20 March 2020 in IEEE Geoscience and Remote Sensing Letters
We propose the detection of humans using millimeter-wave FMCW radar based on time-varying signatures of range-Doppler diagrams using deep recurrent neural networks (DRNNs). Demand for human detection is increasing for security, surveillance, and search and rescue purposes, recently, with a particular focus on urban areas filled with clutter and moving targets. We suggest the classification of targets based on their signatures in range-Doppler plots with time because the signatures can be consecutively observed. We measure five target types: humans, cars, cyclists, dogs, and road clutter using millimeter-wave FMCW radar that transmits fast chirps at 77 GHz. To maximize the classification accuracy using the time-varying range-Doppler signatures of the targets, we investigate and compare the performance of 2-D-deep convolutional neural networks (DCNN), 3-D-DCNN, and DRNN along with 2-D-DCNN. The DRNN plus 2-D-DCNN showed the best performance, and the classification accuracy yields 99%, with the human classification rate of 100%.
14 August 2020 in IEEE Transactions on Aerospace and Electronic Systems
Radar mounted on a moving vehicle returns timevarying detections corresponding to unstable scattering points, unlike optical sensors which produce relatively stable detections. We present an efficient extended Kalman filter-based simultaneous localization and mapping (EKF-SLAM) algorithm for radar, utilizing new techniques of clustering and sifting the timevarying detections. Velocity bias and yaw rate bias, which are inherent in any odometer are also estimated and compensated using the same EKF. For theoretical performance evaluation, the posterior Cramér-Rao bound (PCRB) for SLAM with odometer bias estimation is derived and compared with the root mean squared errors (RMSEs), with and without bias estimation. The simulation results show that the RMSE for SLAM with bias estimation is the closest to the PCRB. The proposed algorithm is also verified experimentally with field data. The comparison with the ground truth trajectory obtained from a differential global positioning system (DGPS) shows that the proposed algorithm yields an accurate trajectory estimate, even under large odometer bias. Also, the real-time capability of the proposed algorithm is verified.
5 January 2021 in IEEE Communications Letters
Direction of arrival (DOA) in MIMO radar is estimated from the phase differences between its channels. Hence, the phase difference vector, called a virtual array response vector, needs to be accurately known. We propose a new model of the virtual array response with a wide-band interleaved orthogonal frequency-division multiplexing (OFDM) signal. The angle is estimated by maximum likelihood estimation (MLE), and we show that the estimation error is lower than the one when using the systematic phase correction method which compensates for the range-dependent phase term. Also, the root mean square error (RMSE) of angle estimation with the proposed method is closer to the Cramér-Rao bound (CRB).
1 May 2023 in IEEE Access
To tackle a Doppler sensitivity problem of orthogonal frequency division multiplexing (OFDM), orthogonal time frequency space (OTFS) has been investigated, where information is carried over delay-Doppler domain. In this paper, to improve communication reliability in doubly dispersive channel, an auto-encoder (AE)-based OTFS modulation and detection scheme is developed, where the transmit OTFS waveform and its associated detection scheme at the receiver are jointly optimized in a deep learning framework. However, the conventional AE architecture which takes one-hot encoded input vector is hard to be reused in OTFS due to its enormous input dimensionality that increases exponentially on the number of grid points in delay-Doppler domain. To overcome it, we divide the delay-Doppler grid into multiple subblocks and associate the one-hot encoded vector with each subblock. Then, by concatenating them, one multi-hot vector is formed and exploited as the input vector for the proposed AE-based OTFS modulation and detection. We also develop a meta-learning scheme to effectively train the AE-based OTFS transceiver for newly updated channel profile.
8 November 2022 in IEEE Sensors Journal
In this paper, a stepped-carrier orthogonal frequency division multiplexing (OFDM) system with a nonlinear hopping pattern is developed for joint radar and communication (RadCom). In the proposed OFDM-based RadCom system, the carrier frequencies of OFDM symbols for multiple RadCom nodes are orthogonally switched to avoid interference while also maximizing the signal-to-radar interference ratio (SRIR). Differently from conventional linear stepped-carrier OFDM radar that does not consider the communication performance, we first develop a subband allocation method that takes into account both the radar and the communication performance for multiple RadCom nodes. Channel state information (CSI) of multiple nodes are shared with a centralized server, which then computes the subband allocation strategy that gives the maximum SRIR. Furthermore, a distributed subband allocation strategy based on stateless Q-learning is also proposed, which is suitable when the RadCom nodes do not share their CSI. We also develop the radar signal processing algorithm for range and velocity estimation when the stepped-carrier OFDM waveforms are transmitted through multiple subbands with a nonlinear hopping pattern. Through computer simulations, we confirm that the proposed stepped-carrier OFDM with the nonlinear hopping pattern exhibits a higher achievable sum rate than conventional linear stepped OFDM while not sacrificing the radar performance of target parameter estimation.
10 October 2019 in IEEE Transactions on Aerospace and Electronic Systems