In real-time communications (RTC) over the internet (WebRTC), an optimal trade-off is sought between the quality and the latency of transmission. For example, a high-resolution video call requires a large throughput, which can cause congestion at network nodes, in turn resulting in unacceptable call delay. Current RTC protocols use congestion control mechanisms to try to achieve an acceptable quality-versus-latency trade-off. Under certain circumstances, these protocols are sensitive to delay spikes, compensate inadequately for large data packets, lack robustness in situations where media encoders over-produce data, and use test probes that often return inaccurate estimates of network capacity.

The techniques of this disclosure address the problems of congestion control mechanisms by estimating the current network state using particle filtering. The network state is used to set the bit-rates of media encoders such that the quality-latency trade-off is optimized.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.