Neural Networks for Classification of Radar Signals
Abstract
Radar Electronic Support Measures (ESM) systems are faced with increasingly dense and complex electromagnetic environments. Traditional algorithms for signal recognition and analysis are highly complex, computationally intensive, often rely on heuristics, and require humans to verify and validate the analysis. In this paper, the use of an alternative technique - artificial neural networks - to classify pulse-to-pulse signal modulation patterns is investigated. Neural networks are an attractive alternative because of their potential to solve difficult classification problems more effectively and more quickly than conventional techniques. Neural networks adapt to a problem by learning, even in the presence of noise or distortion in the input data, without the requirement for human programming. In the paper, the fundamentals of network construction, training, behaviour and methods to improve the training process and enhance a network's performance are discussed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1993
- Accession Number
- ADA279930
Entities
People
- Christopher A. Carter
- Nathalie Masse
Organizations
- Defence Research and Development Canada