Neural Network-Integrated Multistatic Sensing for Joint Angle Estimation in Cell-Free JCAS Systems

Cell-free (CF) systems play a pivotal role in the evolution of next-generation wireless networks, as they improve spectral efficiency and coverage by eliminating the need for traditional cell boundaries. However, these systems encounter significant challenges, such as high computational complexity, scalability issues, and constraints on real-time decision-making. Meanwhile, the joint communication and sensing (JCAS) concept in wireless systems provides a framework that leverages communication signals not only for data transmission but also for accurate environmental sensing, thereby maximizing the utility of available resources. In this paper, we propose a neural network (NN)-based framework to estimate the joint angle-of-arrival (AoA)/angle-of-departure (AoD) resulting from available targets in a CF system after exploiting the communication waveforms generated by the access points (APs). In the simulation results, we first demonstrate that our proposed NN mechanism achieves comparable performance to the maximum likelihood estimation (MLE), and then show that the results are promising, proving the NN's ability to capture complex, non-linear relations between the angle values and channel estimations across a range of received signal qualities after testing with varying numbers of APs and targets.