Kalman Prediction Based Proportional Fair Resource Allocation for a Solar Powered Wireless Downlink

Optimization of a Wireless Sensor Network (WSN) downlink with an energy harvesting transmitter (base station) is considered. The base station (BS), which is attached to the central controller of the network, sends control information to the gateways of individual WSNs in the downlink. This paper specifically addresses the case where the BS is supplied with solar energy. Leveraging the daily periodicity inherent in solar energy harvesting, the schedule for delivery of maintenance messages from the BS to the nodes of a distributed network is optimized. Differences in channel gain from the BS to sensor nodes make it a challenge to provide service to each of them while efficiently spending the harvested energy. Based on PTF (Power-Time-Fair), a close-to-optimal solution for fair allocation of harvested energy in a wireless downlink proposed in previous work, we develop an online algorithm, PTF-On, that operates two algorithms in tandem: A prediction algorithm based on a Kalman filter that operates on solar irradiation measurements, and a modified version of PTF. PTF-On can predict the energy arrival profile throughout the day and schedule transmission to nodes to maximize total throughput in a proportionally fair way. key words: Broadcast, energy harvesting, proportional fairness, time sharing, industrial wireless sensor networks, solar energy, Kalman filter, prediction.