Participant Recruitment and Data Collection Framework for Opportunistic Sensing: A Comparative Analysis
Abstract
Mobile crowdsensing is a novel approach that exploits the sensing
capabilities offered by smartphones and users’ mobility to sense
large scale areas without requiring the deployment of sensors insitu. Opportunistic sensing utilizes users’ normal behavior to
crowd-source sensing missions. In this work, we propose a novel
framework for fully distributed, opportunistic sensing that coherently integrates two main components that operate in DTN
mode: i. participant recruitment and ii. data collection. We
adopt a new approach to match mobility profiles of users to the
coverage of the sensing mission. We analyze several distributed
approaches for both components through extensive trace-based
simulations, including epidemic routing, PROPHET, spray and
wait, profile-cast, and opportunistic geocast. The performances
of these protocols are compared using realistic mobility traces
from wireless LANs, various mission coverage patterns and sink
mobility profiles. Our results show how the performances of the
considered protocols vary, depending on the particular scenario,
and suggest guidelines for future development of distributed opportunistic sensing systems.