Energy-Efficient Orchestration in Wireless Powered Internet of Things Infrastructures

Multipurpose devices, capable of dynamically operating under various sensing modes, have emerged as a key element of the Internet of Things (IoT) ecosystem. In this paper, the holistic orchestration of an energy-efficient operational framework of such interconnected devices is investigated. A reinforcement learning technique is utilized enabling each IoT node, by acting as a learning automaton, to select a sensing operation mode in accordance with the IoT infrastructure's provider interests. Subsequently, a coalition formation among the nodes is realized, relying on their socio-physical relations, namely nodes' spatial proximity, energy availability, and mode correlations. The aforementioned operation is supported by a non-orthogonal multiple access wireless powered communication environment, where the nodes are able of harvesting energy from the base station. The energy efficiency of the overall system, is further improved by a utility-based optimal uplink transmission power control mechanism. The corresponding optimization problem is treated in a distributed manner as a non-cooperative game-theoretic problem, and the existence of a unique Nash equilibrium is shown, while the adoption of convex-based pricing in the utility leads to a more socially desirable Equilibrium point. The performance of the proposed approach is evaluated through modeling and simulation under several scenarios, and its superiority is demonstrated.