Participation Management for Mobile Crowdsensing

The ubiquity of sensor-rich smartphones and other portable digital devices has propelled the research on large-scale mobile crowdsensing applications, where a large number of mobile users can be exploited to collect sensing data. Examples include include traffic monitoring, environmental monitoring, and many others. Fostering and maintaining user participation is crucial yet challenging for mobile crowdsensing applications with limited and opportunistic attention from mobile users. Existing crowdsourcing platforms such as Amazon Mechanical Turk [1] and mCrowd [2] require users to actively search for crowdsensing tasks. Such pull-based systems are inefficient in utilizing potential participants as it is time consuming and cumbersome for mobile users to find sensing tasks from numerous candidates, especially considering the small form factor of mobile phones. Instead of pull-based approaches, an alternative is to proactively push sensing tasks to users in the hope of notifying potential participants with tasks suggestions. However, existing push-based schemes are often in lack of a mechanism to ensure that a task is pushed to the right user at the right time. Without such mechanism, task pushing would easily degrade to spamming. Our objective is to develop a participation management system that pushes tasks to the right set of users. In particular, we focus on the following challenges: 1) how likely a user would participate? 2) how much the data contributed by this user can help the application? and 3) what is the cost of having this user participate? In other words, our goal is to determine the set of participants such that expected contributions meet the crowdsensing application requirements while the total cost is minimized.