From: Michelle Liu (liujing@u.washington.edu)
Date: Tue Oct 05 2004 - 22:10:32 PDT
Review of "A Protocol for Packet Network Intercommunication"
Jing Liu
This paper presents a protocol ¾ TCP, which supports the intercommunication of different packet switching networks.
The very first contribution of the paper is that the author has proposed a protocol to enable the scaling of network sizes. In consequence, remote communication is possible.
In the beginning of this paper, the notion "gateway" has been proposed as an interface between a pair of different networks. Many issues are raised concerning the new architecture of interconnected networks. The author goes through those issues one by one, such as different functions and cooperation of hosts and gateways, uniform addressing scheme, segmentation, fragmentation and reassembling and sequencing problem. The schemes of solving those issues and the motivation for the designing are very clearly stated.
One of the important contributions of this paper is the proposal of retransmission and duplicate detection scheme. With retransmission and duplicate detection scheme, TCP provides reliable transmission through the interconnected networks to some degree. Particularly, congestion control and flow control are possible. Through the received acknowledgement, the source will know how much capacity is available in the network and how much is a safe transfer.
Another contribution is that the connection-free scheme is used in TCP. This scheme saves the cost on connection setup and makes the communication more efficient.
One of the weaknesses of the paper is that the author doesn't mention how the congestion control works, although it mentions that congestion at the TCP level is flexibly handled owing to the robust retransmission and duplicate detection strategy. The author only presents some fundamental issues related to the interconnection of packet switching networks. Thus, there are many things could be improved. For instance, how is the gateway's routing working? How is the performance of this protocol? Is there any simulation result on performance?
This paper has proposed the original model of today's TCP, which is still valid and used in the Internet. Thus, it is quite helpful for us to understand the original design of Internet protocols. The future work for this paper could be detailed specifications of the protocol and performance simulation.
This archive was generated by hypermail 2.1.6 : Tue Oct 05 2004 - 22:10:44 PDT