Automatic Modulation Classification: Principles, Algorithms and Applications

Automatic Modulation Classification (AMC) has been a key technology in many military, security, and civilian telecommunication applications for decades. In military and security applications, modulation often serves as another level of encryption; in modern civilian applications, multiple modulation types can be employed by a signal transmitter to control the data rate and link reliability.

This book offers comprehensive documentation of AMC models, algorithms and implementations for successful modulation recognition. It provides an invaluable theoretical and numerical comparison of AMC algorithms, as well as guidance on state-of-the-art classification designs with specific military and civilian applications in mind.
 

Key Features:

Provides an important collection of AMC algorithms in five major categories, from likelihood-based classifiers and distribution-test-based classifiers to feature-based classifiers, machine learning assisted classifiers and blind modulation classifiers Lists detailed implementation for each algorithm based on a unified theoretical background and a comprehensive theoretical and numerical performance comparison Gives clear guidance for the design of specific automatic modulation classifiers for different practical applications in both civilian and military communication systems Includes a MATLAB toolbox on a companion website offering the implementation of a selection of methods discussed in the book

1. Introduction

1.1. Background

1.2. Applications of AMC

1.2.1. Military Applications

1.2.2. Civilian Applications

1.3. Field Overview and Book Scope

1.4.Modulation and Communication System Basics

1.4.1. Analogue Systems and Modulations

1.4.2. Digital Systems and Modulations

1.4.3. Received Signal with Channel Effects

1.5. Conclusion

2. Signal Models for Modulation Classification

2.1. Introduction

2.2. Signal Model in AWGN Channel

2.2.1. Signal Distribution of I-Q segments

2.2.2. Signal Distribution of Signal Phase

2.2.3. Signal Distribution of Signal Magnitude

2.3. Signal Models in Fading Channel

2.4. Signal Models in Non-Gaussian Channel

2.4.1. Middleton’s Class A Model

2.4.2. Symmetric Alpha Stable Model

2.4.3. Gaussian Mixture Model

2.5.Conclusion

3. Likelihood Based Classifiers

3.1. Introduction

3.2. Maximum Likelihood Classifiers

3.2.1. Likelihood Function in AWGN channels

3.2.2. Likelihood Function in fading channels

3.2.3. Likelihood Function in non-Gaussian noise channels

3.2.4. Maximum Likelihood Classification Decision Making

3.3. Likelihood Ratio Test for Unknown Channel Parameters

3.3.1. Average Likelihood Ratio Test

3.3.2. Generalized Likelihood Ratio Test

3.3.3. Hybrid Likelihood Ratio Test

3.4. Complexity reduction

3.4.1. Discrete Likelihood Ratio Test and Lookup Table

3.4.2. Minimum Distance Likelihood Function

3.4.3. Non-parametric Likelihood Function

3.5. Conclusion

4. Distribution Test Based Classifier

4.1. Introduction

4.2. Kolmogorov-Smirnov (KS) Test Classifier

4.2.1. The KS test for goodness-of-fit

4.2.2. One sample KS test classifier

4.2.3. Two sample KS test classifier

4.2.4. Phase Difference Classifier

4.3. Cramer-von Mises Test Classifier

4.4. Anderson-Darling Test Classifier

4.5. Optimized Distribution Sampling Test Classifier

4.5.1. Sampling Location Optimization

4.5.2. Distribution sampling

4.5.3. Classification Decision Metrics

4.5.4. Modulation Classification Decision Making

4.6. Conclusion

5. Modulation Classification Features

5.1. Introduction

5.2. Signal Spectral Based Features

5.2.1. Signal Spectral Based Features

5.2.2. Spectral Based Features Specialties

5.2.3. Spectral Based Features Decision Making

5.2.4. Decision Threshold Optimization

5.3. Wavelet Transform Based Features

5.4. High-order Statistics Based Features

5.4.1. High-order Moment Based Features

5.4.2. High-order Cumulant Based Features

5.5. Cyclostationary Analysis Based Features

5.6.Conclusion

6. Machine Learning for Modulation Classification

6.1. Introduction

6.2. K-nearest Neighbour Classifier

6.2.1. Reference Feature Space

6.2.2. Distance Definition

6.2.3. K-nearest Neighbour Decision

6.3. Support Vector Machine Classifier

6.4. Logistic Regression for Feature Combination

6.5. Artificial Neural Network for Feature Combination

6.6. Genetic Algorithm for Feature Selection

6.7. Genetic Programming for Feature Selection and Combination

6.7.1. Tree Structured Solution

6.7.2. Genetic Operators

6.7.3. Fitness Evaluation

6.8.Conclusion

7. Blind Modulation Classification

7.1. Introduction

7.2. Expectation Maximization with Likelihood Based Classifier

7.2.1. Expectation Maximization Estimator

7.2.2. Maximum Likelihood Classifier

7.2.3. Minimum Likelihood Distance Classifier

7.3. Minimum Distance Centroid Estimation and Non-parametric Likelihood Classifier

7.3.1. Minimum Distance Centroid Estimation

7.3.2. Non-parametric Likelihood Function

7.4.Conclusion

8. Comparison of Modulation Classifiers

8.1. Introduction

8.2. System Requirements and Applicable Modulations

8.3. Classification Accuracy With Additive Noise

8.3.1. Benchmarking Classifiers

8.3.2. Performance Comparison in AWGN Channel

8.4. Classification Accuracy With  Limited Signal Length

8.5. Classification Robustness Against Phase Offset

8.6. Classification Robustness Against Frequency Offset

8.7. Computational Complexity

8.8. Conclusion

9. Modulation Classification for Civilian Applications

9.1. Introduction

9.2. Modulation Classification for High Order Modulations

9.3. Modulation Classification for Link Adaptation Systems

9.4. Modulation Classification for MIMO Systems

9.5. Conclusion

10. Modulation Classifier Design for Military Applications

10.1. Introduction

10.2. Modulation Classifier with Unknown Modulation Pool

10.3. Modulation Classifier against Low Probability of Detection

10.3.1. Classification of DSSS signal

10.3.2. Classification of FHSS signal

10.4. Conclusion