Through-the-Wall Detection of Stationary Human Targets Using Doppler Radar
Summary (3 min read)
1. INTRODUCTION
- In recent years, there has been a great deal of research directed towards the use of Doppler-radar systems for monitoring human activity.
- Doppler-radar was first demonstrated for remotely monitoring human activity in [1, 2].
- Human activity can be considered as a combination of one or many of these movements, and each activity occurs over a different time scale.
- Based on the reasoning presented in Section 3, the criteria the authors choose to decide on the time-frequency technique are — frequency resolution, ability to resolve time-frequency components of low amplitude, non-linearity of transformation and adaptive selection of time-scales.
- The novel contribution of the present paper is the detection and characterization of Doppler from stationary humans, i.e., wherein the human torso is not moving.
2.1. Introduction to EMD-HS
- The EMD-HS algorithm (also called the Hilbert Huang Transform (HHT)) was proposed in [13] for analyzing non-stationary signals originating from non-linear processes.
- EMD extracts intrinsic oscillatory modes defined by the time scales of oscillation, called IMFs.
- Such functions permit the application of the Hilbert transform and the corresponding definition of instantaneous frequency in [13].
- The functions s(t) and ω(t) are the instantaneous amplitude and instantaneous frequency of the signal, respectively.
2.2. Sifting Process
- The basic step of the EMD algorithm is the sifting process which essentially extracts scales of the signal.
- The points of set S1max are interpolated to form the upper envelope of the signal, x̂max.
- Following this, the function xr1 = x(t) − x1(t) is created, and the sifting process is repeated, resulting in x2(t), the second IMF.
- Empirically, the EMD has been shown to be effective in extracting relevant components in a variety of applications involving non-stationary signals.
- The first IMF represents the fastest modes of oscillation in the signal, and with subsequent IMFs, the frequency, as measured by the number of zero crossings decays exponentially as the index of the IMF.
3. MODELING DOPPLER SIGNATURES DUE TO HUMAN ACTIVITY
- The Doppler modulations due to human activity vary in time according to the dynamics of human movement.
- Non-stationary models for Doppler due to walking human targets were proposed in [8, 14].
- Walking induces high Doppler shifts in the waveform that can be observed over short time durations.
- The finite non-zero dimensions of the human arm and other parts of the body result in a Doppler return that consists of multiple frequency components at each time instant [14].
- The authors conjecture that a human whose torso is not moving can be identified from the Doppler signatures due to activity such as breathing and movements of the arm.
3.1. Modeling Human Arm Motion
- A characteristic Doppler event associated with stationary human targets is the movement of the arm.
- Details of the motion of the arm contains information regarding the intent of humans behind the wall.
- It is desirable to detect and characterize Doppler signatures of human arm motion for through-the-wall monitoring applications.
- The authors present a model for Doppler due to human arm movements.
- Human arm motion is composed of three components, defined by the joints driving its motion.
3.2. A General Model for Human Arm Motion
- The three components of the arm, as represented by the wrist, forearm and arm can each be modeled as a solid shaft exhibiting rotational motion around the corresponding joint- wrist joint, elbow joint or shoulder joint.
- In such a model, the movement of the human arm is defined by the three components: ω1, ω2, and ω3, representing the angular velocities of the three segments OA, AB and BC, around the points defined by O, A and B, respectively.
- For deriving the Doppler shift resulting from this motion, the authors consider an infinitesimal element on each of the line segments OA, AB and BC.
- From Equation (6), it is clear that the length of the moving component controls this ‘spread’ in the frequency implying that a scatterer of larger dimensions results in a higher frequency-spread.
3.3. Velocity of the Human Arm
- The goal of their work is to identify Doppler characteristics that distinguish human activities.
- This information has to be extracted from the time-dependency of the frequency, and the spread of the frequency.
3.3.1. Doppler Modeling Based on the Biomechanics of Human Movement
- Doppler-radar models for human walking based on well known models of human locomotion used in computer animation are presented in [8, 22].
- Figure 2 shows the idealized velocity profile of human arm movement considered in [23].
- The Doppler shift due to a single element of the human arm is integrated over the entire length.
- The authors drop the subscript i from Equation (6) for convenience.
- The unimodal velocity profile over the duration of motion is seen to result in a return signal with four distinct maxima and a region of stationary points close to the time instant of maximum velocity.
3.4. Intermittent Human Activity
- The return signal can then be represented as a linear combination of different waveforms, each of which is non-zero over a different time interval and with each waveform corresponding to the Doppler modulation due to the human activity.
- Over each time interval, a different type of human motion results in a different Doppler modulation of the transmit signal which is represented as a non-stationary signal ai(t).
