Congestion Control Techniques in a Computer Network: A Survey

Abstract

With the advance in technology the focus has shifted from desktops and laptops to handheld devices like tablets, phablets, mobile phones, PDA etc. giving rise in the number of users connected to internet. About 900 million computers are connected to internet [23].In an hour 383 thousand TB of data transmission takes place [24]. As the traffic on internet increases giving rise to the problem of congestion. Several congestion techniques have been proposed, in this paper the summary of several congestion techniques and methods over the years has been discussed. One of the latest approaches to control the congestion is based on Neural Networks is also included in this paper. KeywordsCongestion control, TCP, Neural Networks, ECN, XCN, ATM, MANETS, EWCCP, Back propagation algorithm. INTRODUCTIONA Network is said to be in the state of congestion when the total demands for resources exceeds the total available resources .In other words we can say a network is in a state of congestion when the quality of the service delivered to the user decreases, whenever there is increase of load in a network. Take an example of downloading a 1 GB file .When there is no congestion, the file gets downloaded in few minutes, but on day when there is congestion in a network same file gets downloaded in hours. The problem of congestion has become an important issue these days. BACKGROUNDIn the year 1999 Saverio Mascolo in his paper titled [21] states that there is a large bandwidth delay products in high speed communication networks thus resulting in adverse conditions on closed-loop congestion control algorithm. By using the combination of classical theory and Smiths principle a law for congestion control for high speed communication networks can be designed. This control law is transformed into discrete values and window form for ATM and internet respectively thus guarantying full and stable utilization of queue, links during transit and steady phase.In the year 2001 Deepak Bansal And Hari Balakrishnan in their paper titled [6] states that congestion results in reduction of transmission rate that is problematic for voice and video applications ,therefore an algorithm that is nonlinear in nature is needed naming this algorithm as binomial Algorithm. TheBinomial algorithm makes use of TCP style for Addictive increase and multiple decrease for increasing and decreasing the transmission rate. In addictive increase the rate of increase is inversely proportional to power p of current window size and for multiple decrease, rate of decrease is proportional to power z of current window size. There can be infinite number of such binomial algorithms, those that satisfy p +z =1 and all binomial algorithms converges for the fairness.In the year 2002 J.Alan Bivens in his paper titled [10] predicts the congestion in a network using neural network learning technique, in which neural networks are used to predict the source of congestion, the system has been divided into three programs, one implementation consists of c language used for manipula...

A Radial Basis Neural Network Controller to Solve Congestion in Wireless Sensor Networks

“…data loss due to collisions are independent design aspects and thus can be abstracted when studying a congestion avoidance and control protocol. Table (1) shows the important information about the designed network. In the matlab 250 sensor with three base stations are deployed randomly in a rectangular area as shown in figure (8).…”

“…In (2015), Mirza Waseem Hussain Sanjay Jamwal, and Majid Zaman [1], produced a survey of a several congestion techniques about congestion control technique in a computer networks, One of the latest approaches to control the congestion is based on Neural Networks. Which will be useful for new researchers exploring its use in their research problems.…”

“…The following figure (10) represent the number of traffic in the memory of the base station that coming from ON nodes with and without RBFNNCC. From the above figure one can see the role of RBFNNCC in managing the traffic that received to the base station in 10 times in fig ( 9) the base station has congestion (in rounds time 1,2,3,4,6,7,8,9) and losses some of the received data, and other once has no congestion (time 5,10) in the case without using RBFNNCC. But with using RBFNNCC all times has traffic less than the size of the memory of base station (270).…”

“…This means that the quality of the service (QOS) transferred to the employer decreases. If there is no congestion, when downloading a 1 GB file, the file take a few minutes to downloaded, but the same file gets downloaded in many hours when there is congestion in a network [1]. The Sensor Network can be seen in our lives in different aspects.…”

Traffic Shaping for Congestion Control

“…The algorithm does not add more congestion to the network. (Hussain, Jamwal, & Zaman, 2015) When the sending node is about to send data packets, at the point where connection occurs the sending node together with the carrier links agree as to a specified network traffic pattern. The data is then passed to flow strictly through the specified traffic policy.…”