FMCW Radar Interference Suppression via Neural Network Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Radar sensors in automotive applications face interference issues due to overlapping radar signals from nearby vehicles, which impair their operation and accuracy in detecting objects and maintaining safe distances.
Innovation Solution
A method using a convolutional neural network to filter out interfering signals from received radar signals by generating datasets associated with frequency-modulated chirps and processing them to reduce interference, allowing for effective signal suppression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If radar sensors are deployed in multiple vehicles, then coverage and detection capability are improved, but interference from other radar signals increases
Solution Approach 1:
The patent applies neural network-based signal processing to convert the harmful interference signal into useful information. The neural network is trained to recognize patterns in interference signals and distinguish them from legitimate radar echoes, thereby transforming the harmful RF interference into a learnable pattern that can be filtered or utilized for interference awareness.
Solution Approach 2:
The patent introduces an intermediate processing stage between signal reception and target detection. A neural network acts as an intermediary that processes the raw radar signals, identifying and separating interference components from valid target echoes before final detection, thus mediating between the conflicting signals and the detection system.
2Measurement precision
If neural network filtering is applied to reduce interference, then signal accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent performs preliminary training of the neural network offline using labeled datasets containing both interference and valid radar signals. This preliminary action prepares the model in advance, allowing it to make rapid decisions during real-time operation without requiring complex runtime computations, thus reducing online processing complexity while maintaining high accuracy.
Solution Approach 2:
The patent replaces traditional mechanical or algorithmic signal filtering methods with a data-driven neural network approach. Instead of using fixed filtering algorithms or hardware-based interference rejection, the system uses a trained neural network that learns optimal filtering strategies from data, substituting complex mechanical processing with intelligent pattern recognition.
Data Source
AI summary
A radar device may include a radar receiver to receive a radio frequency (RF) radar signal and generate a digital signal based on the RF radar signal. The digital signal may comprise a plurality of signal segments. The radar device may include a neural network comprising a plurality of layers to process the plurality of signal segments. Each layer of the plurality of layers may have one or more neurons. The plurality of layers may process the plurality of signal segments using weighting factors having values selected from a predetermined set of discrete values. At least one neuron in an output layer of the plurality of layers may provide an output value that indicates whether a respective signal segment or a sample, associated with the at least one neuron, is overlaid with an interfering signal.