Adaptive Request Routing Based on Current Load Analysis

Article deposited for open archiving and discovery (librarian deposit). Original publication: Kliukovkin, G., & Mikhaylov, B. (2020). Adaptive Request Routing Based on Current Load Analysis. Current Research: International Scientific Journal, No. 22(25), 2020. Publisher: Agency for Advanced Scientific Research (APNI), Belgorod.Abstract: The paper presents the theoretical foundations of adaptive routing, including the use of queuing models, the backpressure approach, and AntNet algorithms. Modern hybrid routing schemes in software-defined networks (SDNs) are considered, taking into account load and quality of service (QoS) metrics. Based on the collected data, weight models of routes are formed, and routing rules are updated via the OpenFlow controller. Special attention is paid to development prospects, including integration with machine learning algorithms, predictive models, management decentralization, and cognitive routing. The conducted research confirms the high efficiency of adaptive methods in comparison with classical approaches and justifies the need for their further implementation in network infrastructures with a high degree of variability and fault tolerance.