Browse Prior Art Database

Optimum Decoder

IP.com Disclosure Number: IPCOM000096360D
Original Publication Date: 1963-Apr-01
Included in the Prior Art Database: 2005-Mar-07
Document File: 2 page(s) / 55K

Publishing Venue

IBM

Related People

Brown, DT: AUTHOR [+2]

Abstract

This is a decoder of binary information received over a channel perturbed by independent noise. The decoder stores analog values representing all the binits of a received word. The word is decoded as a whole. The decoder avoids the use of a fixed threshold level for determining binit values, of 1 and 0, which arbitrarily replaces and destroys the received information. Decoding is accomplished in a matrix network. This is a summing network, where one of g (the number of legal words) outputs attains a higher voltage level than all the others in the word set.

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 75% of the total text.

Page 1 of 2

Optimum Decoder

This is a decoder of binary information received over a channel perturbed by independent noise. The decoder stores analog values representing all the binits of a received word. The word is decoded as a whole. The decoder avoids the use of a fixed threshold level for determining binit values, of 1 and 0, which arbitrarily replaces and destroys the received information. Decoding is accomplished in a matrix network. This is a summing network, where one of g (the number of legal words) outputs attains a higher voltage level than all the others in the word set.

Initially, the message is received from the channel and demodulated at 21. The binary information Y is the input to network 22. Y is the transmitted signal X + noise N. The non-linear network 22 can be a diode network, while not a threshold device, does provide levels or a continuous variable output determined by the calculated probability that Y equals 1 or 0. The parameters which are used to determine this probability are the nominal signal level, the noise characteristic in the channel and the distribution of 0's and 1's. The graphic example at the left illustrates this. If the distribution of 1's were increased relative to 0's, the function of the network could be changed in the direction of the dashed line in the output curve.

In the log network 23 the probabilities are converted into logarithms so that sums are calculated instead of products. When a binit is 1, the Log Z output is at a hig...