Fast and reliable congestion control improved algorithm based on user datagram protocol (UDP)

Abstract

The invention provides a fast and reliable congestion control strategy (FRUDP) which is based on a user datagram protocol (UDP) of a transport layer, and is designed to guarantee an information flow to run high-efficiently and stably in a high speed network environment between an application layer and the transport layer on the basis that network address translator (NAT) equipment can be penetrated through and transportation stability are guaranteed. The FRUDP is mainly formed by an estimating module, a window control module, a data processing model and an emergency control module, wherein the estimating module is arranged at a sending end, does not need to modify a network layer protocol, estimates transmission paths and propagation delay through collaboration according to the local information capable of being obtained by the sending end, and accurately estimates the actual number of the queued packages remaining in periodic line buffer areas in each process. The window control module designs congestion control strategies in a collaboration mode at the sending end, taking the queue size of a bottleneck periodic line buffer area as a balance point, actively controls the queuing length of periodic line buffer areas, and actively prevents the overflow of the queues of the buffer areas and congestion phenomena from occurring. The data processing module determines which groups to be sent from a candidate pool. The emergency control module determines when to send the groups.

Description

technical field [0001] The invention relates to the field of network transmission control protocol algorithm improvement, in particular to an improved algorithm that can simultaneously meet the requirements of fast transmission, reliability and penetration of NAT equipment when the transport layer selects the User Datagram (UDP) protocol. Background technique [0002] The current Internet is based on the fourth version of the Internet Protocol (IPv4 for short). With the rapid popularization and development of various network applications, IP address resources are becoming increasingly scarce. In order to solve this problem, various enterprises, companies and Internet service providers have deployed a large number of network address translation devices (Network Address Translation, NAT). Using these NAT devices and corresponding NAT technologies, all hosts in a LAN can access the Internet through one or several public IP addresses. However, due to the different addressing me...

AI Technical Summary

Problems solved by technology

The BA2RUDP algorithm has saved a lot of time and resources compared with the previous two RUDPs, but it still has shortcomings: the sender has only one sending pointer, that is, after the sender receives the confirmation every time, if there is a data packet that needs to be retransmitted, The sending pointer has to temporarily stop the sending operation, find the corresponding data packets that need to be resent, and perform the resending operation, and then continue the subsequent sending after the resending operation is completed

Since the Internet is a complex giant system, coupled with its inherent heterogeneity and large-scale distribution, the congestion information at the link end cannot be fed back to the sending end in time

When the router is congested, the sender needs to go through a round-trip delay time to obtain the packet loss information. To be dropped, excessive packet loss seriously affects system performance

[0008] (2). Global synchronization problem

But the sender starts to increase the sending speed after learning that the network is no longer congested, which eventually causes network congestion

And this phenomenon often goes on again and again, so that the network is in a state of low link bandwidth utilization for a period of time, reducing the overall throughput. This is the so-called "TCP global synchronization" phenomenon

[0009] (3). Poor high-speed network adaptability

When the data sent by the sender exceeds the bandwidth of the bottleneck link, the buffer queue length of the bottleneck link will increase

When the queue length increases to a certain extent, the buffer will overflow, resulting in a large amount of data loss, congestion, and a sharp drop in network performance

[0011] At present, there is no good way for the sender to accurately and timely estimate the network congestion status

Images