Energy-aware online task dispatching and scheduling for edge systems with energy harvesting

Abstract

In this paper, we consider the problem of online task dispatching and scheduling in a system of devices that may possess energy harvesting capabilities. The objective is twofold, namely to maximize the cumulative weight of tasks that can be completed before their deadlines and to minimize the total energy consumption. Our proposed solution, termed ELISE, operates in an online fashion in that for each newly arriving task it decides between three alternatives (execute before another previously scheduled task, replace an existing task, or place in the waiting line) so as to meet the objectives. We analyze the complexity of ELISE and further provide performance guarantees in terms of bounds on the gap to optimality with regards to the two objectives. Extensive simulations attest to superior aggregate weight, energy consumption, guarantee ratio, and energy consumption per task, over baseline algorithms.