Bayesian classification based service-awareness in software defined optical network for big data services

With emerging of multiple services brought by Big Data, the rapid evolution of Software Defined Optical Network (SDON) provides great support for big data services with high dynamic. However, newly emerging big data services have diversified characteristics and demands, which presents great problems for current SDON to match these services with high quality and matching degree. This paper proposes a Bayesian Classification based Service Awareness (BC-SA) mechanism of SDON for big data services. By using Bayesian classification, the BC-SA mechanism is able to be aware of the type of service and to cooperate with the OpenFlow protocol, with the aim to achieve high matching degree between services and SDON. Based on this BC-SA, bandwidth scheduling can be improved in SDON. Simulation results show that the BC-SA can match requirements by big data services with better performances.