Семинар по проблемам Visual Cloud Computing (VCC)
Семинар по проблемам VisualCloudComputing (VCC)
22.12.2016г., 17:00, ауд. 103, Биржевая 14
Тема: Resilient Virtual Path Management for Scalable Data-intensive Computing at Network-Edges
Докладчик: ДмитрийЧемоданов. А senior PhD student in the Department of Computer Science at University of Missouri-Columbia, USA
Data-intensive visual computing requires seamless processing of imagery/video at the networkedge and core cloud platforms with resilient performance that caters to user Quality of Experience (QoE) expectations. Seamless processing is particularly critical in (man-made or natural) disaster incident scenarios, where the locations generating imagery or video content are rarely equipped with high-performance computation capabilities to run computer vision algorithms. Existing network virtualization literature lacks pertinent knowledge on how to satisfy the needs of incident-supporting Visual Cloud Computing (VCC) applications, e.g., on virtual path management algorithms and application-aware network design. In this project, we address such knowledge gaps by investigation of new approaches to bootstrap and manage virtual networks hosting VCC applications and services. We propose novel virtual path management solutions - path creation and maintenance - for VCC, allowing data movement from the network-edge with civilian smartphones and surveillance cameras to pertinent computing sites e.g., fog at a network-edge, or a public cloud. Our approaches are motivated by emerging paradigms such as software-defined networking and hybrid cloud computing that are taking shape, which allow decoupling of the network/system elements and their functions. By extending the decoupling concepts to the data-intensive computing level that involve imagery/video-processing functions, we seek to explore new virtual path management paradigms to enable effective VCC of data generated by applications during disaster incidents.