Research And System Implementation On Video Transmission Load Balancing Algorithm For Safe City

Large-scale security systems such as Safe City,Skynet Project and Xue Liang Project require that video monitoring should be interconnected in scale,shared widely and used intelligently.Since large-scale video transmission generates pressure to the streaming media server cluster,allocating user requests in balance to the streaming media server cluster is the core and key point for improving the performance of the streaming media server cluster and ensuring the real-time performance of video transmission and the quality of the videos.This paper designs load balancing algorithm for different scenarios in Safe City.The specific work is as follows:For the unbalanced load of the streaming media server cluster in emergency command,this paper,by employing global search of genetic algorithm,puts forward a load balancing algorithm based on genetic algorithm.A fitness function with request priority level is designed to guarantee the quick response to video requests when an alarm event or abnormal behavior event occurs.By determining the chromosome coding,the design of the fitness function and genetic operations,an optimal allocation plan for requests is obtained in order to maximize the resource utilization,to improve the performance and to balance load of the streaming media server cluster.Compared with the traditional dynamic feedback load balancing algorithm,the load balancing algorithm based on genetic algorithm reduces by 15.9% in the average responding time,by 37.1% in the load standard deviation of streaming servers.For the unbalanced load of the streaming media server cluster in the request of the same streaming by large scale users,this paper,based on BP neural net,proposes a method: continuous prediction for the video request which achieves the accurate prediction to the numbers of the video requests.Based on this,a load balancing algorithm based on the predicting numbers is proposed with the help of trend in the load change getting form predicting numbers,which improves the load lag in the traditional dynamic feedback load balancing algorithm the load balance of the clusters of the streaming media.Compared with the traditional dynamic feedback load balance,the load balancing algorithm based on the predicting numbers of the requests reduces by 10.2% in terms of the average responding time,by 51.5% in the load standard deviation of streaming servers.Based on the studied techniques and methods,we have succeeded in a monitoring system for the safe city.The two algorithms of load balancing proposed in this paper are used in the system,and tested to achieve load balancing of streaming media server clusters during video transmission in different scenarios.It ensures the real-time video transmission and video quality,and verifies the effectiveness of the method.