Research On Task Allocation Algorithms In Crowd Computing Based On Mobile Social Networks

Crowd computing is a newly emerging computation model in recent years, which involves a crowd of users, who carry various mobile smart devices (such as smart phones, iPads, vehicular devices, wearable devices, etc.), cooperatively accomplish a large complex computation task that an individual cannot easily deal with, through their intelligent collaboration. On the other hand, mobile social networks, which combine delay tolerant networks and online social networks, can support mobile users to exchange or share large-size data by using their carrying smart devices in a delay tolerant manner. Due to the technology advances, current mobile smart devices generally have good capacity of computation, sensing, communication and storage. Consequently, mobile users have actually become a huge computing resource treasure. As crowd computing can effectively utilize such a resource, it is significant to study crowd computing, having broad application prospects.Most of existing crowd computing systems are online systems. The tasks are allocated passively. Moreover, these systems have not taken into consideration the characteristics of mobile social network users, and the effect of the result recovery process on task allocation have not considered. To this end, a task allocation algorithm for crowd computing based on mobile social networks is studied in this dissertation. A user group task assignment algorithm for crowd computing is proposed. Furthermore, this algorithm is proved to be optimal. In order to demonstrate the algorithm’s performance, a crowd computing system for verifying task allocation algorithms is designed, including the user information model, the task generation model, the task assignment model, the task performing model, the result management model, and the corresponding incentive mechanisms. Mobile users can publish crowd computing tasks and recycle results through this system.The main contributions of this dissertation are presented as follows:1 A user group task assignment algorithm for crowd computing is proposed, which can assign tasks according to the mobile characteristics of users. This algorithm selects as less users as possible to effectively perform the crowd computing tasks, which can save the payment for assigning tasks to mobile users. It proves the optimality of this algorithm through the theoretical analysis, and demonstrate the significant performances through simulations.2 A crowd computing system for verifying task allocation algorithms is designed. Unlike the existing systems, this system can support mobile users to deliver large-size data in the manner of short-distance wireless communication. Moreover, a task assignment module is designed to verify the performance of the task allocation algorithms in crowd computing based on mobile social networks. Additionally, two incentive mechanisms are designed for the proposed task assignment algorithm, and the performance of this algorithm is verified by using the verification system.