Synchronization of multimedia streams in distributed environments

A major feature that distinguishes multimedia applications from other traditional applications is the integration of various media streams that have to be presented in a synchronized fashion. We present a set of algorithms for achieving fine-grain synchronization in a best-effort distributed system. Our algorithms, based on numerical time stamps, take into account both the possible frame loss due to the device buffer overflow at the sender, and the display time of a video frame at the receiver. As a proof of concept, we have implemented a robust solution for lip-synchronization. Experimental results show that the skew between audio and video was between the acceptable values. We extended the solution to n continuous stream synchronization that has the same or different sources. A direct application of our research is the Interactive Remote Instruction (IRI) project, which requires synchronization algorithms in a distance learning environment where participants use best-effort operating systems and networks.