Paper Review : A Protocol for Packet Network Intercommunication


Reviewer : Seh Leng Lim


The paper proposes the following 3 key main ideas :

(a)    A transmission control program (TCP) protocol that allows for delivery of packets between dissimilar packet switched networks and from one program on a host to another program on another host

(b)   The idea of a sliding window protocol to handle retransmission and duplicate detection

(c)    The idea of fragmenting larger packets into smaller packets as they pass from one network to another network


The main contribution of the paper is to propose in detail the concept behind the transmission control program protocol which can be and has already been implemented.


I think that the paper was a breakthrough at the time it was published. I see that the paper has a rating of 5. This is because it proposes many elegant and revolutionary ideas to solve the problems of intercommunication between packet networks. However, it was not able to foresee the explosion of the Internet, and hence it makes the claim that 65536 addresses per network is sufficient, leading to the design of addressing schemes that are no longer sufficient for today’s use. Also, it did not foresee the need to provide the different quality of services, and hence IP was not abstracted out as a separate layer. The paper is purely based on abstract reasoning, and there is no real experimental data to support it. However, it has done a very good conceptual job as what it proposes have mostly been implemented.


From this paper, system researchers and builders can better appreciate the considerations behind the TCP protocol. This may provide them with the background knowledge in the design of new protocols for the future.