论文部分内容阅读
基于流媒体客户端缓冲的被动变化可以通过自适应媒体播放(AM P)技术修改播放速度来调节,为了减轻AM P中速度突变对播放质量的影响,提高变速的性能,提出了一种基于多层神经网络控制的AM P方法,产生随当前缓冲动态变化的速度,并且保持缓冲稳定在一定的范围内。该方法引入了多层神经网络控制结构,并采用反向传播学习算法(BP)进行离线训练。仿真结果表明:该方法产生的速度平均幅值和变化增量可以减少0~100%,性能比原AM P方法更好。
Passive changes based on buffering of streaming media clients can be adjusted by modifying the playing speed through adaptive media playback (AMP) technology. In order to reduce the impact of speed abrupt changes on the playback quality in AM P and improve the performance of transmission, a method based on multiple The AM-P method controlled by layer neural network produces a dynamic change with the current buffer speed, and keeps the buffer stable within a certain range. This method introduces a multi-layer neural network control structure, and uses BP back-propagation training algorithm. The simulation results show that this method can reduce the average amplitude and increment of speed by 0 ~ 100%, and has better performance than the original AM P method.