5G-enabled Edge Computing for MapReduce-based Data Pre-processing

Abstract

The notion of edge computing, including fog computing, is to shift tasks to process data generated from sensing systems from the server-side to the network edge. Data directly measured by sensing systems tend to contain noise and loss and be in a non-canonical representation. Pre-processing for such data is often needed to reduce noise and to translate the data to a canonical representation. MapReduce processing, which was originally designed to be executed on a cluster of high-performance servers, is also useful for pre-processing data generated at the edge of a network. To support edge computing, we previously developed an approach to enable processing in embedded computers connected through wired or wireless local area networks in a peer-to-peer manner. The purpose of this paper is to extend our existing approach to give it the ability to work 5G networks. The extended approach connects nodes at the edge to base stations but not directly nodes. This paper describes the extension and its performance. The extension is based on our previous approach but has several contributions in common with other embedded computing systems for 5G networks.