Rate-control for conversational H.264 video communication in heterogeneous networks

摘要

The transmission bit rate available along a communication path in a heterogeneous network is highly variable. The wireless link quality may vary due to interference and fading phenomena and, peered with radio layer reconfiguration and link layer protection mechanisms, lead to varying error rates, latencies, and, most importantly, changes in the available bit rate. And in both fixed and wireless networks, varying amounts of cross traffic from other nodes (i.e., the total offered load on the individual links of a network path) may lead to fluctuations in queue size (reflected again in a path latency) and to congestion (reflected in packet drops from router quenes). Senders have to adapt dynamically to these network conditions and adjust their sending rate and possibly other transmission parameters (such as encoding or redundancy) to match the available bit rate while maximizing the media quality perceived at the receiver. We investigate congestion indicators and their characteristics in different multimedia environments. Taking these characteristics into account, we propose a rate-adaptation algorithm that works in the following environments: a) Mobile-Mobile, b) Internet-Internet and c) Heterogeneous, Mobile-Internet scenarios. Using metrics such as Peak Signal-to-Noise Ratio (PSNR), loss rate, bandwidth utilization and fairness, we compare the algorithm with other rate-control algorithms for conversational video communication.