In the study of networked HD video surveillance, various caches embedded in the IP camera, the data management server and the network proxy server are used for lossless video storage and complete playback. And it has been considered as an emerging solution that utilizing the agent technology to achieve effective and collaborative caching in the end-to-end process of video capture, network data transmission and video data management.
However, on one hand, according to the cache's location in HD video surveillance implementation, all caches can form a top-down hierarchical structure in a vertical direction. On the other hand, according to the cache's storage performance such as free storage space, “similar” caches at the same level can be grouped by clustering in a horizontal direction. Hence, how to discover the complex interaction relationships among caches by considering both vertical and horizontal directions and further effectively make optimized video stream transmission through collaborative caching is still an open and challenging problem.
As a result, in this research, an agent-based collaborative caching model is designed, in which three roles of agents: the Capturer Agent, the Transmitter Agent and the Manager Agent are developed for discovering the intrinsic interaction relationships of caches. The agent-based collaborative caching model has two significant contributions.
The AP clustering mechanism based on caches's free storage space is utilized to cluster “similar” caches on different-levels and enables dynamic cluster updating. From both vertical and horizontal considerations, the intrinsic interaction relationships: the potential relationship and the attraction relationship among caches are deeply exploited and evaluated. Meanwhile, through utilizing the potential relationship and the attraction relationship, a collaborative caching framework for video surveillance is designed, in which the cache interaction relationship-based cache activities control and video stream migration are effectively realized.
In case study, through revealing the detailed whole cache activities control process and migration process, the effective of the model can be evaluated from a qualitative perspective. It proves that through setting proper parameters, the model can effectively decrease the transmission latency of HD video surveillance, and increase the flexibility of cache activities control and video transmission migration.
This research is supported by the National Natural Science Foundation of China (No. 61103158, 60903079, 61035003), Guangxi Key Laboratory of Trusted Software, the National S&T Major Project (No. 2010ZX03004-002-01), the Securing CyberSpace Lab of Deakin University, National Basic Research Program of China (No. 2013CB329502), the Sino-Finnish International S&T Cooperation and Exchange Program (No. 2010DFB10570), the Strategic Pilot Project of Chinese Academy of Sciences (No. XDA06010302), National Science & Technology Pillar Program of China (No. 2012BAH01B03).
The research entitled “Interaction Relationships of Caches in Agent-based HD Video Surveillance: Discovery and Utilization” has been published in Journal of Network and Computer Applications (Volume 37, January 2014, Pages 155–169).