多通道監視系統晶片之架構探索

由於人身安全的需求不斷提高，使得當前的數位監視系統越來越複雜，通常包含多個高畫質的攝影機、即時串流影像編解碼機機制、後端高效能串流伺服器以及連接所有裝置的監視互聯網路。因此，即時完成監視與監控，將成為現代數位監控系統設計的主要挑戰。上述挑戰可以區分成幾個方面。首先，後端串流伺服器內的記憶體子系統，必須同時滿足由多個攝影機擷取的高畫質串流影像所需的頻寬需求。其次，在複雜的多通道監視系統中，必須提高互聯網路的效能與使用率，以克服因監視目標變化導致突然大量資料傳輸所造成的頻寬浪費。因此本論文提出多個硬體排程機制克服上述問題。首先，本論文提出一個具有硬體排程的智慧型記憶體控制器(Smart Memory Controller)，可以依據監視通道實際需求，動態調整多個多媒體串流通道的記憶體存取次序。接著，提出一個中央控管的智慧型硬體模組(Smart Surveillance Hub)來管理多通道串流影像的傳輸，其可透過調整各通道的優先順序、傳輸頻寬與傳輸延遲時間，提高監視系統網路的使用率。當監視網路頻寬需求過載時，仍然能提供有效的頻寬分配以維持監視結果。此外，為了快速建模、設計與實作上述硬體模組，本論文提供一個新的硬體設計方法(Data-oriented Design Methodology)，能以一個簡單且統一的流程完成上述設計階段。Smart Memory Controller與Smart Surveillance Hub的效能分析與晶片實作結果亦將分別於本論文討論。

關鍵字

Real-time Scheduling ； Multi-Channel Surveillance

並列摘要

The continuing need for improving safety and security have increased the complexity of modern digital video surveillance systems, which often consist of multiple high-definition cameras, real-time video streaming encoding/decoding, high performance backend streaming server, and a corresponding surveillance network to connect the above devices. Therefore, the methods to accomplish the monitoring and surveillance in the real-time manner become the major challenges of designing a modern digital surveillance system. The above challenges can be partitioned into several aspects. First, the memory subsystem of the backend streaming server must meet the bandwidth requirements of multiple high-definition data streams generated by continuous video captured by multiple video cameras. Second, the performance and efficiency of the streaming interconnection network used in the surveillance system must avoid bandwidth waste during sudden transfers of huge amounts of data produced by multiple activities of the monitored targets. Accordingly, this dissertation proposes several hardware mechanisms to solve the above problems. Firstly, a novel hardware memory management mechanism, Smart Memory Controller, is proposed for dynamically adjusting memory access by multiple video channels according to their actual memory requirements. Secondly, a new hardware administrating module, Smart Surveillance Hub, is proposed for managing the stream transfer of the attached multiple video channels. Notably, the module also adjusts the priorities, occupied bandwidths, and transfer latencies of each video channel to optimize utilization of the surveillance network. It can also make the decisions needed to maintain surveillancing results during a network overload. For rapid modeling, design, and implementation of the above hardware modules, this dissertation also proposes Data-Oriented Design Methodology, a novel hardware design methodology for completing the above design stages in a simple and unified flow. The experimental results of the Smart Memory Controller and Smart Surveillance Hub are discussed. The fabrication results of the above hardware modules are also provided.

並列關鍵字

即時排程；多通道監視

參考文獻

[1] C. L. Liu and J. W. Layland, “Scheduling Algorithms for Multiprogramming in a Hard-Real-Time Environment,” Journal of the ACM, vol. 20, no. 1, pp. 46-61, 1973.

[2] D. de Niz, K. Lakshmanan, and R. Rajkumar, “On the Scheduling of Mixed-Criticality Real-Time Task Sets,” in Proc. IEEE Real-Time Systems Symposium, pp. 291-300, 2009.

[3] P. Pedreiras, P. Gai, L. Almeida, and G. C. Buttazzo, “FTT-Ethernet: a flexible real-time communication protocol that supports dynamic QoS management on Ethernet-based systems,” IEEE Transactions on Industrial Informatics, vol. 1, no. 3, pp. 162-172, 2005.

[4] L. Almeida, P. Pedreiras, and J. A. G. Fonseca, “The FTT-CAN protocol: why and how,” IEEE Transactions on Industrial Electronics, vol. 49, no. 6, pp. 1189-1201, 2002.

[7] O. Mutlu and T. Moscibroda, “Parallelism-Aware Batch Scheduling: Enhancing both Performance and Fairness of Shared DRAM Systems,” in Proc. International Symposium on Computer Architecture, pp. 63-74, 2008.

國際替代計量

多通道監視系統晶片之架構探索

未授權

主題瀏覽