20.7 INTERPLANETARY (IPN) INTERNET

< Day Day Up >

The Internet, as we know it, is a network of connected networks spread across the earth. The optical fiber–based backbone (the set of high-capacity, high-availability communication links between network traffic hubs) of the Internet supports very high data rates with negligible delay and negligible error rates, and continuous connectivity is assured. If there is loss of packets, it implies congestion of the network.

Now, imagine Internets on other planets and spacecraft in transit. How do we go about interconnecting these internets with the Earth's Internet? Or think of having an Internet service provider to the entire solar system. The brainchild of Vincent Cerf, the Interplanetary Internet (InterPlaNet or IPN), aims at achieving precisely this. IPN's objective is to define the architecture and protocols to permit interoperation of the Internet on the earth and other remotely located internets situated on other planets and spacecraft in transit, or to build an Internet of Internets.

The deep space communication channels are characterized by high data loss due to errors, transient link outages, asymmetric data rates, unidirectional channels, power constrained end systems. To develop protocols to work in this type of communication environment is a technology challenge; also, such protocols will lead to better solutions even to develop systems on the earth.

The three basic objectives of the IPN are these:

Deploy low delay internets on other planets and remote spacecraft.
Connect these distributed (or disconnected) internets through an interplanetary backbone that can handle the high delay.
Create gateways and relays to interface between high delay and low delay environments.

The TCP/IP protocol suite cannot be used for the IPN for the following reasons:

Communication capacity is very expensive, every bit counts, and so protocol overhead has to be minimized.
Interactive protocols do not work and so:
- Reliable sequential delivery takes too long.
- Negotiation is not practical.
- It is difficult to implement flow control and congestion control protocols.
- Retransmission for error recovery is expensive.
- Protocols need to be connectionless.

The proposed IPN architecture is shown in Figure 20.7. The thrust areas for implementation of this architecture are:

Deployment of internets on various planets and spacecraft
Inter-internet protocols
Interplanetary gateways (IG)
Stable backbone
Security of the user data and the backbone

click to expand
Figure 20.7: Architecture of Interplanetary Internet.

Presently work is underway by IPNSIG (Interplanetary Internet Special Interest Group: http://www.ipnsig.org) to define and test the new set of protocols and the backbone architecture. The work includes defining new layers in the place of TCP and IP, using RF instead of fiber for the backbone network, as well as the addressing issues. (A few years from now, if you have to send mail to someone, you may need to specify .earth or .mars extension to refer to the internets of Earth and Mars.)

The vision of Vincent Cerf, the Interplanetary Internet, will interconnect the various internets located on the Earth and other planets and spacecraft. To develop this network, new protocols need to be designed because TCP/IP protocol stack does not work due to enormous propagation delays, low speeds, and the need for connectionless services.

The time frame for the IPN protocol testing is 2003+ for Mars Internet. It is proposed to use schools as test beds for testing the new protocols. Though the IPN may be of practical use perhaps 20 years from now, it is expected that the outcome of this research will solve many of the high-delay problems encountered on the networks on the earth itself.

Summary

This chapter presented an overview of the TCP/IP protocol stack. Above the physical and datalink layers, the Internet Protocol (IP) layer takes care of addressing and routing. IP provides a connectionless service, and there is no guarantee that all the packets will be received. The packets also may be received out of sequence. It is the job of transport layer protocol to take care of these problems. The transport layer provides end-to-end reliable service by taking care of flow control, error control, and acknowledgements. Above the TCP layer, different application layer protocols will be running, such as Simple Mail Transfer Protocol (SMTP) for e-mail, File Transfer Protocol (FTP) for transferring files, and Hypertext Transfer Protocol (HTTP) for the World Wide Web. The user datagram protocol (UDP) also runs above the IP, but it provides a connectionless service. Since the processing involved in UDP is less, it is used for network management and real-time communication applications. The TCP/IP protocol stack presents problems when used in satellite networks because satellite networks have high propagation delay. The TCP layer has to be suitably modified for use in satellite networks.

