Neural Network Pulse Parameter Detection in Low SNR Radar
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional radar and communications systems face challenges in accurately detecting RF pulses in high noise or complex noise environments, particularly when the signal-to-noise ratio (SNR) is low, leading to reduced accuracy and reliability.
Innovation Solution
Training a neural network, such as a Pulse Parameter Estimation Neural Network (PPENN), to detect multiple pulse parameters within a noisy RF signal, which can be further enhanced by pre-filtering using a matched filtering system or a second neural network to remove noise and channel effects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional filtering and thresholding techniques are used for pulse detection, then the system is simple to implement, but the detection accuracy deteriorates in high noise or low SNR environments
Solution Approach 1:
The patent replaces conventional mechanical filtering and thresholding systems with a neural network-based detection system. The neural network learns optimal detection parameters and pulse characteristics from training data, enabling accurate pulse parameter estimation even in high noise environments where traditional methods fail. This substitution transforms the detection approach from rule-based to learning-based, significantly improving measurement precision at the cost of increased computational complexity.
2Reliability
If thresholding is used to ignore signals below a threshold value, then simple noise rejection is achieved, but pulse detection fails when noise levels are as high as or higher than the pulse signal
Solution Approach 1:
The patent converts the harmful noise interference into a beneficial training opportunity. By including noisy signals with known pulse parameters in the training dataset, the neural network learns to recognize pulse patterns even when buried in noise. The system transforms the previously harmful high noise environments into valuable training conditions that enhance the detector's robustness and reliability against noise interference.
Solution Approach 2:
The patent applies preliminary action by training the neural network beforehand with extensive noisy data before actual pulse detection. The training phase pre-computes optimal detection strategies and adapts the network weights to handle various noise conditions. This preliminary training enables the system to reliably detect pulses in high noise environments without requiring complex real-time noise filtering or adaptive threshold adjustment during operation.
3Measurement precision
If matched filtering is used to account for noise, then some noise reduction is achieved, but the system becomes less effective when signal-to-noise ratio drops below 5 dB
Solution Approach 1:
The patent applies parameter changes by transforming the detection approach from fixed threshold-based methods to adaptive neural network-based methods. The neural network dynamically adjusts its internal parameters (weights and biases) based on the input signal characteristics and noise conditions. This parameter adaptability allows the system to maintain high measurement precision across a wide range of signal-to-noise ratios, including the challenging regime below 5 dB where conventional matched filtering becomes ineffective.
Data Source
AI summary
Techniques are described for a computing device to estimate pulse parameters. A method includes (a) receiving a digitized radio frequency (RF) signal as a digital input signal; (b) feeding the digital input signal into a plurality of input nodes of a trained Pulse Parameter Estimation Neural Network (PPENN), the PPENN having been trained using machine learning; (c) operating the trained PPENN to estimate a plurality of pulse parameters of a set of pulses embedded within a waveform of the digital input signal and to output the plurality of pulse parameters from the trained PPENN; and (d) processing the plurality of pulse parameters to further quantify the set of pulses. A system, apparatus, and computer program product for performing this method and similar methods are also described.