Modelling wireless sensor networks with energy harvesting: A stochastic calculus approach

Wireless Sensor Networks (WSN) have been increasingly applied to industrial monitoring and control, where soft Quality of Service guarantees under worst-case scenarios are often required. Many performance analysis frameworks including Network Calculus are proposed to derive those guarantees. Those frameworks were originally designed for wired network and typically do not consider energy constraints, which play a crucial role in the operation of wireless sensor networks. In this paper, a novel model is proposed to integrate energy harvesting into the network calculus framework to stochastically bound the worst-case performance. In our framework, energy is converted into tokens to stochastically bound the service provided to the traffic. Meanwhile, behaviors of nodes during the situations of energy overflow and underflow are accurately characterized by two virtual queues. Various performance metrics are analyzed and bounded for a single node and for a tree-based data collection wireless sensor network. Additionally, procedures are presented to show the model instantiation for a specific wireless sensor network based on hardware parameters and empirically obtained energy harvesting traces.