Fuzzy Model of Dynamic Traffic Management in Software-Defined Mobile Networks

Nowadays, in mobile networks, latency-sensitive services may compete for the bandwidth of other services, thus degrading overall performance. Software-defined mobile networks opening many new possibilities for dynamic traffic management. This gives chances to ensure strict requirements of the service quality in changing conditions. We considered two requirements for the quality of service: bandwidth and latency. Providing the required bandwidth is relatively simple compared to the end-to-end delay, as its guarantee requires a complex model that takes into account the mutual influence of flows throughout the entire path. Our model includes the following metrics: maximum channel utilization, traffic priority, the number of "hops". Fuzzy balancer module has been developed for the Floodlight controller in Java. This module calculates the weights of the links in accordance with the proposed method. Simulation network was held in Mininet environment. During the experimental implementation, it has been shown that simple algorithm based on mathematical apparatus of fuzzy logic allows dynamically adapting the network to the change of the traffic volume, as well as its structure.