ML-Assisted RIS for ISAC Systems: Initial Results in the 6G Study Band

Integrated sensing and communication (ISAC) is essential for 6G networks to alleviate spectrum congestion while simultaneously cater to rising demands for sensing and communication. While reconfigurable intelligent surfaces (RIS) improve ISAC performance, accurate channel state information (CSI) remains a challenge. This paper proposes a machine learning (ML) approach to estimate the angle of arrival (AoA) in RIS-aided systems. By training an ML model using data collected in the 6G study band, RIS is observed to be capable of predicting AoA in different scenarios. For a limited number of scanning angles and at 6G study band from 6.0-7.5 GHz, prediction errors are below 1° with 91.7% scanning reduction. Additionally, by utilizing extra sensors to measure the distance from the receiver (Rx) to the origin and its x-coordinate, estimation errors are reduced to 0.8°. The results offer practical insights for balancing AoA estimation accuracy and system complexity of the ML-intergrated RIS for ISAC applications.