Model-Driven Sensing-Node Selection for Maneuvering Target Tracking

Maneuvering target tracking will be one of the es-sential applications for future perceptive mobile networks, where multiple sensing nodes (SNs) collaboratively track the same tar-get. However, the associated SN selection is a substantial challenge due to stringent latency requirements of sensing applications. In this paper, we propose a model-driven approach by unfolding conventional optimization-based methods to tackle this problem. To this end, we first propose an iterative selection method based on the majorization-minimization (MM) framework and the alternating direction method of multipliers (ADMM). Then, we unfold the MM-ADMM approach as a deep neural network (DNN) to reduce the computational complexity and improve the performance, by leveraging an enhanced surrogate function. Simulation results demonstrate that the unfolded DNN outper-forms conventional methods with much lower computational complexity.