SVRGGA APPROACH TO INVESTIGATE AND PREDICT THE DYNAMIC TRANSFORMATION AMONGST THE NODES IN THE DISTRIBUTED SYSTEMS

Over the decades of years there has been a huge demand for the networked systems. A networked system can be thought of as a collection of objects or nodes which are scattered and connected through links. The parameters that define Network are topology and the behavior / interaction of individual node with the rest of the nodes in the network. The fascinating point in understanding the behavior of the nodes is obtaining a better payoff based on the behavior or strategy of the other node. The outcome of an individual node depends on the behavior of the nodes to which it is connected. Thus the node's action depends on the behavior of the rest of the connected nodes. Say for example the popularity of the product depends on the reviews given by the customer, thus if a new customer who tries to buy the product gets connected to the existing customers, views the reviews and later behaves based on the reviews received. Objective: Thus it becomes necessary to study the dynamic behavior of the nodes to learn about the status of the nodes. Game theory is the field of mathematics which provides models of individual behavior where the outcome depends on the behavior of others. Method: In the previous works, load balancing problem is addressed using genetic or game theory approach and these approaches has given a fair results compared to the traditional approaches (static) used for load balancing. In this research the influence of the node's communication in the distributed systems with the other nodes is studied with the help of game theoretic and genetic approach by making the prediction of the cost of the nodes for a server in a distributed network using SVR models. Findings: 98 % of accuracy is obtained using SVR's RBF models for cost prediction of the nodes done in distributed systems and around 97% efficiency is realized by employing the nodes in the network with genetic and game theoretic approach. Application: Dynamic behaviour of the nodes influence the performance of the network. Thus this work can be used in any networked domain to achieve higher performance.