Anti-packet loss method, device and system in real-time communication

Abstract

The invention discloses an anti-packet loss method, a device and a system in real-time communication. The method comprises steps: a network data packet is transmitted to a receiving end, wherein the network data packet comprises video frame data after coding and FEC (Forward Error Correction) redundant data, and the FEC redundant data determine a corresponding FEC redundant level; feedback data transmitted after the receiving end analyzes and restores the network data packet are received; when the feedback data are packet loss feedback data, according to the current packet loss model, the corresponding level of FEC redundant data is reselected from preset multiple groups of redundant levels of FEC redundant data, and the data are transmitted to the receiving end; besides, when the feedback data are decoding error feedback data , according to the current decoding error model, an independent decoding frame corresponding to the decoding error frame is acquired and transmitted to the receiving end. The problems that processing on the network packet loss can not meet requirements of high quality and low bandwidth in the prior art, and the anti-packet loss ability is poor can be solved, and the anti-packet loss ability in the real-time communication process is enhanced.

Description

technical field [0001] The present invention relates to the technical field of communication, in particular to a method, device and system for anti-packet loss in real-time communication. Background technique [0002] In real-time communication, audio and video media are usually transmitted using UDP (User Datagram Protocol, User Datagram Protocol). Since UDP is a connectionless transmission protocol and a best-effort service method, it cannot provide reliable transmission. Therefore, packet loss will inevitably occur in the network, resulting in discontinuous sound, blurred video, etc., seriously affecting communication quality and user experience. Therefore, anti-packet loss technology is an important technology related to user experience. [0003] Currently, technologies for anti-packet loss generally include: FEC (Forward Error Correction, forward error correction), retransmission, and interleaving. Generally speaking, the time delay of interleaving is relatively large,...

AI Technical Summary

Problems solved by technology

Generally speaking, the delay of interleaving is large, and when packet loss is caused by congestion, blind retransmission will aggravate network congestion. Therefore, the application of retransmission and interleaving in real-time communication systems is subject to some restrictions

[0004] Based on the above considerations, FEC is usually used for anti-packet loss in real-time communication, and because FEC achieves the purpose of anti-packet loss by increasing redundancy, it will lead to further aggravation of congestion when the network is congested

Moreover, in the era of mobile Internet, another negative effect of increasing redundancy through FEC is that too much FEC will consume a large amount of network traffic, therefore, expensive communication costs will be generated

If the redundancy of FEC is relatively high, the packet loss can be recovered to the maximum extent, but the bandwidth of redundant data will be high; on the contrary, if the redundancy of FEC is relatively low, all packets cannot be guaranteed to be fully restored, which will cause video The display has problems such as blurred screens and freezes, which reduces user experience

[0005] A common way to solve this problem is to determine the FEC redundancy level based on the packet loss rate of network statistics. However, since the packet loss rate fed back by the network is generally based on historical statistics and cannot fully reflect the current network status, FEC redundancy The level corresponds to the average status of the network in the historical time. Under the current network status, it is not guaranteed to restore all packets. Therefore, FEC cannot achieve the expected effect, and thus cannot fundamentally solve the problem.

Images