Towards Self Driven Networks
Today’s networks have scaled to a size and complexity that is beyond human ability to manage, maintain availability, and achieve optimality. Artificial Intelligence and Machine Learning are seen as the only way out to continue growing.
Indeed, SDN and NFV are bringing about programmability by decoupling control plane from the network devices and running softwarized network function on commodity hardware. The SDN-NFV controller is expected to play a key role for service provider network management by programming their customized network management policy. This controller requires computational intelligence.
During the past several years there have been remarkable advancement in academic research for applying machine learning to network management. Machine learning is applied to diversified data including traffic flow, performance, quality, syslog, configuration file, trouble tickets, SNS. In fact, AI can be practically used in many networking areas, such as fault isolation, intrusion detection, event correlation, log analysis, capacity planning, and design optimization, just to name a few.