Architectural Enhancements for Efficient Sensing Data Utilization in 6G ISAC

Current architecture proposals within standards development organizations such as ETSI and 3GPP enable sensing capabilities in mobile networks; however, they do not include a repository for storing sensing data. Such a repository can be used for AI model training and to complement ongoing sensing service provisioning by improving efficiency and accuracy. One way of realizing this is through the fusion of historical sensing data with live sensing data. In this paper, we study historical and live sensing data fusion for Integrated Sensing and Communication in future 6G systems and introduce a Sensing Data Storage Function to store historical sensing data and sensing results. We show how the Sensing Data Storage Function can be used with other network functions in a 6G architecture proposition for Integrated Sensing and Communication. We validate our proposal with a measurement model and show performance improvements in terms of detection probability and false-alarm rate. The network functionality to fuse and process sensing data combines live sensing measurements with previously sensed historical sensing data using a map-aware hard filter that rejects detections consistent with known static structures. Our simulation illustrates that, for a traffic junction scenario, map-aware hard filtering substantially reduces false alarms without degrading detection probability.