Asymmetric Bilinear Inference for Joint Communications and Environment Sensing

Beyond-fifth-generation (B5G), sixth-generation (6G) and Internet-of-Things (IoT) applications, such as autonomous driving and robot-assisted living anticipate scenarios where multiple user equipment (UE) not only communicate with multiple access points (APs), but also collaborate among themselves and utilize communication signals in the detection and identification of objects in the environment. Motivated by this emerging paradigm, joint communications and sensing (JCAS) systems have been heavily investigated in recent years. This paper contributes to this field with a novel JCAS algorithm to extract environmental information from the scattered paths of communication signals typical of the high frequency bands envisioned for B5G and 6G systems. In particular, we model the environment as a voxelated occupancy grid and employ a bilinear Gaussian belief propagation (BiGaBP) framework to jointly estimate the voxel occupancies (filled/empty space) and the data symbols. Simulation results verify the effectiveness of the proposed algorithm in both signal and environment estimation, where the three-dimensional (3D) environment objects are perfectly reconstructed with sufficient signal-to-noise ratio (SNR) while achieving a low communications error rate.