Load-Balanced Task Allocation for Improved System Lifetime in Mobile Crowdsensing

Abstract

Mobile CrowdSensing (MCS) applications rely on sensor data collected from a number of mobile participant devices; the participant devices need to sustain in the system for longer duration in order to services multiple requests. In this work, we propose two online load-balanced algorithms, that use available resources on mobile devices, to efficiently allocate tasks to a subset of participants. We have conducted extensive simulations to compare our algorithms with three baseline approaches and observed significant improvements in the system lifetime and the total number of tasks serviced. To further validate our results, we also conduct real-world experiments on 8 smartphones. We achieve 29.3% increase in the number of tasks serviced, with drastic improvements in system lifetime (in resource constrained cases) over the state-of-the-art approaches.