Mobile Resource Aware Scheduling for Mobile Edge Environment

Abstract

In stream processing applications, a data stream is a continuous stream of data items that are generated from multiple sources distributed at various geographic locations. A common method of streaming processing is to transfer raw data streams to a data center for unified processing. However, the method does not scale well when a huge amount of data for stream processing is generated at the edge of the Internet, with the development of smartphones, Internet of things, 5G and other technologies in recent years. For stream processing applications, processing data at the edge can significantly reduce the response latency of the applications. However, the mobility of edge nodes in a mobile edge environment poses a significant challenge to scheduling stream processing tasks efficiently to achieve high system throughputs. In this paper, we introduce a scheduling algorithm, referred to as Mobile Resource Aware (MRA) stream processing scheduling, for mobile edge environment. Compared with other existing scheduling algorithms, our MRA algorithm can optimally schedule resources for stream processing tasks through adapting to the mobile edge environment with limited node resources. We implement MRA scheduling algorithm in Storm through a custom scheduler and we evaluate the performance of MRA in an emulation mobile edge environment. Our experimental results have demonstrated that our MRA algorithm can achieve significantly higher system performance than the other two existing scheduling algorithms.