Leveraging Edge Computing for Real-Time Data Security in Multi-cloud Infrastructures

Abstract

In the recent era, prosumer growth plays a significant role and makes a lot of advantages and flexibility to create interest in other activities in order to manage the energy to be consumed or produced. As in that kind of system, a lot of data is involved as it needs privacy among the users. In federated learning, the problem of dependent and identical data distribution problem occurs and those data contain user information. As that information should be more effective to make a decision-making process by ensuring privacy among the user information. In this paper, a novel federated-based data clustering technique is proposed as it improves the degree of data sampling based on multiple iterations. Using this model degree, weight of each client is measure by considering the client distribution of independent and identical problem. Based on the proposed model, communication rounds get reduced and number of iteration as compared to FedAvg algorithm. Then the data accuracy also gets improved using clustering and weighted clients based on the distribution of non-independent and identical data.