Collaborative Optimization of Sensing, Communication, and Computation Functionalities in MEC-Assisted ISAC Systems

With the unprecedented growth within the field of integrated sensing and communication (ISAC), tremendous data can be collected and transmitted in a spectrum-efficient way. However, limited by the power budget and computation capability of the ISAC device, it remains a prominent problem to expand the computation functionality in the ISAC network, while exploring the intricate interaction and achieving the balanced improvements among the performance of sensing, communication, and computation (S&C&C) pose enormous challenges. In this paper, to unravel the comprehensive interplay among S&C&C functionalities, we propose a mobile edge computing (MEC)-assisted ISAC framework that centers on enhancing the sensing performance and satisfies the communication and computation requirements. In the framework, the ISAC devices sense a set of targets and offload the sensing data to the edge server for further model training. We formulate an optimization problem aimed at maximizing the minimum sensing signal-to-interference-plus-noise ratio by jointly optimizing the sensing scheduling, beamforming of sensing and offloading signals, and computation resource allocation, subject to task processing delay constraint and model training error constraint. Then, we develop a three-layer vertical algorithm to decouple and iteratively solve the formulated mixed-integer non-convex problem. Numerical results showcase the advantages of the developed approach over benchmark algorithms and reveal the plexiform interplay among the performance of S&C&C functions, which is a notable absence in the baseline algorithm.