Hello Prathamesh!
The TCP sequence number is random because, as Rene mentioned earlier, TCP is vulnerable to security issues like spoofing, injection or connection resetting. If an a attacker is able to determine the sequence number, he/she would be able to spoof a trusted source, and thus compromise the TCP session.
For an example of such a spoofing attack, take a look at the following link: http://www.citi.umich.edu/u/provos/papers/secnet-spoof.txt
(Notice that this paper is almost 20 years old and yet it’s still valid today!!! Shows just how well designed and robust these protocols are.)
The sequence number is generated by the OS of the device sending the data. Each OS may use a different method for that calculation, but in general it is calculated as a function of the current time. The method has to be random enough so that the sequence numbers cannot be predicted.
Concerning the ending of a TCP session, the initiator is responsible for sending a TCP segment with the FIN flag set in the header. This indicates to the receiver that the initiator is requesting to end the session. The receiver responds with the ACK flag set in the header. Next, the receiver responds with a FIN flag and the initiator responds with an ACK. The session is ended. Take a look at the following image that depicts this process.
http://www.iptel.org/~jiri/publications/da/html/da-6.gif
I hope this was useful for you!
Laz