How does the TCP reliability algorithm work?

The Transmission Control Protocol (TCP) is the internet protocol that provides a set of rules for communication between different devices on a network. There are several benefits of using TCP over UDP in many cases. For instance, TCP is better for sharing documents, and UDP is better for playing online games. Moreover, TCP provides reliable communication, flow control, congestion control, and packet retransmission if the packet is lost or has not arrived at the destination. In this Answer, we will discuss how the TCP reliability algorithm works.

TCP reliability algorithm

We will divide the TCP reliability algorithm into the following three steps:

1. Establishing the connection.

2. Transferring the data.

3. Terminating the connection.

Establishing the connection

As discussed below, connecting the client and the host is a three-way handshake process.

• The client has to initiate the connection with the host by sending a TCP segment to the host. This TCP segment is a synchronized packet (SYN). The SYN packet contains the control flags and an initial sequence number to identify the connection.

• As a host receives the SYN packet, it generates its sequence number and sends back the SYN-ACK packet as an acknowledgment that the SYN packet has been received. When reaching the client, this SYN-ACK packet indicates that the host can connect with the client.

• Lastly, the client acknowledges the connection by returning the ACK to the host.

The diagram of the three-way handshake process is illustrated as,

Transferring the data

After the connection is established successfully, the client will send the data, and the host will receive the data. The main features of transferring the data between the client and the host are given below:

• When the client sends the packet of data, it must receive an acknowledgment from the host that the packet of data is received within the interval.

• The acknowledgment sent by the host is such that it increases the acknowledgment number by the length of the received packet.

• In the above diagram, acknowledgment number 32 indicates that the host received and successfully processed all the data up to byte 31 and expects the next byte to be byte 32.

• In the worst case, if the packet gets lost due to flow control or congestion control, then after the time interval, the acknowledgment sent by the host will be 1.

• After receiving the acknowledgment of 1, the client will understand that the host does not receive the packet. Then it will re-transmit the packet of data.

This is how the sequence and acknowledgment numbers help us determine which packet the host receives. This way, we can ensure the reliability of TCP.

Terminating a connection

When the data is transferred from the client to the host or vice versa, it is time to terminate the TCP connection between the client and host. Remember, only the client can instantiate a connection with the host, but in case of terminating a connection, either the host or the client can do the termination process.

Let's take a case where the client's computer is closing the connection, the illustration of which is shown below,

1. A client computer sends a packet with FIN ( Finish = 1) byte.

2. A host computer sends the packet with ACK = 1 and FIN = 1.

3. A client computer sends the packet with FIN byte = 1.

This results in the termination of the connection between the client and the host.

Conclusion

From this Answer, we have learned about the TCP reliability algorithm. We have also discussed the TCP retransmission of the packets to maintain the reliability of transferring the data so that every packet reaches the host. Furthermore, we have understood the establishment of a connection, terminating a connection, and the role of the acknowledgment and sequence numbers in maintaining reliability.