Channelized RF Receiver Using Compressive Sensing for Emitter Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional channelized receivers and compressive sensing systems face challenges with signal-to-noise degradation and complex processing requirements due to noise folding and pulse-on-pulse interference, especially in high-bandwidth RF environments, which limits their effectiveness in detecting and identifying RF emitters efficiently.
Innovation Solution
A smart receiver integrating compressive sensing and machine learning capabilities, which channelizes RF signals, encodes them using a Compressive Multiplexor technique, and employs machine learning to characterize and identify RF emitters, reducing computational complexity and noise folding by preserving signal structure and using deep learning for efficient pulse detection and classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If conventional compressive sensing is used to detect sparse signals across wide band, then signal detection capability is improved, but signal-to-noise ratio deteriorates due to noise folding
Solution Approach 1:
The receiver divides the wideband signal into multiple channelized sub-bands before compressive sensing encoding. Each channel processes a narrower bandwidth segment, reducing noise folding within each channel while maintaining overall wideband detection capability. The channelized structure allows parallel processing of segmented signal portions.
Solution Approach 2:
The system performs channelization and frequency segmentation before the compressive sensing encoding stage. This preliminary organization of the wideband signal into structured channels prepares the signal for more effective compressive sensing by reducing the noise folding problem that would occur with direct wideband compression.
2Difficulty of detecting and measuring
If conventional compressive sensing decoding is performed to reconstruct folded signals, then signal separation is achieved, but computational complexity increases
Solution Approach 1:
The decoding process is divided into channel-specific operations rather than a single complex wideband reconstruction. Each channel performs its own decoding on segmented data, reducing the overall computational burden compared to reconstructing the entire folded wideband signal in one operation.
Solution Approach 2:
The system performs decoding only on channels where signals are detected rather than processing all channels uniformly. This partial action approach reduces computational complexity by focusing processing resources only on relevant signal segments.
3Measurement precision
If full Nyquist rate sampling is used to capture wideband signal, then signal fidelity is improved, but data rate and processing requirements increase
Solution Approach 1:
The wideband signal is divided into multiple channelized sub-bands, each processed at a lower sampling rate appropriate to its narrower bandwidth. This segmentation allows the system to capture the full wideband spectrum with high fidelity while reducing the overall data rate by avoiding Nyquist-rate sampling of the entire bandwidth simultaneously.
Solution Approach 2:
Each channel is allocated sampling resources proportional to its bandwidth requirements rather than uniformly across the entire wideband spectrum. This local quality approach ensures high signal fidelity in each channel while optimizing the overall data rate by matching sampling rates to local bandwidth needs.
Data Source
AI summary
System and method for identifying an RF emitter include: channelizers for channelizing RF signals into several channels; a compressive sensing (CS) encoder for each channel to CS encode the channelized signal to produce an encoded channelized signal in each of the plurality of channels; a summer to sum the encoded channelized signals of all of the plurality of channels to produce an I/Q data; a channelized pulse detection circuit to detect pulses in each channel and produce encoded pulse snippets from the I/Q data; a CS decoder for each channel to CS decode the encoded pulse snippets; a first machine learning device to characterize the decoded pulse snippets and to produce pulse description words (PDWs); and a second machine learning device to associate the PDWs with one or more RF emitters and identify the one or more RF emitters.