Research And Implementation Of Task Scheduling Mechanism In Data Center For Intelligent Video Surveillance

In recent years, video surveillance has a wide variety of applications. The technology revolution of intelligent video surveillance systems has promoted the development of video data center. The main function of the online video analysis system in the video data center is to provide users with live surveillance video and real-time intelligent video analysis services. When there are multiple video streams, multiple users and a variety of analysis tasks, in order to make full use of system resources, we need reasonable task allocation method. Therefore, an effective task scheduling mechanism is of great importance.This thesis aims to study the task scheduling method in the surveillance system according to the characteristics of the real-time video services. First, we analyze the functional requirements of the online surveillance video analysis system, and then design the system based on B/S architecture. We also describe the design of each server in the system. Next, this thesis proposes an online video analysis task scheduling method. This method can quantify the processing ability of each video analysis server, and also quantify the resource consumption of different task. During the task execution phase, the method can update each server’s weight and task load through the dynamic feedback of each server’s resource utilization in real-time, and then choose the most appropriate one to process the current task. Finally, we implement the online surveillance video analysis system and test the scheduling method. The experimental results show that the system runs well. Our scheduling method can balance the system load and make a good decision for task allocation in real-time.