Congestion control bitrate algorithm

Abstract

The method may include sending, via an unreliable protocol, a stream of data packets to at least one recipient device over a network. The stream of data packets may include source packets and forward error correction (FEC) packets. The method may include, during said sending, receiving one or more feedback reports from the at least one recipient device, each said periodic feedback report characterizing packet loss during a corresponding period of time. The method may also include, during said sending, adjusting a rate at which said data packets are sent in the stream in response to at least one of said feedback reports. Adjusting the rate may include increasing a FEC rate at which the FEC packets are sent while maintaining a source rate at which the source packets are sent in response to a one of the feedback reports which characterizes the packet loss as within an acceptable level.

Description

[0001] field [0002] This disclosure relates to data transmission over a network. In particular, various aspects of the present disclosure relate to systems and methods for controlling congestion using unreliable transport protocols in packet-switched networks. Background technique [0003] With the growing popularity of digital streaming services and various cloud-based computing solutions, the ability to quickly and accurately transfer large amounts of data between remote devices is mission-critical. Sending digital data to a target system over a shared resource of a network such as the Internet, a Wide Area Network (WAN), or a Local Area Network (LAN) typically involves an arrangement of converting the data into formatted blocks, called packets, which may have fixed or variable length. Each data packet generally includes a payload or body with basic user data being conveyed to a destination and some supplementary information, usually at least partially contained within t...

AI Technical Summary

Problems solved by technology

This may be tolerable for less time-sensitive applications involving reliable data transfer (such as TCP), since retransmission of lost data is guaranteed; however, this may be unacceptable for many real-time applications and others involving unreliable transfers Yes, because packet loss can reach such a level that the receiver cannot reconstruct the lost data, leading to undesirable results such as loss of signal

On the other hand, it is also undesirable when the maximum available bit rate is changed far beyond the bit rate offered by the sender, as the result is that the full transmission capacity of the network channel is used inefficiently and the signal quality at the receiver side may be unnecessarily poor

[0009] Unfortunately, transmitting data using unreliable protocols in a manner that efficiently uses the available bandwidth of a network channel without incurring congestion conditions that result in unacceptable packet loss is a huge challenge

Traditional congestion control techniques are often only suitable for reliable protocols (such as TCP), which have feedback to the transport layer of the built-in sender, but are ineffective for many unreliable protocols (such as UDP), which often lack the required Feedback, unless added separately on the transport layer by enabling the user

Furthermore, congestion control or congestion avoidance algorithms designed for TCP or other reliable protocols will generally not be fast enough for real-time streaming applications or and may not be suitable for many data transfer applications involving unreliable protocols, because the response to the congestion bit rate Exponential reduction may result in an unduly affected quality of the real-time signal

Furthermore, while packet loss caused by increasing the bit rate to the point of congestion may be tolerable in less time-sensitive applications (which use TCP or other reliable protocols to retransmit data), it may not be possible in many real-time applications Accepted, this is due to the inability of the sink to reconstruct the data

Images