面向大规模多层监控网络的资源优化方法(英文)

Adaptive deployment optimization for a multi-layer surveillance network

在建设大规模的视频监控网络时,面对监控终端随时传输过来的动态数据流问题,提出一种动态的数据量预测方法和优化设计方案.通过对系统均衡状态的理论分析,建立多层排队网络模型用以分析和预测视频监控网络的视频动态上传行为.基于此模型,对监控网络建设中的两个资源配置问题提出了优化方案.针对满足需求的最少资源配置问题,通过建模给出一种量化计算方法.对因突发事件随机暴增的上传数据流,设计了一种动态调整计算量的算法,可将超出负荷的数据流动态转移到其他可用的计算点,以保持整个系统的负载均衡,保证有效的响应时间.最后,通过一组实际环境下的实验验证了该分析和算法的有效性.

A profound design for a large distributed surveillance network system is proposed in this paper to efficient-ly quantify and predict the dynamic incoming data flows of video system surveillance. Through making a theoretical analysis on the equilibrium demand of the system, a new queuing network model is introduced to characterize the dy-namic video stream uploading behaviors from surveillance network. Two optimization problems in the system are dis-cussed, including minimal server deployment, and dynamic routing for a satisfied quality of service（ QoS） . Further-more, the new design and the proposed optimization algorithms are evaluated in a realistic environment. It is able to dynamically redirect the bursty requests flow to the sufficient nodes to avoid the overflow of the queueing and to keep the system response in time.