A Long-Term Rate Scheduling Framework for Bulk Data Multicast Dissemination in Hybrid Heterogeneous Satellite-Terrestrial Networks

Abstract

We study a problem of long-term rate control for multicasting bulk data to heterogeneous receivers through hybrid satellite-terrestrial networks. By setting one of the control parameters called the {em 'knob'}, our proposed multicast rate control systems can trade off between multicast bandwidth requirement, which is the administrative issue, and reception latency, which is the users' satisfaction issue. Through a proactive use of the forward error correction (FEC) in the form of Reed-Solomon erasure (RSE) correcting codes, our proposed multicast rate control can also probabilistically guarantee reception reliability. <p>In this dissertation, there are four contributions. One, our real experiments on a hybrid satellite-terrestrial internet channel that result in our proposed Gaussian approximation approach. Two, our proposed end-to-end version of the multicast rate control system which utilizes the Gaussian approximation approach. Three, our study on the issues of multicast traffic's co-existence with terrestrial TCP background traffic. Four, our proposed two-staged version of the multicast rate control system which also utilizes the Gaussian approximation approach and allows significantly improved scalability. We validate our proposed systems through simulations, some of which are trace-driven using real satellite traffic traces. We have found our proposed mechanisms to be very effective.<p>There are a variety of applications for our proposed multicast rate control systems. Some examples include: a military application where commands or strategic information need to be disseminated at different priorities, a financial application where a faster access to certain key information can make a difference in profitability, and a news application where news-related contents such as images, weather, headlines are to be disseminated based on their urgency.<p>Keywords: satellite multicast, reliable multicast, forward error correction (FEC), multicast rate control, Gaussian approximation