Radar Signal Noise Separation via Sparse Recovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Low-frequency ultra-wideband radar systems face significant challenges in detecting targets due to corruption by radio frequency interference (RFI) signals, which result in a severely reduced signal-to-noise ratio and ineffective target detection, especially when the interference is dynamic and unpredictable, and existing techniques like notch filtering lead to side lobes and reduced target amplitudes.
Innovation Solution
A system and method that jointly estimate and extract interference noise components in the time domain using a sparse recovery framework, modeling RFI as a sparse linear combination of cosine and sine waves, without requiring prior knowledge of interference sources, and separates noise from radar signals using low-rank and sparse recovery optimization, allowing for adaptive processing and accurate interference removal.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If notch filtering is used to suppress interference noise, then the noise suppression effect is improved, but side lobes appear and target amplitudes are reduced
Solution Approach 1:
The patent extracts interference noise components from the received signal by modeling them as a sparse linear combination of cosine and sine waves. The processor identifies and separates these noise components in the time domain, then subtracts them from the original signal to obtain a cleaned radar signal, avoiding the need for frequency-domain notch filtering that causes side lobes and amplitude reduction.
Solution Approach 2:
The patent introduces an intermediate processing step where the received signal is first modeled as a sum of radar signal and sparse interference components. By using a sparse recovery framework with cosine and sine wave bases, the interference is extracted as an intermediate result before final signal reconstruction, preserving target amplitude information while removing noise.
2Device complexity
If frequency domain spike detection is used to identify corrupted bands, then interference detection is simplified, but severe sidelobes appear in the time domain output
Solution Approach 1:
The patent replaces the mechanical approach of zeroing out frequency bins (which causes sharp transitions and side lobes) with a time-domain sparse recovery method. Instead of manipulating frequency domain data with hard cutoffs, the system uses a linear algebraic approach with cosine and sine wave bases to smoothly extract interference components, avoiding the creation of time-domain artifacts.
3Ease of manufacture
If simple suppression of contaminated frequency bands is applied, then processing is simplified, but the signal-to-noise ratio is reduced
Solution Approach 1:
The patent changes the approach from frequency-domain suppression to time-domain extraction by transforming the problem parameters. Instead of suppressing entire frequency bands (which loses signal information), the system models interference as a sparse combination of sinusoidal components with specific frequencies, amplitudes, and phases. This parameter-based modeling allows selective removal of only the interference components while preserving the radar signal, maintaining SNR while achieving noise suppression.
Data Source
AI summary
A system for reception of electromagnetic waves in a spectrum in which interference with radio frequencies of other electronics devices occurs comprising a transmitter; at least one receiver configured to receive the received signal; each received signal organized into a digital vector; at least one memory portion configured to store a plurality of received signals in a vector form; the vectors being combined into a matrix, each vector of the matrix being a digital data record representing a received signal; at least one processor operatively connected to the at least one memory portion; the at least one processor configured to estimate that portion of the received signal attributable to noise; the at least one processor operating to jointly estimate a minimal number of distinctive noise patterns and minimize the simplicity of the data of interest; the at least processor operating to process the noise and data of interest separately.