ARCHITECTURE AND FRAMEWORK FOR RELIABILTY OF EDGE-ASSISTED IOT SYSTEM

Abstract

Internet of Things (IoT) is the interconnection of Internet-enabled devices that can share information with each other or to Cloud servers. The communication between the delivering nodes and the Cloud server is vulnerable to link failures. In real time scenarios, this connection loss might hinder the decision making at the Cloud due to non-reception of needed data at the right time. Eventually, affecting the reliability of the IoT systems. Edge servers, placed near to the IoT node on the edge of the core network, will have limited computing and storage resources that can address the reliability of the systems. They will collect data from the IoT node and forward data to the Cloud. In this way even if the connection to the Cloud is lost, the Edge server would still be able to process and analyze sensed data to produce meaningful results similar to Cloud. Yet another challenge is the failure of links between IoT nodes and the Edge servers. This research work is aimed at making the data processing and alarm generation as reliable as possible so that even if the connection is lost between the IoT node and Edge server or Cloud server, the data could still be processed and results are obtained. The work in this thesis is expected to produce a reliable data processing system which can be used for real-time response. The proposed mechanism has been implemented on the laboratory setup to prove that, though there might occur transmission and processing delays during switching of the system, after all, is reliable. The proposed model is tested for Landslide Early Warning System (LEWS) where connection failure between the IoT node and Cloud can be catastrophic and cause danger to human lives.