IP in Space FAQ ------------------------------------------------------------------------ [Home] ------------------------------------------------------------------------ This FAQ is under construction. Last update on 24-Jul-2000. Please check back often! A text only version of this FAQ is also available for download. 1. General 1. Define what you mean by "IP". 2. What do you mean by "layers"? 3. Doesn't IP fail in space due to long propagation delays? 4. Isn't IP designed for clean terrestrial links instead of noisy, intermittant space links? 5. What is required to adapt a ground station for IP to a spacecraft? 6. Isn't the Space Communication Protocol Standards (SCPS) suite of protocols designed to overcome the shortcomings of IP used in space networks? ------------------------------------------------------------------------ 1. General 1. Define what you mean by "IP". "IP" stands for Internet Protocol. Strictly speaking, this is a layer 3 (networking) protocol that is only concerned with end-to-end network addressing. However, when speaking generally, the term "IP" is oftern used to refer to the entire suite of Internet protocols, including TCP, UDP, FTP, NTP, SMTP, etc., and not just the network layer protocol. 2. What do you mean by "layers"? "Layers" refers to a set of Open Systems Interconnect (OSI) network services that was originally defined by ISO in the mid 1980's. Each layer has a well defined subset of services it provides to the layer above, and a well defined interface to the layers above and below. The OSI model had 7 layers, but as the Internet evolved, layers 5 and 6 were rarely used. The layers are: Layer 7: Application Layer High-level protocols that use lower layer services. (FTP, Telnet, SMTP, NFS, DNS, SNMP) Layer 6: Presentation Layer - not generally used Layer 5: Session Layer - not generally used Layer 4: Transport Layer End-to-end stream multiplexing Protocols. (TCP - reliable byte-stream delivery, UDP - unreliable packet delivery) Layer 3: Network Layer End-to-end addressing, packet routing, simple packet forwarding. (IP, RIP, OSPF, IGRP, BGP) Layer 2: Data Link Layer Data unit delivery across a single data link, data unit framing, link level addressing. (Ethernet, HDLC, ATM, PPP, SLIP) Layer 1: Physical Layer Physical media and signal levels. coding specifications, bit-level operations. (10baseT, RS-422, UTP, Reed-Solomon) 3. Doesn't IP fail in space due to long propagation delays? This is a misconception that is brought about by confusing IP, a layer 3 network protocol, with TCP, a layer 4 transport protocol. IP is strictly an end-to-end addressing mechanism. IP packet headers have a source address, an destination address, and a few bytes of options. These get slapped on a packet which is then sent on it's merry way and forgotten about. Network layer protocols do not perform handshaking or retransmission. These functions get performed at a higher layer. As a result, IP is insensitive to transmission delay. It will work equally well across the room or out to Pluto. It will even work over uindirectional links. UDP, a layer 4 transport protocol, will also work over unidirectional links, allowing applications to be built that are not sensitive to delay. TCP, on the other hand, requires a bidirectional link to perform handshaking. The round-trip-time and the TCP "window" buffer size combine to restrict the maximum throughput possible over a TCP connection. TCP experiments have been done with the ACTS satellite in geosynchronous orbit that achieved a throughput of 540 Mbits/sec. It appears that TCP is quite useable in a planetary neighborhood, out to about lunar distance. Beyond that, UDP-based applications are called for. 4. Isn't IP designed for clean terrestrial links instead of noisy, intermittant space links? This is another misconception. One of the commonest physical links for the terrestrial Internet, the telephone line, has a bit error rate (BER) than is similar to most spacecraft RF links! The CCITT recommendations for voice circuits that have been conditioned to carry data is a BER of 10-5 (Source: ITT Reference Data for Radio Engineers). On the other hand, NASA typically specs it's spacecraft RF links at a BER of 10-5! The reason IP works over the phone lines is that the modem applies error correction down at the physical layer, transparently to the upper layers. This allows the upper layers to behave as though they have a clean link. Similarly, NASA applies error correction to it's space links, achieving BERs down to 10-7 or better. At 10-7, handshaking protocols, such as TCP/IP, work well. 5. What is required to adapt a ground station for IP to a spacecraft? Connect the clock and data lines from the groundstation RF system to a synchronous serial WAN port on a router. 6. Isn't the Space Communication Protocol Standards (SCPS) suite of protocols designed to overcome the shortcomings of IP used in space networks? At the time the SCPS Rationale, Requirements, and Application Notes document (CCSDS SCPS-710-0-G-0.4) was generated, that was true; however, in the two years since then, while ongoing industry research has solved, implemented, standardized, and deployed IP solutions to many of these problems, and are continuing to do so. For example, IP now supports constrained bandwidth channels via the header compression standard defined in RFC-2507, which was SPECIFICALLY designed for channels with non-trival packet loss rates. And asymetric channels are better supported by Selective Acknowledgement (SACK) RFC-2018, which is now deployed standard with most operating systems. Selectable routing, such as flood routing, and managed connection operation is being handled under the evolving IETF draft for the Dynamic Source Routing (DSR) protocol, and congestion signaling is being handled by Explicit Congestion Notification (ECN) and Random Early Detection (RED) spelled out in RFC-2481. ----------------------------------------------------------------------- ----------------------------------------------------------------------- [Home] ----------------------------------------------------------------------- Responsible NASA Official: Jim Rash Technical Contact: Keith Hogie, Computer Sciences [Image] Corp. [Image] NASA Privacy Statement Last Modified: Monday, 24-Jul-2000 10:25 EDT