Joint Position and Velocity Vector Sensing in mmWave/THz ISAC Systems

Abstract

This paper exploits the line-of-sight (LoS) and dominant single-bounce non line-of-sight (NLoS) propagation characteristics of millimeter wave (mmWave)/terahertz (THz) channels to facilitate joint position and velocity vector sensing in integrated sensing and communication (ISAC) systems. We develop a signal model that captures how the velocity vector is reflected in the angular velocity of the LoS path and the Doppler shifts of these paths, and characterize the fundamental accuracy limits through the position error bound (PEB) and the velocity error bound (VEB). Next, we propose a two-stage estimation framework to approach these bounds in practice. In the first stage, we propose a space-alternating generalized expectationmaximization (SAGE)-maximum likelihood (ML) algorithm to detect propagation paths and estimate the associated channel parameters. In the second stage, the estimated channel parameters are mapped to sensing parameters, and a weighted least squares (WLS) based fusion algorithm is developed to recover the velocity vector by adaptively combining the radial and tangential velocity information provided by the LoS path with the additional radial velocity information extracted from the dominant single-bounce NLoS paths. Furthermore, a weighted non-linear least squares (WNLS) refinement is proposed to adjust estimated sensing parameters. Simulation results demonstrate that the performance asymptotically approaches the corresponding bounds at medium to high signal-to-noise ratios (SNRs), and the proposed algorithms outperform the baselines in both position and velocity estimation accuracy.