MediaServ: Resource optimization in subscription based media crowdsourcing

Abstract

In this paper we propose resource optimization for subscription based media content crowdsourcing. In this form of crowdsourcing, interested entities (we refer to them as Campaigners) announce their `interests' expressing what media content (such as pictures, audio, and videos) they want to receive from participant users whereas mobile users subscribe to those interests as an intention to serve content satisfying the respective interests. Campaigners solicit content generated by users by mentioning explicit criteria that the media content should satisfy, for example a `noise pollution' campaigner who wants to measure noise level of a city neighborhood, may ask potential users for audio clips recorded at a certain location at peak hours of weekdays. Subscribed users voluntarily or on paid terms generate content against those interests. Given that a user may subscribe to different campaign interests and its generated content may satisfy different interests in varying degree of accuracy, we propose methods to evaluate contents based on the degree of satisfaction against the subscribed interests, and then develop techniques for delivering those contents to the campaign end points so that it optimizes the user's resource utilization, such as energy and bandwidth.