Delay and jitter minimization in active diffusion computing

In the active computation paradigm the problem of delay and jitter control takes a complex form from its classical counterpart. Stream processing time also becomes a new factor and the concept of path bandwidth degenerates as data volume can change while in transit. The paper presents a dynamic feedback based algorithm for this problem. We have recently implemented a concept transport system, where a video stream can receive 'transformation' in a distributed manner over an active subnet while the packets diffuse via multiple paths by a scheme called cooperative transcoding. In this paper, we present the dynamic scheduling algorithm and present its performance on live runs on the test-bed system.