Continuous-Wave radars – applications, security, and target emulator

Radars have been increasingly used for human-aware detection/activity recognition in the smart living sector, and to realize advanced driver-assistance systems (ADAS) and autonomous driving in the automotive industry. Prior to large-scale deployment of these sensors, it is essential to study the potential threats against these sensors that can interrupt their designed functionality, and comprehensively test them in scenarios that mimic real-world conditions. This dissertation presents three interlinked aspects related to continuous-wave (CW) radars: 1) their application as a sensing modality for smart homes, cities, and infrastructure, (2) a feasibility study on the possible spoofing attacks against CW radars, and (3) a low-cost CW radar target emulator for extensive testing under various background conditions. It is crucial for the radar sensors deployed indoors to isolate the target of interest from unwanted clutter sources. A novel approach to suppress both stationary and moving clutter sources based on exponential moving average (EMA) filtering is proposed for indoor sensing. Although EMA-based filtering techniques were used for stationary clutter suppression in existing works, the proposed work addresses moving clutter suppression as well. The proposed approach removes all motion artifacts outside the characteristic frequency range of the human cardiopulmonary motion. A filter-and-subtract method is employed, where the output from two filters with different cut-off frequencies is subtracted so that all moving clutter signatures are attenuated while retaining the human target signatures. For experimental validation of the proposed approach, a 60-GHz frequency-modulated continuous-wave (FMCW) radar with single-input multiple-output (SIMO) architecture is used to uniquely identify the 2-D location of a stationary human subject, with various moving and stationary clutter sources in the background. By leveraging the digital beamforming (DBF) capability of the 60-GHz radar, a sideways hand gesture recognition technique is proposed that determines the instantaneous 2-D position of the hand at the start and end of the gesture. To detect the start and end time of the gesture, EMA filtering was applied to attenuate the reflections from the human body so that the signature of the gesture is prominent. The results presented can differentiate left-to-right and right-to-left lateral hand gestures. A novel frequency-domain spoofing attack model is proposed to investigate the vulnerability of FMCW radars against malicious attacks. The proposed model avoids the need for precise nanosecond synchronization with the victim radar’s chirp transmission time. A single-sideband (SSB) mixer is the key component of the attack system, which introduces a frequency shift to the radio frequency (RF) chirp signal transmitted by the radar and retransmits the modulated chirp signal. Upon deramping operation on the radar's receiver chain, the introduced frequency shift by the SSB mixer translates to the beat frequency of the baseband signal, thereby creating an illusion of a real target. The frequency shift introduced can be varied to alter the range of the fake target. The theory of the spoofing model is developed, and a 5.8-GHz proof-of-concept spoofing system is designed to provide experimental validation. A hybrid-chirp FMCW waveform is proposed to distinguish a real target from a spoofing target to mitigate spoofing attacks. Finally, a low-cost radar target emulator (RTE) for Doppler and FMCW radars is proposed. The proposed RTE is designed to emulate the radar’s response generated by two human activities: the Doppler artifacts of a human elbow-down gesture as seen by a Doppler radar, and the inherent chest motion of a stationary human subject measured by an FMCW radar. For the Doppler mode RTE, an SSB mixer is utilized to electronically modulate the CW signal transmitted by the radar to resemble a human hand gesture. The vital sign motion of a stationary human target is emulated using an analog phase shifter that serves as an integral part of the FMCW mode RTE. An SSB mixer integrated with the FMCW RTE showcased the ability to vary the range of the human target as well. The Doppler- and FMCW- mode target emulators are realized using 5.8 GHz commercially available off-the-shelf RF components. Due to the high similarity between the system-level architecture of an RTE and a spoofing system, the RTEs discussed above are set up as spoofing systems to deceive two state-of-the-art human detection algorithms.

Embargo status: Restricted until 09/2023. To request the author grant access, click on the PDF link to the left.