Abstract: The virtual broadcast system of the present invention optimizes the routing of digital content among nodes along overlay networks that are dynamically reconfigured based upon forecasts of frequently-changing congestion levels of component interconnections within an underlying network. In the context of delivering streaming video over the Internet to large numbers of concurrent users, the present invention makes efficient use of the limited capacity of congested ASN peering points by employing deep learning techniques to forecast congestion levels across those ASN peering points and, based on those forecasts, to optimize the routing of video content along dynamically reconfigured overlay networks. The virtual broadcast system handles unscheduled as well as scheduled events, streams live as well as pre-recorded events, and streams those events in real time with minimal delay in a highly scalable fashion that maintains a consistent QoE among large numbers of concurrent viewers.

Abstract: The virtual broadcast system of the present invention optimizes the routing of digital content among nodes along overlay networks that are dynamically reconfigured based upon forecasts of frequently-changing congestion levels of component interconnections within an underlying network. In the context of delivering streaming video over the Internet to large numbers of concurrent users, the present invention makes efficient use of the limited capacity of congested ASN peering points by employing deep learning techniques to forecast congestion levels across those ASN peering points and, based on those forecasts, to optimize the routing of video content along dynamically reconfigured overlay networks. The virtual broadcast system handles unscheduled as well as scheduled events, streams live as well as pre-recorded events, and streams those events in real time with minimal delay in a highly scalable fashion that maintains a consistent QoE among large numbers of concurrent viewers.

Abstract: The virtual broadcast system of the present invention optimizes the routing of digital content among nodes along overlay networks that are dynamically reconfigured based upon forecasts of frequently-changing congestion levels of component interconnections within an underlying network. In the context of delivering streaming video over the Internet to large numbers of concurrent users, the present invention makes efficient use of the limited capacity of congested ASN peering points by employing deep learning techniques to forecast congestion levels across those ASN peering points and, based on those forecasts, to optimize the routing of video content along dynamically reconfigured overlay networks. The virtual broadcast system handles unscheduled as well as scheduled events, streams live as well as pre-recorded events, and streams those events in real time with minimal delay in a highly scalable fashion that maintains a consistent QoE among large numbers of concurrent viewers.