Method to control error floor in analog neural ldpc decoder

H - Electricity – 03 – M

Patent

Rate now

  [ 0.00 ] – not rated yet Voters 0   Comments 0

Details

H03M 13/11 (2006.01) G06N 3/02 (2006.01) H03M 13/19 (2006.01)

Patent

CA 2651256

Analog decoder for (16,4) Hamming code is simulated with neural design of the check nodes. Error distributions of 3D and 4D mapping for various check nodes based on multilayer perceptron architecture is obtained in simulation after supervised training of neural mappers. For random code-words performance of neural decoder is found to be suboptimal to atanh(.) based ideal decoder. Error floor is shown to be a function of the error distribution in the check nodes. Optimal distribution is achieved when error for training points and those most remote from training points are close to one another with overall precision been about twice better. Control of the error floor via optimization of the error distribution in the check node's mappers is demonstrated, improving the achieved error floor for neural analog type decoder down to BER values of about 10 -6. Over-enhanced error distribution (with precision for intermediate points higher than that for training ones) exhibit deterioration in the performance, demonstrating error floor of about 10 -4 BER.

LandOfFree

Say what you really think

Search LandOfFree.com for Canadian inventors and patents. Rate them and share your experience with other people.

Rating

Method to control error floor in analog neural ldpc decoder does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Method to control error floor in analog neural ldpc decoder, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method to control error floor in analog neural ldpc decoder will most certainly appreciate the feedback.

Rate now

     

Profile ID: LFCA-PAI-O-1377208

  Search
All data on this website is collected from public sources. Our data reflects the most accurate information available at the time of publication.