Sensing–Communication–Computing Integrated Resource Allocation for AI-Empowered Trustworthy IoT

Abstract

Advancements in 5G, cognitive Internet of Things (IoT), and Artificial Intelligence (AI) technologies have revolutionized consumer electronics in smart grids. AI-empowered IoT enables efficient data processing and decision support, providing significant benefits alongside security challenges. While blockchain ensures trustworthy data sharing, transaction management, and attack protection, consumer electronics still face threats from resource allocation fragmentation, device access adversary, and paradox between data processing and blockchain consensus. Therefore, a joint optimization problem for sensing-communication-computing multi-dimensional resource allocation is constructed. By jointly optimizing the time scheduling ratio, channel allocation, power control, and computing resource allocation, it minimizes the combined delay of device-to-device transmission, data computing queuing, and blockchain authentication. Then, the problem is decoupled and solved in two stages. In the first stage, the adversarial Deep Q-network (DQN) based joint optimization algorithm of time scheduling, channel allocation, and power control method is proposed. It achieves adversary awareness by augmenting DQN with edge-end collaborative Signal-to-Interference plus Noise Ratio (SINR) ranking. In the second stage, a delay and security tradeoff computing resource allocation method is proposed to jointly guarantee low-latency data processing and high-security blockchain consensus. Simulation results demonstrate that our algorithm effectively reduces data processing delay and enhances blockchain authentication security.