Maximizing Sum Rate by Joint Sensing and Communication Scheduling for IRS-Assisted Multi-User Downlink Wireless Communications

Millimeter-wave (mmWave) technology has emerged as a promising approach to deliver ultra-high data rates for Automated Guided Vehicles (AGVs), thereby supporting critical Industrial Internet of Things (IIoT) applications such as path planning. However, its narrow directional beams are prone to blockages and difficult to align with a moving AGV, leading to beam misalignment errors and subsequent capacity degradation. By integrating sensing and communication (ISAC), mmWave can sense and extract environment information and AGV location from echo signals to optimize beamforming, thereby mitigating misalignment effects. However, frequent sensing consumes time resources, reducing data transmission time. Hence this paper proposes a multiple Intelligent Reflecting Surface (IRS)-assisted ISAC framework with a location sensing activation mechanism to balance communication performance and sensing overhead. Within this framework, multiple semi-passive segmented IRSs are deployed to simultaneously support location sensing and data transmission for multiple AGVs by allocating sub-IRS arrays independently. Moreover, we formulate a fairness-aware system rate optimization problem. We then propose an ISAC resource allocation algorithm based on heterogeneous agents proximal policy optimization (HAPPO) to jointly optimize the activation factor for location sensing and the allocation of limited wireless communication resources. Simulation results demonstrate that the proposed framework achieves centimeter-level localization accuracy, delivers superior communication performance with balanced multi-AGV fairness, and operates with low inference latency suitable for real-time industrial ISAC applications.