基于混合流拥塞控制的Ad Hoc网络视频传输

Ad Hoc Network Video Transmission Based on Hybrid Flow Congestion Control

  • 摘要: 基于TCP反馈的主动队列管理算法会无差别地丢弃UDP分组,导致视频质量下降. 根据TCP/UDP发送特性和混合流排队机制,推导拥塞窗口、丢弃概率及混合流队列长度的微分方程,通过线性化获得Ad Hoc网络混合流AQM(active queue management)控制模型,提出了一种减少UDP丢弃率的PI-V主动队列管理算法. 当视频画质下降时,算法依据视频帧权重与分组长度,动态地调整UDP视频流的丢弃概率. NS(network simulator)仿真表明,PI-V算法优于PI与去尾算法,提高了视频传输质量0.82 dB~2.22 dB的峰值信噪比(PSNR).

     

    Abstract: An active queue management (AQM) algorithm, based on TCP feedback, will drop UDP packets indistinguishably, leading to a dropin UDP video quality. Congestion window-size, drop probability, and the differential equations of the hybrid flow queue length are deduced based on TCP/UDP transmission characteristicsand hybrid flow queuing mechanisms. An Ad Hoc network hybrid flow AQM control model is obtained through the linearization of these equations. A PI-V AQM algorithm is proposed, which reduces the drop-rate of UDPs. This algorithm dynamically modifies the drop probability for UDP video stream, adjusting to video frame weight and packet length as the transmission quality changes. NS (network simulator) simulations demonstrate that this PI-V algorithm's performance is better than that of the PI and drop-tail algorithms. The to-be-discussed PI-V AQM algorithm improves video transmission quality by 0.82 dB-2.22 dB, as determined from the peak signal-to-noise ratio.

     

/

返回文章
返回