Application and optimization of packet traffic monitoring depending on software defined network controller

This study has been done for packet traffic monitoring depending on software defined network controller. For this purpose, floodlight is used as software defined network controller and Eclipse is used as working platform. In order to increase the accuracy, packet statistical information was obtained after follow up on five separate topologies. These statistical data were processed on matlab and the desired data sets were created. These data sets are used in artificial neural networks by nntool on matlab to predict packet flow at any time. Artificial neural networks were tested on five separate topologies and their correctness was tested with MAPE and R2 functions. As a result of the tests performed, optimizations were applied on topology with the highest accuracy. For this purpose, a total of four optimization techniques have been tested with a combination of linear search, tabu search, modified tabu search and mixture algorithm (tabu search - simulated annealing). As a result, the most intensive route was determined as soon as possible with the optimization algorithm performed with the mixture algorithm.