LAUSR

Optimal resource allocation is crucial for successful deployment of energy harvesting wireless sensor networks (EH-WSNs) such as Internet of Things (IoT) devices. Nonorthogonal multiple access (NOMA) can significantly improve the network throughput compared to orthogonal multiple access (OMA). This article considers optimal power management and data scheduling in multihop EH-WSN using NOMA. The EH-WSN consists of $M$ sensor nodes aiming to transmit their data to a sink node. Assuming network connectivity, the multihop EH-WSN is represented by a directed graph. The resource allocation problem is formulated to efficiently utilize the available harvested energy to send the available data to the sink node with minimum cost. The resource allocation problem given the system dynamics is nonconvex due to the nonconvex constraints. Assuming high signal-to-interference and noise ratio (SINR), the nonconvex constraints are lower bounded by convex constraints. With the aid of variable transformation, the constrained nonconvex problem is approximated with a convex problem. The convex problem is solved using finite-horizon dynamic programming considering offline and online operations. The offline problem is formulated assuming noncausal information of the harvested energy and data arrival. The model predictive control (MPC) framework is used to obtain the solution of the online operation of the EH-WSN. A distributed MPC (DMPC) is proposed to overcome the computational complexity of solving the centralized MPC problem, assuming each sensor node is allowed to exchange information with its neighboring nodes. In the simulations, we use energy efficiency and average data transmitted to compare the performance of the EH-WSN using NOMA and OMA. Simulation results confirm that NOMA in multihop EH-WSN results in higher throughput compared to OMA.