Semantic Communication (SC) aims to reduce communication traffic by transmitting semantic embeddings rather than raw data. However, SC faces challenges such as reconstruction errors and the inability to effectively integrate cross-modal information, which can limit its accuracy in distributed sensing networks. We propose JESAC, a JEPA-Aided Self-enabled ISAC Framework, that addresses these limitations by incorporating Integrated Sensing and Communication (ISAC) using naturally available RF signals (e.g., WiFi CSI, RSSI) as complementary sensing modalities. Our framework leverages the Joint Embedding Predictive Architecture (JEPA) for cross-modal semantic sharing, allowing the system to predict missing data in a self-supervised manner.

By integrating ISAC into SC, JESAC enhances the performance fo SC significantly.

(???achievements)

This approach provides an efficient and scalable solution for distributed multimodal sensing networks.