Deliverable D5.2 Control Strategies and Adaptive L1L2 Functionalities for Semantic Communications First results

This deliverable presents the initial results on control strategies and Layer 1 and Layer 2 (L1/L2) functionalities for semantic and goal-oriented communications. It introduces the basic framework and key functionalities for implementing policy-based semantic control in the Open Radio Access Network (O-RAN) architecture. First, the current O-RAN architecture is reviewed to identify the necessary enhancements to components, standardized interfaces, and operations to support semantic network control. Subsequentially, the Semantic RAN Intelligent Controllers (S-RICs) are introduced as core elements of the 6G-GOALS architecture, enabling dynamic, semantic-aware policy enforcement in the RAN. The potential benefits of Machine Learning-based network control via S-RICs are demonstrated in a commercial-grade 5G network deployment. Furthermore, the deliverable proposes semantic-aware L1/L2 functionalities and control strategies for three use cases: 1) Semantic content selection in the Internet of Things (IoT); 2) Heterogeneous IoT; and 3) Collaborative edge inference. Methods for designing and optimizing Medium Access Control protocols and device grouping strategies are developed and evaluated. These methods yield significant improvements in key performance indicators, including energy efficiency, task accuracy, throughput, communication overhead, and top-k Query Age of Information. Overall, this deliverable presents important advancements toward the realization of efficient semantic control in O-RAN, supported by novel L1/L2 functionalities and control approaches.