Another innovative project is the Interplanetary Internet, which plans for interconnection of internets of different planets and spacecrafts. The results of this research will help improve the TCP/IP protocol stack performance in high-delay networks.

References

L.L. Peterson and B.S. Davie. Computer Networks: A Systems Approach. Morgan Kaufman Publishers Inc., CA, 2000. This book gives a systems approach, rather than a layered approach to computer networks.
A.S. Tanenbaum. Computer Networks. Prentice Hall, Inc., NJ, 1996. This book gives a layered approach to describe the computer networking protocols.
N. Ghani and S. Dixit. "TCP/IP Enhancements for Satellite Networks". IEEE Communications Magazine, Vol. 37, No. 1, July 1999.
http://www.ietf.org The Requests for Comments (RFCs) that give the complete details of the TCP/IP protocol stack can be obtained from this site. Each protocol specification gives the complete details of implementation.
http://www.ipnsig.org Web site of Interplanetary Internet.

Questions

Explain the functions of different layers in the TCP/IP protocol architecture.
Explain the operation of TCP and IP.
IP does not provide a reliable service, but TCP provides end-to-end reliable service. How?
What are the limitations of the TCP/IP protocol stack?
Differentiate between TCP and UDP.
List the problems associated with running the TCP/IP protocol stack in a satellite network.
Explain how congestion is controlled in TCP/IP networks.

Exercises

1.	Write a technical report on Interplanetary Internet.
2.	Prepare a technical report on running the TCP/IP protocol stack on a satellite network.
3.	Two systems, A and B, are connected by a point-to-point link, but the communication is only from A to B. Work out a mechanism to transfer a file from A to B using UDP as the transport protocol.
4.	Discuss the benefits of using UDP for data applications if the transmission link is very reliable and if there is no congestion on the network.
5.	Compare the performance of stop-and-wait protocol and sliding window protocol in terms of delay and throughput.

Answers

1.	You can obtain the details of Interplanetary Internet at http://www.ipnsig.org.
2.	The TCP/IP protocol stack does not perform well on a satellite network because of the large propagation delay. There will be timeouts before the acknowledgement is received, so packets are retransmitted by the sender though the packets are received at the other end. This causes unnecessary traffic on the network. To overcome these problems, spoofing and link accelerators are used.
3.	When the communication is one-way only, the TCP protocol cannot be used at the transport layer because acknowledgements cannot be sent in the reverse direction. In such a case, the connectionless transport protocol UDP has to be used. To transfer a file, the file has to be divided into UDP datagrams and sent over the communication link. At the receiving end, the datagrams have to be assembled by the application layer protocol. It is possible that some of the datagrams are received with errors, but retransmission cannot be done because of lack of the reverse link. The receiver has to check every datagram for errors, and if there is an error even in a single packet, the whole file is discarded. The sender may send the file multiple times so that at least once all the datagrams are received without error.
4.	If the transmission medium is very reliable, the packets will be received correctly. Also, if there is no congestion, the packets are likely to be received without variable delay and in sequence. Hence, UDP provides a fast data transfer, and the transmission medium is utilized effectively.
5.	Stop-and-wait protocol is very inefficient because the communication channel bandwidth is not utilized well. After the first packet is sent, an acknowledgement has to be received, and then only the second packet can be sent. On the other hand, in sliding window protocol, a number of packets can be sent without waiting for acknowledgements. Hence, channel utilization is better if sliding window protocol is used.

Projects

Interconnect two PCs using an RS232 link. Simulate a high delay in the network, run the TCP/IP protocol stack, and observe the throughput on the LAN to study the impact of the delay on the TCP/IP stack.
Develop software for spoofing—when the acknowledgement receipt is delayed from the other machine, an acknowledgement can be locally generated to cheat the TCP layer.

< Day Day Up >