A new bandwidth prediction based wireless Ad-Hoc video communication

In this paper, we propose a new bandwidth prediction and control scheme for transmission of real-time video stream over wireless ad hoc network. Our proposed strategy is based on cross-layer, feedback and neural network techniques. We first determine several relevant parameters which directly impact the quality of video transmission. We obtain the topology change, the signal power information by using cross-layer mechanism. We obtain the packet loss statistics and the network jitter from receiver feedback. Subsequently, we extract their relevant features from those sample data, and take them as the input for training data of the neural network. The predicted result of the neural network would be used as the bandwidth of mobile ad-hoc network (MANET). This parameter will be used to control video encoder to adjust the bit rates of real-time video stream. We have implemented an integrated video streaming system simulation for proving our proposed scheme. In contrast to the conventional transfer scheme, the results of our proposed control strategies indicate that there are considerable improvements in the packet loss and the real time visual quality.