Modulation Classification via Signal Graphs for SISO MIMO Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current modulation classification methods for radio communication signals in multipath fading channels are complex, sensitive to model mismatches, and not applicable to MIMO systems, requiring improvements for efficient classification in both SISO and MIMO systems.
Innovation Solution
A method and system that extract modulation classification features from a graph representation of the Fourier transform of signal samples, constructing undirected graphs based on amplitude transitions and using binomial distribution probabilities to classify modulation formats, applicable to both SISO and MIMO systems with reduced computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If likelihood-based algorithms are used for modulation classification, then classification performance is improved, but implementational complexity and sensitivity to model mismatches increase
Solution Approach 1:
The patent extracts only the essential spectral features (peak frequencies and amplitudes from FFT) needed for classification, discarding redundant signal information. This extraction approach achieves effective classification while avoiding the computational burden of processing entire signal models, thus resolving the contradiction between performance and complexity.
Solution Approach 2:
The patent uses simple, computationally inexpensive features (frequency bins and amplitudes) that can be quickly computed and discarded after classification, replacing complex likelihood functions. These simple features provide sufficient classification accuracy without the high computational cost of full likelihood-based methods.
2Ease of operation
If feature-based methods are used for modulation classification, then ease of implementation and robustness are improved, but classification performance decreases
Solution Approach 1:
The patent changes the parameter representation from time-domain or complex-constellation features to spectral domain features (FFT magnitudes at specific bins). This parameter transformation enables simple implementation through basic spectral analysis while achieving high classification performance by focusing on the most discriminative frequency components.
Solution Approach 2:
The patent transitions from analyzing signals in the time domain or I-Q plane to the frequency domain, where modulation characteristics manifest as distinct spectral patterns. This dimensional change simplifies feature extraction while enhancing classification performance through spectral fingerprinting.
3Adaptability or versatility
If prior modulation classification techniques are applied to MIMO systems, then adaptability is improved, but computational complexity and sensitivity to frequency offset increase
Solution Approach 1:
The patent segments the MIMO received signal into multiple independent streams (one per receive antenna), applies FFT and spectral analysis to each stream separately, and combines the results. This segmentation approach enables MIMO classification with the same simple per-stream processing used in SISO, avoiding exponential complexity growth while maintaining adaptability to multiple antennas.
4Measurement precision
If long observation intervals are used for modulation classification, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The patent uses only a partial observation interval (one symbol period or fraction thereof) focused on the most informative spectral components, rather than requiring long observation intervals. By concentrating on critical frequency bins where modulation signatures appear, sufficient classification accuracy is achieved with minimal time investment.
Data Source
AI summary
This invention relates system for classifying a modulation format of a communication signal. The system includes a receiving antenna system, to receive the communication signal, a preprocessing module, to evaluate a Fourier transform of a plurality of samples of the received communication signal, normalize each discrete sample of the Fourier transform, quantize each normalized sample based on a number of quantization levels (Q), a graphical analyzer to construct a undirected graph by tracing amplitude of the each quantized sample; and a classification module to extract one or more modulation classification (MC) features from the undirected graph; and to determine the modulation format of the received communication signal based on the extracted MC features.