Author :
Santos, Ricardo ; Pedreiras, P. ; Almeida, L.
Author_Institution :
Inst. of Telecommun., Univ. of Aveiro, Aveiro, Portugal
Abstract :
Video applications, which include industrial uses like machine vision, object tracking, surveillance, driving aids, etc. are becoming increasingly common. These sensors produce large amounts of data, being normally compressed at the source nodes to save network bandwidth. As a side effect, video streams exhibit a large variability in their bandwidth utilization. On the other hand, many video applications are highly dynamic. For instance, a video surveillance application can meet its goals with a low frame-rate video, when the environment being monitored is static, but require an high frame-rate when the environment changes. Another example is applications that have several video sources that are activated on demand. For instance, a rear-view video camera, nowadays commonly found in cars and trucks, is only necessary during certain maneuvers. Applications like the ones presented above can be efficiently supported by enhanced Ethernet switching that provides hierarchical server-based traffic scheduling, in particular by the FTT-enabled switch [1] [2] that resulted from instantiating the FTT (Flexible Time-Triggered) communication paradigm [3] onto a customized Ethernet switch. Servers are, in fact, adequate abstractions to handle real-time video transmission since:1) servers discipline the use of the network, thus coping with the large variability of compressed video sources, 2) the adaptability, obtained by changing the servers budget, permits varying dynamically the bandwidth allocated to each video stream, 3) the reconfigurability permits enabling and disabling video-streams. We demonstrate these features with a simplified video surveillance demo, using the setup shown in Figure 1.
Keywords :
bandwidth allocation; computer vision; local area networks; object tracking; quality of service; telecommunication switching; telecommunication traffic; video streaming; video surveillance; Ethernet switching; FTT-enabled switch; QoS; bandwidth allocation; bandwidth utilization; driving aids; flexible time-triggered communication; hierarchical server-based traffic scheduling; machine vision; object tracking; surveillance; video sensing; video streaming; Bandwidth; Cameras; Monitoring; Quality of service; Servers; Streaming media; Switches;