A Synergic Model For Trust Management System For Fog Computing

Abstract

In order to successfully implement trust levels in fog networks an efficient routing protocol design is required. The routing protocol chooses nodes based on their high trust values. As a result data that is sent through these nodes is transmitted with great energy efficiency low latency high throughput and little packet loss. In order to develop such a protocol several network properties need to be analyzed and nodes need to be grouped according to the results of these analyses. In addition to this the categorization has to be complemented by dynamic routing based behaviour which enables routers to adjust the category of nodes in order to improve routing efficiency. A machine learning model for Quality of Service QoS aware routing is proposed as part of this body of work in order to facilitate the construction of a dynamic trust based routing network. This model takes into account the distance between nodes the energy of each node and the clustering of nodes in order to increase the overall routing trust levels. It is based on the Dempster Shaffer theory. When the suggested algorithm is compared to an adhoc on demand distance vector routing AODV based non trust routing algorithm a 10% increase in quality of service is shown as a result of the comparison. When measured against the previously implemented AODV based non-trust routing network this enhancement resulted in a 20% decrease in the amount of time spent waiting as well as a 15% decrease in the amount of energy used. Fog networks are subject to persistent assaults from enemies both within and outside of the organization. Researchers have developed a broad range of dynamic security models that are based on machine learning in order to identify and stop these attacks. Each of these models has its own set of functional properties that make it unique from the others. The vast majority of these models make use of a dynamic encryption technique of some kind which raises the bar for the complexity of its implementation while other simpler models are vulnerable to