Multicommodity Flow Reliability for Energy Harvesting Wireless Sensor Networks

Abstract

In this paper, we consider energy harvesting wireless sensor networks (EH-WSN) with multiple sinks that supports multiple concurrent applications. Each application is associated with a number of sensor nodes that generate and send traffic to the application associated sink. Each node can relay any application traffic towards the application sink. In addition, each node uses an energy management unit to control the amount of traffic that the node can relay based on its available energy. To model the fluctuations of the nodes’ energy levels, we formalize a multicommodity flow reliability problem (denoted MultiFlowRel) that calls for estimating the likelihood that at least certain amount of each application traffic is delivered to the associated application sink. We show that our proposed problem is #P-hard and propose a bounding framework for deriving lower bounds for the exact reliability solutions. Numerical results show the performance of our proposed framework and its use in solving some interesting network design problems (e.g. optimal sink locations and appropriate transmission parameters).

Index Terms—Energy harvesting wireless sensor network, network reliability, energy management, multipurpose wireless sensor networks, iterative methods, probabilistic graphical mod- els

Faculty Mentor: Dr. Mohammed Elmorsy