmission and distributed multiple-input multiple-output (MIMO) radar techniques, we propose an innovative networked ISAC scheme, where multiple transceivers are employed for collaboratively enhancing the S&C services. Then, stochastic geometry is exploited for characterizing the S&C performance, which allows us to illuminate the key cooperative dependencies in the ISAC network and optimize salient network-level parameters. Remarkably, the derived Cramér-Rao lower bound (CRLB) expression of the localization accuracy unveils a significant finding: Deploying N ISAC transceivers yields an enhanced average cooperative sensing performance across the entire network, in accordance with the $\ln ^{2}N$ scaling law. Crucially, this scaling law is less pronounced in comparison to the performance enhancement of $N^{2}$ achieved when the transceivers are equidistant from the target, which is primarily due to the substantial path loss from the distant base stations (BSs) and leads to reduced contributions to sensing performance gain. Moreover, we derive a tight expression of the communication rate, and present a low-complexity algorithm to determine the optimal cooperative cluster size. Based on our expression derived for the S&C performance, we formulate the optimization problem of maximizing the network performance in terms of two joint S&C metrics. To this end, we jointly optimize the cooperative BS cluster sizes and the transmit power to strike a flexible tradeoff between the S&C performance. Simulation results demonstrate that compared to the conventional time-sharing scheme or a non-cooperative scheme, the proposed cooperative ISAC scheme can effectively improve the average data rate and reduce the CRLB, hence striking an improved S&C performance tradeoff at the network level.