- Without a knowledge of ti, it is not possible to pre-define optimum time and frequency resolutions for computing the joint time frequency distributions.
- If for some ai(t), the events are non-stationary within the width of the window function w(t), then the spectrogram will fail to capture the complete time-frequency distribution of ai(t).
4. EXPERIMENTAL RESULTS
- A human target located behind a brick wall of about 16 cm thickness was imaged using a radar system operating in the ultrahigh frequency (UHF) band.
- The transmitted power was −5 dBm and the antenna gain was 5 dB.
- The 750-MHz radar system was used to extract Doppler signatures associated with different activity associated with a stationary human.
- The IMFs are indexed inversely as the scales of oscillations.
- The Doppler oscillation caused by a person shuffling from a stationary position for about 2 s produces features of about 1.5.
4.1. Signatures of Different Types of Arm Movement
- The human arm model was earlier described as consisting of three components, centered at the shoulder, the elbow and the wrist joints.
- Figure 4 illustrates Doppler features extracted from the two experiments.
- In the first experiment, the human target repeatedly moved the wrist around the wrist joint for a duration of 10 seconds, while keeping the rest of the arm stationary.
- The energy distribution across the IMFs is considerably flat, and demonstrates that the energy is concentrated within a small number of IMFs.
- The number of non-stationary oscillatory components in the former is higher than in the latter.
4.2. Experimental Verification of the Kinematic Model
- The authors present the experimental results of human arm movement.
- The return waveform was processed as described earlier.
- The resulting plots for three different trials with different subjects are shown in Figure 5.
- This demonstrates the viability of characterizing activities associated with a stationary human using a model based approach.
- The experimental results validated the theoretical results as given by the Gaussian velocity profile model even when different individuals were used as targets.
5. CONCLUSION
- The authors have developed a system for through-the-wall detection of a stationary human, based on the empirical mode decomposition-Hilbert spectrum algorithm.
- The Doppler detection system was validated by testing the algorithm on real data.
- The modeled waveform compared favorably with experimental results.
- A model based approach for classifying human activity was thus shown to be feasible.
- Doppler modulations due to different types of human activity were shown to occur over different scales.
Did you find this useful? Give us your feedback
Citations
14 citations
Cites methods from "Through-the-Wall Detection of Stati..."
...Using this concept, X-band radars were used to detect heartbeat and breathing from a rabbit confined in a cardboard box and a human at a distance of 30 cm [1], and from a human at a distance of 30 m behind a cinder block wall [2]....
[...]
12 citations
12 citations
12 citations
11 citations
Cites result from "Through-the-Wall Detection of Stati..."
...Similar results for a radar operating at 750 MHz were presented in [15] and [16] and for a Noise Radar for through-the-wall imaging in [17]....
[...]
References
18,956 citations
"Through-the-Wall Detection of Stati..." refers background or methods in this paper
...The EMD-HS algorithm (also called the Hilbert Huang Transform (HHT)) was proposed in [13] for analyzing non-stationary signals originating from non-linear processes....
[...]
...The EMD-Hilbert spectrum (referred to hereafter as EMD-HS) algorithm is a recent development in the field of time-frequency analysis [13]....
[...]
...The instantaneous frequency of this analytic signal is defined as the derivative of the instantaneous phase defined in [13]....
[...]
...Such functions permit the application of the Hilbert transform and the corresponding definition of instantaneous frequency in [13]....
[...]
2,304 citations
"Through-the-Wall Detection of Stati..." refers background or methods in this paper
...The application of the EMD algorithm to characterize random noise has been discussed in [20]....
[...]
...Similarly, the energy of the IMFs also reduces according to an exponential rule [20]....
[...]
1,306 citations
939 citations
"Through-the-Wall Detection of Stati..." refers background or methods in this paper
...The measured trajectory is given in [23]....
[...]
...The results in [23] and [24] suggest that a human arm moving in response to a stimulus follows a similar velocity profile across different human subjects and trials....
[...]
...In this section, we propose a model for the motion of the human arm, primarily based on [23]....
[...]
...In [23], the authors monitored the velocity of the human arm using a set of light emitting diodes placed on the human arm....
[...]
...Figure 2 shows the idealized velocity profile of human arm movement considered in [23]....
[...]
808 citations
"Through-the-Wall Detection of Stati..." refers background in this paper
...The velocity profiles for the movement of the human arm in response to different types of stimuli are presented in [24]....
[...]
...The results in [23] and [24] suggest that a human arm moving in response to a stimulus follows a similar velocity profile across different human subjects and trials....
[...]