Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Error Control for Fast Packet Networking

IP.com Disclosure Number: IPCOM000039568D
Original Publication Date: 1987-Jul-01
Included in the Prior Art Database: 2005-Feb-01
Document File: 3 page(s) / 47K

Publishing Venue

IBM

Related People

Lien, YC: AUTHOR [+4]

Abstract

A data communication network usually comprises a multiplicity of switching nodes and links interconnecting pairs of these nodes. One of the purposes of the network is to transport packets of data from a source node to a destination node, typically visiting one or several intermediate nodes along the path. Each link can span considerable distances, using a variety of available transmission and switching facilities, such as those provided by the telephone company. In transit, one or several bits of a packet can get corrupted, giving rise to transmission errors. Error control is one of the functions of the data link control component of a data communication network. It has two aspects: error detection and error correction.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 51% of the total text.

Page 1 of 3

Error Control for Fast Packet Networking

A data communication network usually comprises a multiplicity of switching nodes and links interconnecting pairs of these nodes. One of the purposes of the network is to transport packets of data from a source node to a destination node, typically visiting one or several intermediate nodes along the path. Each link can span considerable distances, using a variety of available transmission and switching facilities, such as those provided by the telephone company. In transit, one or several bits of a packet can get corrupted, giving rise to transmission errors. Error control is one of the functions of the data link control component of a data communication network. It has two aspects: error detection and error correction. A frequently used technique for error detection encodes the information field conveyed by a packet into a 2-byte Cyclic Redundancy Check (CRC) that is transmitted as part of each packet. The receiving station re- computes the CRC upon arrival of the packet. If the received CRC and the recomputed CRC disagree, it is highly probable that a transmission error has occurred (Fig. 1). Error correction is effected by returning a negative acknowledgement (NAK) to the transmitting station, causing it to repeat the transmission. If no error is detected, an affirmative acknowledgement (ACK) is returned, enabling the transmitting station to release the buffer that stored the packet for possible retransmission. In case the corruption occurs in the packet header, the system usually relies on higher level protocol, such as time-out and using retransmission mechanism. Typically, the path taken by a packet through the network comprises several links. It is well known that link-level error control achieves highly reliable packet transport in the presence of moderate link error rates. In such an environment link-level error control is also efficient, since it avoids forwarding corrupted packets. On the other hand, link-level error control exacts a price in two respects. Firstly, it requires additional buffer space, since every packet must be buffered and held in every node along its path until an ACK confirms correct arrival at the next node; secondly, it causes additional delays, since forwarding of a packet cannot begin before the entire packet has been received, because only then is it known whether the CRC check has been successful. In case the links have low error rates, these cumulative buffering delays slow down packet transmission unnecessarily, since retransmissions are rare. In this case, end-to-end error control is preferable, where the CRC is checked only at the destination node, and retransmission of a corrupted packet retraces the entire path through the network. End-to-end error control, given low link error rates, thus permit a considerable reduction in transport delay and improved throughput (since packets can be forwarded as they arrive), and a reduction in buffer storage req...