A Mobile Crowd Sensing Ecosystem based on Fog Computing Infrastructure

Mobile crowd sensing (MCS) applications are rapidly increasing and becoming a vital technique affecting the daily lives of people. However, large scale mobile sensing systems and the rapid growth of diverse applications have generated a huge volume of sensing data that need to be efficiently and effectively processed and managed. This poses the need of a new assistive platform powerful enough to boost the development of MCS. Researchers and practioners have considered to adopt cloud computing as an approach to promote such development. With the assistance of the cloud, massive data can be processed more effectively. However, centralized cloud datacenters still have some issues that need to be addressed: (1) High latency: since most of the sensed data need to be transmitted to the remote cloud, it introduced high latency to MCS applications. (2) Heavy load: the centralized cloud datacenter undertakes most of the load in the MCS system that may lead to hotspot. (3) Limited scalability: the centralized cloud datacenter may not be easily extended as the scale of MCS increases. To overcome these issues, in this paper, we propose a novel Mobile Crowd Sensing framework based on Fog computing (MCSF). Instead of using a centralized cloud, MCSF adopts fog computing infrastructure with decentralized micro datacenters, which can be used for large-scale deployment with better scalability. In MCSF, each micro datacenter of the fog just needs to handle the portion of the load within its service area. This reduces the oversubscription probability that occurs with the system of using a centralized cloud datacenter. Since the distributed micro datacenters are close to mobile nodes, fog computing is more effective than cloud computing for assisting MCS applications, as it reduces latency. MCSF enables the data to be preprocessed in micro datacenters and mobile nodes, which can reduce the amount of transmitted data and thereby reducing the transmission latency.