Look At Transmission Control Protocol Information Technology Essay

Transmission Control Protocol is one most reliable ,connection oriented communication protocol used in the internet traffic .The main aim of this section is to conduct a research on TCP friendly protocols and find a suitable answer to the questions like the features of TCP that are not suitable for real-time multimedia application, TCP friendly congestion control for non-TCP application etc.

I. Features of TCP not suitable for real time multimedia application.

The uses of multimedia applications are increasing day to day in the internet traffic. In recent future it is sure that the real-time multimedia applications increase rapidly and make up a considerable portion of the total internet bandwidth. TCP is full duplex, connection oriented communication protocol that is used for transmitting data that mainly focuses on the accuracy of data. But some features of TCP are not suitable for real time multimedia application which includes the following.

Reliability through Retransmission

Since TCP is a reliable protocol it uses error control mechanisms like retransmission when an error occurs. Retransmission delay is one of the most important problems in multimedia application for TCP oriented connection. This retransmission is useful in traditional application such as http, FTP, telnet etc which require the loss free data transfer between the communicating media. In the case of multimedia application there is an massive emission of data into the network especially in the case of video. This will cause network congestion, since TCP is a reliable protocol it will reduce the window size and data emission rate to get out of this congestion. The retransmission causes end-to-end latency which is not suitable for real time multimedia applications where the receiver needs the data at the same time when the sender sends it.

Congestion control

In TCP, the available bandwidth can be monitored by using congestion algorithm which uses the transmission rate for the manipulation results in some deliberate waste. In the cases where there is no congestion, TCP congestion control fairly shares the available bandwidth. Over shorter time scales, the TCP instantaneous rate just looks like a saw- tooth. The cycles are between periods of additive increase separated by multiplicative decrease (AIMD). This rate fluctuation is not suitable for multimedia streaming application because it affects their smooth functioning.

Socket based application programming interface

The sockets based application programming interface does not support multihoming (a technique to improve the reliability of internet connection of an IP network). This result in an application can use only a single IP address to a particular TCP connection with another host. In an occasion where the interface associated with that IP address goes down which result in the connection loss and reestablishment. This is not suitable or multimedia applications like VoIP.

TCP does not support multicasting

In most of the multimedia communication like audio and video conferencing, the network traffic will be high and large amount of data flow across the network. In such situations data are transmitted by using multicasting techniques between the sender and the receivers. But TCP does not support multicasting

TCP does not support time stamping.

The major problem faced in real time multimedia applications like video and audio conferencing is the jitter. The major solution for the problem associated with jitter is the use of time stamps. The timestamp at each packet shows the time at which each packet is produced when compared with the previous one. So that the receiver knows when each packet is to be played. But unfortunately TCP does not provide support for time stamping.

TCP is connection oriented

The TCP is connection-oriented protocol. In this first a connection is established between the source and the destination before data transfer. The main disadvantage of this connection oriented protocol in real time communication like VoIP is the consumption of resources spent setting up the call. This connection oriented feature made a little overhead during connection establishment because it is time taken during certain occasions. Time overhead is not suitable for real time multimedia applications.

Do not require full duplex transmission

In TCP, the processes that reside in the internet establish a reliable serial communication path and exchanges full duplex stream of bytes. The processes within the communication are identified by the IP addresses. The full duplex connection is not compulsory for most of the multimedia application. The best example is TV broadcast over LAN; in most situations it uses simplex continuous media connection.

Acknowledgement

In TCP, there is the need of acknowledgement during the transmission of each packet. The positive acknowledgement causes overhead because all packets are sent with a fixed rate. This is not suitable for multimedia.

II) UDP is TCP Unfriendly protocol

In order to discuss about this topic we need to describe the basic functions of the TCP and UDP. In order to clearly define about the TCP, UDP unfriendliness we need to understand the basic concepts in detail. With the increased use of real time applications internet users looking for more reliable protocol that provide more faster medium for audio and video.

Since TCP is a reliable protocol it has the capacity to reduce the amount of data loss, noise and out of sequence data. TCP uses connection based communication between the communicating hosts. This connection based feature of TCP helps in the guaranteed delivery of the packets. It has flow control and congestion control.

In the case of UDP, it is unreliable. There is no guarantee for packet delivery and for the sequence. In this protocol there is no connection establishment before the communication starts and it sends the data according to its availability.

In real-time communication (for audio or video), both the communicating ends look for speed rather than accuracy. The connection less feature of UDP, that sends the data when they are available and not try to resend once there is an error or packet loss made UDP much faster than TCP. This increased speed made more popularity for UDP in streaming media over internet.

When compared to TCP the UDP has a main drawback, since UDP continuously sending packets through the network without considering there is congestion on the network. It does not take any measures to reduce the congestion in the case where congestion occurs. But TCP make use of the congestion control mechanisms like reduce the transmission rate. Since UDP does not make use of any congestion control algorithms when congestion occurs this tend to take the utilisation of the available bandwidth over the internet. For this reason UDP is known as a TCP unfriendly protocol. According to the opinion of Ross (2005, p.198) in his book ‘Computer Networking’ specifies that “UDP is the most commonly used protocol for multimedia application such as Internet phone, real-time video conferencing and streamed of audio and video”. So in order to overcome these problems of congestion controls with UDP and improves the band width utilisation and also a fair towards competing TCP connection result in the development of new technology called TCP friendly technique. With the increased popularity of the streaming media the UDP traffic becomes a problem for the internet as a whole.

