Learning of Mobile-Traffic Patterns for Resource Management and Dynamic Power Controlling

Recently, the topology control solutions that use static transmission power, transmission range, and link quality, might not be useful. The objective of this paper adapts the transmission power to be adjusted with external changes by applying a machine learning algorithms. We develop a traffic signature algorithm based on traffic clusters of the network sites that have the same behavior then we predict their upcoming changes and correspondingly. The contribution of this work is using this model to create an optimal power distribution function based on traffic load. Furthermore, we propose a new algorithm that deals with all clusters and determines the optimal number of links at each hour to minimize network resources. As an illustration, results show that the proposed solution can compute coefficients of the power distribution function at each group and their needed resources.