Radar Waveform Recognition Using Hidden Markov Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current radar systems face challenges in recognizing pseudo-random waveforms, especially when they are mixed with pulses from other transmitters, due to limitations in handling unlearned waveforms and requiring extensive recalibration, and they struggle with efficient real-time identification in noisy environments.
Innovation Solution
A method using a hidden Markovian process to model the waveform pattern, determining a score by traversing a decision tree based on pulse parameter measurements, and learning a statistical model from reference data to improve waveform recognition performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If machine learning-based multi-class classification is used for waveform recognition, then recognition capability is improved, but computational complexity increases and the system cannot handle untrained waveforms
Solution Approach 1:
The patent extracts the essential statistical structure of radar waveforms by focusing on the sequence of pulse arrival times and their probabilistic transitions, rather than processing the entire signal through complex neural networks. This extraction of key statistical features enables simpler processing while maintaining recognition capability.
Solution Approach 2:
The patent replaces complex neural network-based mechanical processing with a statistical model based on Markov chains that uses probabilistic transitions to characterize waveforms. This substitution reduces computational complexity while maintaining the ability to recognize waveforms including untrained ones.
2Reliability
If traditional multi-class classification is used, then trained waveform recognition is improved, but the system fails when encountering untrained or unknown waveforms
Solution Approach 1:
The patent creates a universal waveform recognition system that can handle both trained and untrained waveforms using the same statistical model framework. The Markov chain model serves multiple functions: it characterizes known waveforms for recognition and also provides a framework for detecting and adapting to unknown waveforms without requiring separate classification systems.
Solution Approach 2:
The patent introduces dynamic adaptability through the Markov chain model that can evolve to characterize new waveform types as they are encountered. The model's transition probabilities can be updated online, enabling the system to adapt to new waveforms dynamically without requiring retraining of a fixed classification model.
3Speed
If waveform recognition is performed in real-time with small analysis windows, then response time is improved, but the amount of available data for analysis is reduced
Solution Approach 1:
The patent applies partial action by analyzing only the essential statistical features (arrival time sequences and their transitions) rather than processing all pulse data. This selective analysis of key probabilistic characteristics enables real-time processing with small data windows while maintaining sufficient information for accurate waveform recognition.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
Method for recognizing the waveform of a signal received from a transmitter, comprising electromagnetic pulses, using a waveform model according to a hidden Markovian process, comprising a comb of values µ representing process states corresponding to possible values of a pulse parameter, a transition matrix P between consecutive values of the comb, and the standard deviations σ of measurement of the states, the method comprising: - receiving N measurements of the pulse parameter, - determining a score, by traversing (606) a tree whose each node corresponds to one of the process states µk and providing a prediction of the measurement of the pulse parameter, - recognizing the waveform of the signal from the score, the score being updated (607) iteratively, for each node traversed of the tree associated with a process state µk according to a precondition indicating whether the prediction reaches a pulse.