ILBPS: An Integrated Optimization Approach Based on Adaptive Load-Balancing and Heuristic Path Selection in SDN
Z. Ye, Gang Sun, Mohsen Guizani
Abstract
Load-balancing and path selection optimization are efficient solutions to network optimization problems. Since the emergence of software-defined network (SDN), more research in these directions appears because SDN provides a better environment for these topics. However, some current load-balancing optimization methods take a conventional approach. Moreover, we only find a few works attempting to combine load-balancing and path selection optimization to improve the network performance. To address these problems, in this article, we propose an integrated optimization approach (ILBPS) which combines adaptive load-balancing and heuristic path selection. ILBPS first uses singular spectrum analysis (SSA) to extract information from network traffic data and divides the data into payload data and noise data. Then, we use a deep learning model (GCBAC model) to predict the payload data and the convolutional neural network (CNN) model to predict the noise data. After combining the predicted results and calculating the weight of the links, we design a heuristic path selection approach based on the artificial bee colony (ABC) algorithm to compute the optimal path. We refer to this method as ABC-SP. Finally, these optimal paths are sent to the data plane. To test the effectiveness of our approach, we construct an experimental environment based on the GÉANT network and SDN. We compare our approach with the existing approaches. Experimental results show that the proposed ILBPS outperforms the existing approaches in terms of throughput, jitter, and load-balancing factor.