III) TCP Friendly congestion control protocols

The popularity and development of internet leads to the extensive use of real-time multimedia applications like video conferencing, chatting etc. All these applications make use of non congestion controlled protocols like UDP for transmitting continuous video, audio through the network. This type of non congestion controlled traffic is called unresponsive flows. The major feature of the unresponsive flow is that they do not reduce their load on the network when there is a packet loss. They continuously transmit the data through the network at the rate with which is encoded without bothering about the network congestion. This type of UDP traffic causes traffic congestion in the network as well as it produce an extreme unfairness towards the network. In order to solve these problems, we need to improve the UDP for proper bandwidth utilisation, reducing loss ratio and also produce a fair towards the competing TCP connection and this is called as TCP Friendly. In this section we are discussing about the congestion control mechanisms used by TCP friendly protocols like TFRC and TFMC.

TFRC (TCP-Friendly Rate Control)

TFRC is mainly defined as a congestion control mechanism for unicast flows in the Internet. TFRC is evolved from TFRCP protocol and it is mainly used for unicast communication but sometimes it is used for multicast communication with some modifications. The lower variation of throughput over time makes this much more suitable for telephony or streaming media and which has a smooth sending rate.

TFRC uses equation based congestion control for unicast traffic. In this mechanism it explicitly adjust its sending rate as a function of the measured rate of loss events where the loss events contain one or more packet dropped within a RTT. As it follows the equation based congestion control, TFRC directly make use of the TCP throughput equation which makes it possible to compete fairly with TCP. The following is the throughput equation used in the TFRC for congestion control:

In this equation

X is the transmit rate in bytes/second.

S is the packet size in bytes.

R is the round trip time in seconds.

B is the no of packets acknowledged by a single TCP acknowledgement.

t_RTO is TCP transmission timeout value in seconds.

The general working of the TFRC congestion control mechanism as follows:

As the first step the receiver calculate the loss event rate and gives this information back to the sender.

Using this data the sender calculates round-trip time (RTT).

Using this loss event rate and RTT, calculate acceptable transmit rate using the TFRC throughput equation.

Finally the sender adjust the transmission rate to match the calculates rate.

By using the receiver based mechanism TFRC is well suited for applications like the sender are a large server handling many concurrent connections and receiver is has more memory and CPU cycles available for computation.

TFMC (TCP friendly Multicast Congestion Control)

TFMC is the single rate congestion control scheme for multicast transmission. TFMC is single rated simply means that the receiver experiencing worst network condition will adapt to the sending rate. It has relatively low variation of throughput over time, which is suitable for applications that require a smooth sending rate like video streaming.

The general working of TFMC congestion control as follows

First step is the calculation of loss event rate and RTT by the receiver. By using this information with the help of TCP throughput equation the receiver derive a TCP friendly sending rate. In order to prevent feedback implosion at the sender the feedback is supplied only to a subset of the receivers using feedback suppression mechanism. In the receiver side the feedback is not suppressed and it calculates the transmission rate and gives it to the sender. This report has two main purposes, it helps to calculate the RTT for receiver and give the appropriate transmit rate for the sender.

IV) TCP friendly congestion control for non-TCP applications.

There are many studies conducted on TCP friendly congestion control for non TCP application. The congestion control mechanisms are used to achieve the following aims:

Use of congestion control mechanisms help the applications to use the bandwidth more efficiently and thus increases the performance

Those applications which adapt to the network can easily make use of the wider bandwidth and more useful of the Internet

Congestion Control algorithms helps to reduce the Congestive Collapse ie the situation in which network performs a little work due to the over utilisation of network links.

There are a number of TCP friendly congestion control mechanisms are introduced now a day’s which include end-to-end schemes and hop-by-hop mechanism. The end- to- end mechanism consist of AIMD window based and rate based protocols and hop-by-hop consist of rate based hop-by-hop mechanism. The end-to-end congestion control mechanisms are necessary for responsive as well as unresponsive flows which reduce the traffic load when packet loss occurs. In rate based hop-by-hop congestion control scheme it matches the sending rate of the connection to the service rate observed at the downstream node.

In window based AIMD schemes which utilises the increased and decreased congestion window of the sender. This mechanism has some advantage as well as some disadvantages. The major advantage is that it is familiar, reasonable in the field of fairness, stability, oscillations and other properties. The disadvantage is that it has more abrupt changes and oscillation in sending rate which is not suitable for real time streaming application.

In rate based AIMD schemes, the most common examples are RAP (Rate Adaptation Protocol) and LDA (Loss-Delay based Adjustment Algorithm). The LDA is mainly proposed for adapting the transmission rate of multimedia applications to the congestion level of the network. It mainly relies on end-to-end Real Time transport Protocol (RTP) and its control protocol (RTCP) for feedback information. The major difference between rate adaptation schemes and window based schemes is that the latter is not ack-clocked. The main advantage is that the adaption of the sending rate deduces the no of oscillations and also helps to reduce the heavy traffic in real-time applications.