DESIGNING A BUFFER MANAGEMENT ALGORITHM FOR EDGE-ASSISTED IOT SYSTEM

Abstract

The data loss due to connection failure from an Internet-of-Things (IoT) device to a Cloud may occur due to disruption or many other reasons. Buffering is one of the ways to handle such data losses. The work in this thesis proposes a buffer management algorithm that is implemented in Edge-assisted architecture. The algorithm buffers the data with minimum loss in case of connection lost between Edge node and Cloud, and also manages to buffer the data of other IoT nodes in case of connection lost between IoT node and the dedicated Edge node. The algorithm efficiently deletes and reconstructs the data entries if it runs out of the storage. After the connection re-establishment with the Cloud, the buffered data is transmitted to the Cloud. The proposed system is very useful for environments like landslides early warning systems (LEWS) as there are high chances of failure of core network due to rough mountain terrains and less support. The landslide has devastating effects that can lead to multiple deaths and loss of property. The proposed system is tested and analyzed for its performance in an Edge-assisted LEWS.