Envelope-Based Modulation Classification Without IQ Demodulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional automatic modulation classification (AMC) methods relying on IQ data are vulnerable to imperfections such as amplitude or phase imbalance, especially in the presence of high-power interferers, and fail to identify symbol rates, limiting their utility in dynamic spectrum environments.
Innovation Solution
A deep learning-based AMC system that utilizes the envelope amplitude and frequency of radio frequency signals, detected in the RF domain without downconversion, employing a feature extraction circuit and a deep learning neural network, such as an LSTM, to classify modulation types and symbol rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional IQ data-based AMC methods are used, then the system can classify modulation types, but the system becomes vulnerable to amplitude or phase imbalance and high-power interferers
Solution Approach 1:
The patent extracts and uses only the envelope amplitude and frequency components of the RF signal, discarding the IQ data that is vulnerable to interference. By taking out only the essential features (envelope and frequency) that remain robust under interference, the system achieves reliability without being affected by amplitude or phase imbalance.
Solution Approach 2:
The patent introduces an envelope detector as an intermediary component that processes the RF signal before classification. This intermediary extracts the envelope amplitude, which serves as a robust feature that is not affected by the harmful factors (interferers, imbalance) that directly impact traditional IQ-based methods.
2Measurement precision
If downconversion is performed to extract IQ data, then modulation classification can be performed, but the system loses robustness against high-power interferers
Solution Approach 1:
Instead of the traditional approach of downconverting RF to baseband IQ data for analysis, the patent inverts the approach by staying in the RF domain and extracting envelope amplitude and frequency directly. This inversion eliminates the vulnerability introduced by downconversion while maintaining detection accuracy through robust envelope-based features.
3Loss of information
If traditional AMC methods focus only on modulation identification, then the classification process is simple, but the system fails to quantify symbol rate
Solution Approach 1:
The patent makes the envelope-based classification system universal by enabling it to perform multiple functions: both modulation type identification and symbol rate quantification. By using the same envelope amplitude and frequency features for both tasks, the system gains multi-functionality without proportionally increasing complexity.
Solution Approach 2:
The patent adds symbol rate quantification as an additional dimension to the traditional modulation classification output. By incorporating symbol rate information alongside modulation type identification using the same envelope features, the system expands its capability without requiring entirely separate processing paths.
Data Source
AI summary
Method and system for automatic modulation classification (AMC) use time-series voltage signals representative of radio frequency (RF) signal envelope and frequency components as input features to a deep learning-based neural network, which enables classification of both modulation type and symbol rate without requiring in-phase and quadrature (IQ) demodulation. A feature extraction circuit captures RF signal envelope amplitude and frequency using stub-based sensing, and a Long Short-Term Memory (LSTM) neural network processes these features in a digitized form to classify the modulation and symbol rate with high accuracy and minimal latency.