QOS of Video Transmission

This project demonstrates an approach to distributing video-on-demand (VOD) across multiple Video-on-Demand (VOD) instance servers using a client-server streaming technique based on the machine learning decision tree algorithm. One main challenge with using this technique is that a client-server approach must both efficiently utilize the strength of video data from the server(s) being requested at the same time as providing a means for asynchronous communication between the client-server video-on-demand (VOD) streaming instance servers and the video data being streamed from the server(s). In addressing this issue, we are able to provide a viable solution for robust recovery via parallel VOD video streaming instance server filling techniques, and encourage clients to maintain required levels of connectivity through alternative server score calculations. We have developed analytical models and classified server(s) according to different server type group classifications with respect to the capabilities the server(s) have available. By allowing QoS services to be utilized as a mechanism to alter traffic flows to avoid throttling capabilities of critical applications when the available network resources are limited. By using QoS technology, organizations can shift their overall network usage by providing preferential priority for specific, high throughput or high performing applications. Implementing parallel VoD streaming into parallel VOD instance servers that contain different buffer zones for providing incentive to maintain high levels of client peer connectivity will also provide insight concerning the relationship among efficiency, fairness, and incentive within the VOD client-server video-on-demand (VOD) streaming based upon the provisioning of QoS services. Through analytic analysis we demonstrate that robust Quality of Service (QoS) can be attained between clients and servers. The proposed method models the streaming process as an aggregation of end-to-end latency, bandwidth utilization, buffer occupancy, and packet delivery behavior analysis. The streaming rate is dynamically adjusted in response to current real-time network conditions to avoid both buffer starvation and excessive re-buffering.