Fairness and social welfare in service allocation schemes for participatory sensing

Leveraging on the pervasiveness of mobile phones and their rich built-in sensors, partici-patory sensing recently emerged as a promising approach to large-scale data collection. Whilst some contributors may be altruistic, many contributors are motivated by receiving something in return for their contributions, proportional to their level of contributions. In this paper, we adopt a service allocation approach that motivates users by allocating a determined amount of compelling services to contributors, as an alternative to other credit or reputation based incentive approaches. To address two major concerns that would arise from this approach, namely fairness and social welfare, we propose two service allocation schemes called Allocation with Demand Fairness (ADF) and Iterative Tank Filling (ITF), which is an optimization-based approach. We show that: (i) ADF is max–min fair and scores close to 1 on the Jain's fairness index, and (ii) ITF maximizes social welfare and achieves the unique Nash equilibrium, which is also Pareto and globally optimal. In addition , we use stochastic programming to extend ITF to handle uncertainty in service demands that is often encountered in real-life situations. The vast penetration of mobile devices, each with a variety of built-in sensors such as GPS, accelerometer, microphone and camera, has recently spurred the emergence of a new sensing paradigm called participatory sensing [1] for large-scale data collection. It refers to an approach whereby individuals use their personal mobile devices to sense the environment and report sensed data in a non-obligated manner. Compared to conventional wireless sensor networks that use deployed sensors, participatory sensing removes the cost and hassle of sensor installation and network maintenance, while achieving much broader geographical coverage. The big challenge of prolonging network lifetime in conventional sensor networks also becomes a non-issue in participatory sensing because charging the battery on each mobile device is now taken care by the users themselves. Hence, participa-tory sensing is a promising paradigm and has attracted a lot of research attention and efforts such as [2–4]. However, the success of participatory sensing strongly relies on user participation in order to obtain a sufficient and continuous influx of user contributions. So far, most participatory sensing projects focus on developing applications , e.g., for environmental monitoring [2,3] and transportation [4], in which participants are recruited on a voluntary or remunerated basis, which is not sustainable in the long run [5].