Error control codeword generating system and method based on a neural network

• US 5,293,453 A
• Filed: 02/13/1992
• Issued: 03/08/1994
• Est. Priority Date: 06/07/1990
• Status: Expired due to Term

• Current Assignee: Texas Instruments, Inc.
• Original Assignee: Texas Instruments, Inc.
• Inventors: Frazier, Gary A.
• Primary Examiner(s): Fleming, Michael R.
• Assistant Examiner(s): Downs, Robert W.
• Application Number:
  US07/836,186
• Time in Patent Office: 754 Days
• Field of Search: 395/21, 395/22
• US Class Current: 706/17
• CPC Class Codes: G06F 11/08 Error detection or correction by redundancy in data representation, e.g. by using checking codes
• CPC Class Codes: H03M 13/6597 Implementations using analogue techniques for coding or decoding, e.g. analogue Viterbi decoder
• CPC Class Codes: H04L 1/0045 Arrangements at the receiver end
• CPC Class Codes: H04L 25/03165 using neural networks

Abstract

A communication system and method that translates a first plurality of information symbols into a plurality of code words, transmits the plurality of code words through a communication channel receives the plurality of code words transmitted through the communication channel, deciphers the plurality of code words transmitted through the communication channel into a second plurality of information symbols that correspond to the first set plurality of information symbols, wherein the plurality of code words are derived from a reverse dynamical flow within a first neural network.

Claims

1. A method for generating a sequence of K N-bit binary code words for information transmission encoding and decoding, comprising the steps of:

(a) providing a neural network with N neurons, each neuron has two states with output values 0 and 1, interconnection strengths T_ij connecting the output of the jth neuron to an input of the ith neuron, and dynamical evolution given by

space="preserve" listing-type="equation">V.sub.i →

1 if Σ

.sub.j T.sub.ij V.sub.j <

θ

.sub.i and

space="preserve" listing-type="equation">V.sub.i →

0 if Σ

.sub.j T.sub.ij V.sub.j >

θ

.sub.i

where V_i is the output value of the ith neuron, V_j is the output value of the jth neuron and θ

is a threshold;

(b) providing an initial N-bit binary code word U¹ with components U¹_j ;

(c) setting said interconnection strengths T_ij =U¹_i U¹_j for all i and j except setting T_ii =0;

(d) inputting U¹ into said neural network, and letting said neural network evolve to a stable state V;

(e) taking said stable state V to define a new binary code word U² ;

(f) updating said interconnection strengths by adding U²_i U²_j to the current T_ij for all i and j except i=j;

(g) repeating steps (d)-(f) with U^k replaced by U^k+1 for k=1,2, . . . , K-1; and

(h) applying said code words U¹, U², . . . U^k to establish an information transmission encoder and decoder.

2. The method of claim 1, wherein:

(a) said providing an initial N-bit binary code word U¹ is by random selection.

3. A method for generating a sequence of K N-bit binary code words for information transmission encoding and decoding, comprising the steps of:

(a) providing a Hopfield neural network with N neurons but with dynamical evolution away from stored states defining the interconnection strengths;

(b) providing an initial N-bit binary code word U¹ ;

(c) setting the interconnection strengths of said neural network T_ij =U¹_i U¹_j for all i and j except setting T_ii =0;

(d) inputting U¹ into said neural network, and letting said neural network evolve to a stable state V;

(e) taking said stable state V to define a new binary code word U² ;

(f) updating said interconnection strengths by adding U²_j U²_j to the current T_ij for all i and j except i=j;

(g) repeating steps (d)-(f) with U^k replaced by U^k+1 for k=1, 2, . . . , K-1; and

(h) applying said code words U¹, U², . . . U^k to establish an information transmission encoder and decoder.

4. The method of claim 3, wherein:

(a) said providing an initial N-bit binary code word U¹ is by random selection.