Adaptive Spatiotemporal Multi-UAV Sensing in Low-Altitude ISAC Networks

Abstract

In low-altitude integrated sensing and communication (ISAC) networks, the sensing of multiple unmanned aerial vehicles (UAVs) assisted by ground base stations remains challenging because weak-target echoes are easily masked by strongtarget echoes and non-stationary clutter. Current threshold-based sensing methods mainly rely on single-domain target screening, in which a high threshold may miss weak targets, while a low threshold may introduce false alarms by retaining clutter responses and residual strong-target responses. This tradeoff makes single-domain sensing insufficient to balance weak-target sensing probability and false-alarm control. To address this issue, this paper proposes a dual-domain adaptive multi-UAV sensing method based on spatiotemporal threshold design. Specifically, a temporal threshold first retains potential weak-target candidates from continuous echoes, and a spatial threshold further suppresses clutter responses and residual responses induced by strong echoes. The joint refinement of these two thresholds is deep-unfolded into a trainable multi-step refinement network, where a deep deterministic policy gradient (DDPG) agent adaptively regulates the threshold-update step sizes. Simulation results for multi-UAV sensing in a low-altitude ISAC network show that the proposed method improves the weak-target sensing probability while effectively controlling the false-alarm ratio.