Video multicoder with neural network control

The simulation model of statistical multiplexer for possible use in video multicoder is considered. The scheduling control is realized by neural network, using modified Kohonen learning law. The modifications are derived in order to avoid the packet loss of the overall input streams even in case of bursty traffic. Scheduling mechanism includes the priority of input streams and the rate control mechanism in case of multicoder for video streams. Proposed model was tested over real input signals with multimedia content and outgoing signal was analyzed from the multifractal point of view applied to delays introduced by the multiplexer. Obtained results are compared with those obtained using classical RR scheduling method, and it was confirmed that NN control is better than RR.