RU-2861411-C1 - METHOD FOR TRANSMITTING VIDEO DATA IN UNSTABLE NETWORK ENVIRONMENTS

RU2861411C1RU 2861411 C1RU2861411 C1RU 2861411C1RU-2861411-C1

Abstract

FIELD: data transmission systems. SUBSTANCE: invention relates to distributed systems for streaming video data over communication networks, and is intended for operation in unstable network conditions with low latency and high fault tolerance requirements. The field of application includes telemedicine and industrial video monitoring. The essence of the invention lies in the implementation of a hybrid error correction scheme based on Reed-Solomon codes with parallel pipeline data processing on the transmitting side. A FIFO ring buffer with double indexing is used, and data transmission is carried out in UDP datagrams with a Predictive NACK mechanism, which reduces the number of retransmission requests by predicting losses based on transmission statistics. On the receiver side, a multi-level priority queue and binding of computational threads to processor cores are used to ensure reliable decoding and playback synchronisation. EFFECT: ensuring guaranteed delivery of video streams with minimal delays and high reliability in complex network conditions. 1 cl, 3 dwg

Inventors

Babanov Zakhar Dmitrievich
Kukunin Dmitrii Sergeevich
Maksimenko Sergei Olegovich
Podaiko Svetlana Viktorovna

Dates

Publication Date: 20260505
Application Date: 20251124

Claims (1)

A method for transmitting video data in unstable network environments, which consists in the fact that on the sender's side, an array of video data is divided into blocks that are encoded for error correction, transport frames are formed and transmitted to the network using a transport protocol, and on the recipient's side, frames are received that are decoded, restoring the lost data, characterized in that on the sender's side, parallel processing of video data is carried out using independent computing threads assigned to processor cores, while a hybrid error correction scheme is used, including non-systematic encoding of the main payload of video data and systematic encoding of packet headers using Reed-Solomon codes, the original data are buffered using a FIFO circular buffer with double indexing, while encoding is activated when a threshold block size is reached, and the formed frames are encapsulated in UDP datagrams without waiting for confirmation of receipt, and on the recipient's side, reception and initial verification of the integrity of frames is carried out, while multi-threaded decoding is performed using distributed Atomic counters are used to track the block recovery status, a predictive retry mechanism is used to manage retransmissions based on the analysis of anomalies in the packet sequence, decoded packets are buffered in a multi-level priority queue to smooth out jitter and prioritize critical frames, and after synchronization by timestamps, the video is output to the playback device.

Description

The invention relates to the field of data transmission systems, in particular to distributed systems for transmitting video data over communication networks, and is intended for operation in unstable network conditions with requirements for low latency and high fault tolerance. A patent known as "Method for transmitting data over packet communication networks with the ability to recover data in the event of data loss due to connection errors in communication networks" (Russian Federation Patent No. 2706879, IPC H04L 12/805, published November 21, 2019) pertains to methods for transmitting data over communication networks and includes the steps of dividing the source data into blocks by forming data frames with a synchronization sequence, a header with protocol adaptation parameters, and data groups containing data blocks and error correction codes. The method includes forming a composite frame, in which the data frames located within are interleaved at the byte level, thereby increasing the stability of the transmission. Transport data frames contain information for adaptively adjusting the operating modes of the receiving side, which ensures the recovery of lost data without repeated requests. The disadvantage of this method is that the error correction mechanism used cannot cope with high data losses, leading to an increased number of retransmissions, reduced transfer speeds, and buffer overflows. This increases the volume of service traffic, which reduces overall network throughput. A known analogue of the "Device and Method for Transmitting a Transmission Stream and a Processing Device" (RU Patent No. 2677572, IPC H04L 12/851, published on January 17, 2019) discloses a device and method for transmitting a data stream, comprising an information insertion module and a transmission module. The information insertion module is configured to determine the presence of first identification information in data packets encapsulated in transmission packets and to set second identification information in the transmission packet, which has a first value in the presence of the first identification information or a second value in its absence. As a result, transmission packets containing priority data can be identified and processed with increased priority. The transmission module is a device transmitting a data stream with the second identification information set, located outside the data packets. The transmission packets are multi-layered packets containing multiplexed transport packets, such as TLV, GSE, MMT, RTP or FLUTE packets, which ensures flexibility and efficiency of data transmission. The disadvantage of the specified device and method is that in the absence of information identifying priority packets, the data processing speed is reduced, which leads to an increase in the system response time, thereby reducing throughput under high load conditions. The patent "Information Processing Device and Method and Program" (RU Patent No. 2673102, IPC H04L 1/00, H03M 13/35, published November 22, 2018) is a method and device for processing information. The invention is a circuit that automatically selects a forward error correction (FEC) coding system, such as XOR or Reed-Solomon code, based on the current bit rate and its comparison with a predetermined threshold. When the bit rate changes during a communication session, the circuit re-compares it with the threshold and, if necessary, switches to a more suitable FEC system, also taking into account the data compression state. This ensures low-latency and highly reliable data transmission. The disadvantage of this device is that when using a fixed FEC coding system, the efficiency of recovering lost packets under variable bit rate conditions decreases, which leads to data loss. The prototype is the "Method for Transmitting Information in Hybrid Public Communications Networks Based on Equivalent Reed-Solomon Codes" (RU Patent No. 2838539, IPC H04L 1/00, published April 21, 2023). The method for transmitting information in hybrid public communication networks is based on the use of equivalent Reed-Solomon codes represented as linear recurrent sequences of maximum length. On the transmitting side, the data array is divided into frames, each of which is transformed into a homogeneous recurrent sequence taking into account the characteristics of the Galois field. The sequences are then packed into IP packets and transmitted over the network using the UDP protocol. On the receiving side, the IP packets are accumulated until a complete set is formed, determined by the cyclic closure of the set. Upon its receipt, the termination of the transmission session over the TCP connection is initiated. Decoding is performed using the dual basis of the Galois field. A drawback of the known prototype method is the inability to quickly recover information elements at the receiving end due to the need to accumulate a full set of packets before decoding, which leads to delays. Furthermo