To encourage users for applying our SDRs to real world applications, we created this Ancortek Research Center to share with our users some of the experiment results using our SDR units (Matlab source code available with every purchase).


Research Based on SDR Data



SDR KIT can be used as an inverse synthetic aperture radar to form 2-D range and cross-range images of moving targets. Since the cross-range can be derived from Doppler frequency, the 2-D image is a range-Doppler image (Figure 1).


To form a range-Doppler image of a moving target, the first step is to carry out the translational motion compensation (TMC). It estimates target’s translational motion parameters and removes the extra-phase term, such that target’s range is no longer varying with time (Figure 2). Then, by taking Fourier transform along the slow-time domain, a range-Doppler image of the moving target can be reconstructed (Figure 3).


If the target also rotates about an axis, the rotational motion can make Doppler shifts to be time varying. Thus, the range-Doppler image may be smeared in the Doppler domain. In this case, a rotational motion compensation must be taken to correct for the rotation.


MATLAB source codes and SDR-KIT data can be obtained upon request.



Figure 1 – Inverse synthetic aperture radar imaging of a moving vehicle.



When a person is walking, we can see the head, torso, and swinging feet from the range-velocity image. Figure 1 shows radar image sequence of a person walking toward the radar. Figure 2 shows radar image sequence of a person walking toward the radar and then walking away from the radar. Figure 3 shows radar image sequence of a person circular running in front of the radar.


MATLAB source codes and SDR-KIT data can be obtained upon request.



Figure 1 – Radar range-velocity imaging of a walking person.



Figure 2 – Radar range-velocity imaging of a person walking around.



Figure 3 – Radar range-velocity imaging of a person circular running.



A DJI Phantom 3 was used for the data collection. For experimental purpose, the speed of the Phantom 3 was less than ±3 m/s. A circular flying trajectory was used for the data collection as shown in Figure 1.



Generally, the rotation speed of propellers is about 33 rps when idle, 150 rps when at full power and 110 rps while hovering. The diameter of the rotor blade is 24 cm. The main radar return is around 3 cm away from the center of the blade. When the blade rotates at the speed of 110 rps, the 3 cm-away-from-the-center-point is at 20 m/s, hence the main expected Doppler frequency shift we can observe of a rotating blade is around 774 Hz, which is higher than the one that current FMCW configuration can afford. Hence, the CW mode with 128kHz of sampling rate was used.


     A 30-second recording time was selected for collecting the data while the UAV was flying back and forth relative to the radar. The micro-Doppler signature of the Phantom 3 is shown in Figure 2. The body of the drone is the main Doppler component. Majority of the energy reflected from the blades is ± 500Hz around the main Doppler shift, spreading beyond the 2kHz limit. Due to arbitrary initial phase of the four rotors in the UAV, information such as blade length, rotation speed cannot be identified from the micro-Doppler signature. However, the maximal micro-Doppler shift can serve as a feature to classify the UAV from birds.



Extraction of Heart Beat and Respiration 


Any frequency band can be used for the vital sign study. K-band SDR-KIT 2400 was used for the study. The waveform used was CW with 128 kHz data sampling rate and 10 sec recording time. After DC cancellation, data rate reduction, and I-Q imbalance correction, the arctangent demodulation operation, phase unwrapping, differencing, and detrending processing are applied to the digitized baseband signals. The estimated spectrum of the heartbeat and respiratory signals using the K-band radar is shown in Figure 1.


FMCW waveform can also be used for vital sign study for providing additional distance information. By taking FFT along the fast-time, a 2-D range profiles in the range and slow-time domain can be formed as shown in Figure 2(a). After two bandpass filtering, the estimated spectrum of the heartbeat and the respiratory are shown in Figure 2(b) and (c).


Both CW radar data and FMCW radar data and a set of MATLAB codes for processing vital sign data can be obtained upon request.



Figure 1 – The spectrum of vital signs estimated by the experimental CW radar.



Figure 2 – FMCW radar data for human vital signs: (a) 2-D range profiles; (b) spectrum of heartbeat signals; and (c) spectrum of respiratory signals.


Research Based on Simulated Data

Sample Description