We are pleased to announce that our paper, “Trustworthy AI-Driven Dynamic Hybrid RIS: Joint Optimization and Reward Poisoning-Resilient Control in Cognitive MISO Networks,” has been published in IEEE Transactions on Network and Service Management (Early Access). [: arxiv ]

In this work, the authors — Dr. Deemah H. Tashman, and Prof. Soumaya Cherkaoui — propose an adaptive, energy-aware hybrid reconfigurable intelligent surface (RIS) framework for underlay multiple-input single-output (MISO) cognitive radio networks. The proposed approach dynamically alternates between passive and active RIS operation modes based on harvested energy availability, enabling efficient and practical deployment in energy-constrained environments.

To jointly optimize transmit beamforming and RIS phase configuration, we leverage soft actor-critic (SAC) deep reinforcement learning. Importantly, this work presents the first systematic study of reward poisoning attacks on DRL agents in RIS-enhanced cognitive radio networks and introduces a lightweight, real-time defense mechanism to ensure robust and trustworthy operation under adversarial conditions.

Our results demonstrate improved throughput–energy trade-offs compared to fully passive and fully active RIS architectures, while maintaining resilience against malicious reward manipulation.