Improving IPC Video Lag in CCTV Systems by Machine Learning and Software Defined Networks
Wei Chen, Jiachen Luo, Yihuang Luo, Pei Ye · 2023
Many small and medium-sized hotels around the world deploy closed-circuit television (CCTV) networks. In a CCTV network system, IP cameras (IPCs) and other network devices are available, such as Wi-Fi and routers. Usually, when a CCTV network is initially deployed, IPC video surveillance screen jams and distorted images still occur even if network technicians reserve enough uplink bandwidth for the IPCs. In this paper, based on machine learning library by pytorch, we learn and identify the features of IPC traffic data through the deep belief network (DBN) and logistic regression (LR) neural network models to accurately identify IPC traffic. Then, we combine it with hash polarization optimization scheme and use software defined network (SDN) controller to distribute the openflow flow table to handle load balancing of IPC network traffic to solve the problem. At the same time, the quality of service (QoS) policy is applied to the IPC traffic by cooperating with automatic configuration, which can also solve the problem of IPC video surveillance screen lag and screen distortion in the case of network congestion occurring in the surveillance system.