An analytical framework for the design of intelligent algorithms for adaptive-rate MPEG video encoding in next generation time-varying wireless networks

Adaptive rate video encoding is required to maximizeefficiency when wireless links are involved in the communication.In fact, wireless channels are characterized by high, time-varyingbit error rates. To cope efficiently with this problem adaptiveforward error correction schemes have been proposed. Theseschemes introduce an amount of redundancy dependent on thechannel conditions. Accordingly, the bandwidth available at theapplication layer changes: it increases when channel conditionsimprove, and decreases when channel conditions worsen. Obviously,the encoding parameters must be tuned to adapt thevideo source transmission rate to the available bandwidth. Thisadaptation is achieved by means of appropriate feedback laws,which are relationships between the encoding parameters to beused and other variables representing the state of the system. Inthis paper, an analytical framework is introduced which can beused for the design of the feedback laws. To this purpose both thechannel and the video source are modeled by means of Markovmodels. The resulting model of the whole system is denoted asSBBP/SBBP/1/ . Analysis is derived which allows to evaluate themost significant performance measures and, therefore, to designoptimal feedback laws.