Browse Prior Art Database

Adaptive Cyclic Redundancy Check over 16-Bit Bytes: Method and Apparatus

IP.com Disclosure Number: IPCOM000111350D
Original Publication Date: 1994-Feb-01
Included in the Prior Art Database: 2005-Mar-26
Document File: 2 page(s) / 42K

Publishing Venue

IBM

Related People

Blaum, M: AUTHOR [+4]

Abstract

The data that is sent between nodes in a parallel/distributed machine is typically protected by an error detecting/correcting code. A common approach is to use a fixed block length and to add the redundancy bytes at the end of a block. In an asynchronous communication network the redundancy bytes are added at the end of a packet/message. However, the packet size varies and the challenge is to add a variable number of redundancy bytes to packets to achieve uniform reliability. It is natural that, for a smaller number of information bytes, we would like to add a smaller number of redundant bytes, since the probability of error is also smaller.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 72% of the total text.

Adaptive Cyclic Redundancy Check over 16-Bit Bytes:  Method and Apparatus

      The data that is sent between nodes in a parallel/distributed
machine is typically protected by an error detecting/correcting code.
A common approach is to use a fixed block length and to add the
redundancy bytes at the end of a block.  In an asynchronous
communication network the redundancy bytes are added at the end of a
packet/message.  However, the packet size varies and the challenge is
to add a variable number of redundancy bytes to packets to achieve
uniform reliability.  It is natural that, for a smaller number of
information bytes, we would like to add a smaller number of redundant
bytes, since the probability of error is also smaller.

      One possible solution to this problem is to use different codes
in parallel, each one with its own redundancy.  However, this would
complicate the hardware.  Our idea is to use only a single circuit to
encode and to decode, avoiding hardware duplication.  Let L be the
maximum packet size and let M be the number of redundancy bytes we
would like to add to a packet.  For every packet of size $ 0 < l =<
L, we have a predetermined parameter  m_l  that specifies the number
of redundancy bytes that we would like to add to the packet.

      At the encoder: for every packet of length l compute M CRC
bytes by using a standard CRC encoder which is based on a linear
feedback shift register (LFSR) that operates in real time.  Select
the...