Pulsed Radar Motion Classification Using Hidden Markov Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Pulsed radars are limited in classifying the type of motion of a target, such as human activities like running, walking, or creeping, and existing methods require high computational power or constant transmission, making them inefficient for real-time classification and noise resistance.
Innovation Solution
The method employs Hidden Markov Models to create motion-specific models, generates feature vector time series from pulsed radar signals, and uses principal component analysis to reduce computational burden and noise, enabling rapid and accurate classification of target motion by comparing the time series to a trained database.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Fourier-based feature vectors are used for target classification, then classification accuracy is improved, but computational complexity and processing power requirements increase significantly
Solution Approach 1:
The patent extracts only the most relevant features from the radar signal using Principal Component Analysis (PCA), selecting top components that capture the essential motion characteristics while discarding redundant information. This extraction approach maintains classification accuracy by preserving key motion patterns while significantly reducing the dimensionality and computational burden of feature vectors.
Solution Approach 2:
The patent employs simplified feature representations that are computationally inexpensive to calculate and process. By using PCA-transformed features with a limited number of components rather than full Fourier-based feature vectors, the system achieves effective classification with reduced computational resources, making the solution more suitable for real-time radar applications.
2Reliability
If constant transmission is used for target identification, then signal strength is improved, but energy consumption increases and noise resistance decreases
Solution Approach 1:
The patent employs pulsed radar transmission instead of continuous wave transmission, sending radar signals in periodic pulses rather than constantly. This periodic action allows the system to maintain sufficient signal strength for target detection and classification while significantly reducing energy consumption compared to constant transmission, and also provides better noise resistance by allowing integration over multiple pulses.
3Difficulty of detecting and measuring
If classical radar techniques are used for target detection, then detection capability is improved, but motion classification capability is lost
Solution Approach 1:
The patent segments the target motion signal into multiple coherent processing intervals (CPIs) and extracts features from each segment. By dividing the continuous radar signal into discrete temporal segments and analyzing the evolution of features across these segments, the system can classify different types of human motion (walking, running, creeping) while maintaining the detection capability of classical radar techniques.
Solution Approach 2:
The patent adds a temporal dimension to the analysis by creating feature vector time series from multiple CPIs. Instead of analyzing a single static feature vector, the system examines how features evolve over time across multiple radar pulses, enabling motion type classification while preserving the detection capabilities of traditional radar.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach allows for rapid and effective classification of target motion with reduced processing power, improved noise resistance, and the ability to identify specific actions of targets at a distance, even in the presence of system noise.
Implementation Method 1
Pulsed radars can be used to detect various moving objects in an environment at long distances. These radars illuminate the direction to be observed with short-time radiofrequency (RF) signal pulses that are emitted at certain intervals, and after the illumination is completed, said radars detect the location of targets by processing radio frequency echo reflected from the targets
Implementation Method 2
By means of Continuous-Wave Radars, velocity of a target can be determined and by virtue of additional processing, the change in motion-related Doppler frequency over time can be observed
Data Source
AI summary
With the present invention, there is provided a method for classification of the motion of a moving target by means of a radar signal received from the target. The method comprises the steps of creating different motion-specific models using Hidden Markov Model for all the motions desired to be classified; obtaining a database which includes the created different motion models; detecting a target by a pulsed radar; determining the range cell of the detected target within a coherent processing interval; storing signals received, at each pulse within a coherent processing interval, from the target within said range cell; finding feature vectors, at a certain frequency, from a continuous signal obtained by processing the stored signals and adding them together; generating a feature vector time series by arranging the found feature vectors in succession; comparing the generated feature vector time series to the Hidden Markov Model-based motion models in the database so as to calculate its probability for each motion model that is present in the database; selecting the motion determined by the motion model with the highest calculated probability as the motion of the target.