AUTOMATIC MODULATION RECOGNITION BASED ON SPATIAL PATTERN CLASSIFICATION USING ARTIFICIAL NEURAL NETWORKS

PDF

Authors
  1. Ghani, N.
Corporate Authors
McMaster Univ, Hamilton ONT (CAN) Communications Research Lab;Defence Research Establishment Ottawa, Ottawa ONT (CAN)
Abstract
An automatic modulation recognition system has been designed and tested based on backpropagation neural networks. This system successfully distinguishes between the specified 10 signal classes to a very high degree and uses a Welch periodogram preprocessor. Experimental results show that the neural networks match and even outdo the performance of the conventional K-nearest neighbour classifier for the same preprocessor, upper 90's percentages. Significant optimization of the neural networks is also shown using the optimal brain damage pruning algorithm. Discusses the design issues, preprocessing techniques, backpropagation algorithms, implementation details, and the specifics of the software developed during the project.
Date of publication
28 Jul 1992
Number of Pages
198
DSTKIM No
93-02565
CANDIS No
128500
Format(s):
Hardcopy;Originator's fiche received by DSIS;Document Image stored on Optical Disk

Permanent link

Document 1 of 1

Date modified: