Incentive Mechanism for Mobile Crowd Sensing Using Reverse Auction Dynamic Pricing and Recent History

Abstract

Mobile crowd sensing is a technique that allows collection of real-time data from a large number of mobile users who carry a mobile device with sensing capabilities. It is widely used for data sensing applications, such as traffic monitoring, environmental monitoring, health and fitness, retail marketing, and emergency response. It requires individual users to perform the sensing task based on the location of the task and the user. Ensuring privacy and security of individuals and accuracy and reliability of the data collected are primary challenges in a mobile crowd sensing system. To motivate more users to collect data, it is required for the system to be built in a manner that each user is rewarded for the task done while maintaining the budget balance. As users are of heterogeneous nature, they must be rewarded for the task done based on their own true valuation of the task. The reverse auction method for mobile crowdsensing is becoming one of the widely used incentive mechanism for its choice to the mobile users, who act as the participating workers, for fixing the price for which they want to sell the sensed data. For a reverse auction system to work, it is required that there are enough users who are willing to bid in an auction round. Maintaining a participant pool with enough competition while keeping the bid values near to true values is a key challenge to be addressed. Failing to maintain enough participants can result in higher bid prices with each round and hence increasing total reward value to be distributed. This may lead to incentive explosion where the bid price is too high for the available budget. In this work, we propose a novel approach of retaining users by considering the frequency of winning and participation of users. This mechanism is built on top of RADP-VPC which is reverse auction mechanism based on reverse- auction with dynamic pricing with virtual participation credit. The experimental results show that the proposed approach using participation history for each user performs better than RADP-VPC in terms of retaining users and incentive explosion.