Cooperative Sensing in Short-Packet ISAC Systems Based on Superimposed Pilots

The stringent requirements for low-latency communication and high-precision sensing in the industrial Internet have spurred the development of short-packet integrated sensing and communication (SP-ISAC). However, it faces a fundamental trade-off: low-latency communication demands shorter signals, while high-precision sensing requires longer signals to accumulate energy. We propose a novel superimposed pilots based SP-ISAC system, which reduces the time-frequency resource overhead of pilots and communication latency. To improve sensing accuracy, we introduce a multi-user cooperative scheme via data fusion at base station (BS). Furthermore, we formulate a max-min rate optimization framework to ensure user fairness, solved efficiently via unsupervised deep learning. Simulation results demonstrate that our system significantly reduces sensing error while guaranteeing the communication quality of service (QoS) for worst-case users.