MCGR-PB: A Multi-Player Cooperative Game Based Routing for Performance Balancing in Crowd Sensing Networks

Abstract

Mobile crowd-sensing network is a kind of networks that take advantage of pervasive mobile devices to efficiently collect data and enable the completion of large-scale and complicated sensing tasks through the wide participation of mobile users. Data are delivered based on the opportunistic contacts among mobile uses, and the selection of relay nodes plays a vital role in the performance of such network. In most of existing routing mechanisms, data exchange is still limited between two nodes and they always pursue the maximization of single performance objective, which results in extreme imbalance of performance. Unfortunately, this issue has not yet aroused the full attention of researchers by now. In this paper, to deal with this issue, we estimate the network performance metrics such as delivery success ratio and delivery delay according to the movement model. Aiming at the multi-node meeting scene that often appears in real life, the action of selecting relay nodes is modeled as a multi-player cooperative game with the game goal of maximal performance balancing, and an MCGR-PB was proposed, which can redistribute the packets among the contacting nodes based on Nash equilibrium solution to achieve higher performance balancing. Extensive simulations based on a real-life mobility trace and a synthetic trace have been done, and the results show that compared with other protocols, MCGR-PB can greatly balance the overall performance while ensuring the desired delivery performance of the network.