Size Efficient Key-Value Type Context Sharing in Mobile Edge Computing

Abstract

Mobile Edge Computing (MEC) aims to extend the edge of cloud computing networks in remote servers. MEC promises advantages such as fast response time and balancing processing load because MEC devices can process the information between the user devices and cloud servers intelligently and dynamically depending on the situation. An increasing number of Internet of Things (IoT) devices share information with cloud servers. MEC can address issues arising from situations where billions of IoT devices share information with other devices, MEC, and cloud servers. In this case, a load balancing model and architecture are needed for better storage and communication bandwidth control to avoid problems such as data congestion among MEC servers. Considering that the Key-value type context is one of the most popular representations for IoT information sharing, a new approach to address the communication load imbalance in MEC is needed. This paper proposes a novel model and architecture to share key-value type context among mobile devices, cloud servers, and MEC servers. We use probabilistic data structures, Bloomier Filters, to reduce the key-value type information footprint. The data quality degradation caused by false positives can be controlled and eliminated with a particular type of Bloomier filter and a common dictionary shared by devices and servers. We analyze our approach to show the proposed model’s performance, architecture, and algorithm, demonstrating that we can reduce the footprint size to represent the key-value type context information with zero or practically zero data degradation.