On Implementing Social Community Clouds Based on Markov Models

Abstract

Social networks reflect, to a wide extent, the real-world relationships that allow users to connect and share information. The number of people that interact in social networks keeps increasing, and the devices used are equipped with more and more computational capacities. This gives rise to the formulation of social clouds, which refer to resource-sharing infrastructures that enable friends to share their resources within the social network. As modern applications become more and more sophisticated, users should be able to share their own services and computing resources through social networks. This poses many challenges for the design options of a computing system composed of a set of trusted friends. The spotlight turns on the design of a proper trust model that considers the suitability of the trusted users to execute an application’s tasks and on the fair distribution of these tasks among these users. Therefore, social networks and their trust-based applications in a distributed environment have seen increasing attention in the research community. In this regard, we present a social community cloud implementation model, where friendly relationships determine resource provisioning. The issues of fairness and allocation of time are of great importance and they are thoroughly investigated. We use extensive simulations to illustrate that the communities can be employed to construct cloud infrastructures, such that the shared resources can be utilized fairly and efficiently. Our experiments have shown that our model achieves a higher allocation rate (percentage of tasks successfully allocated and completed) than competitive models and reduces the average response time and the total execution time. Finally, our work does not overutilize the resources.