Radio frequency signals have the potential to convey rich information about a node’s motion and surroundings, which can ultimately assist in characterizing environmental mobility and assisting in uninterrupted Tb/sec scale communication. Unfortunately, extracting such information is challenging. Usually, it involves accurate phase measurement, large antenna array structures, or extensive training. In this work, we present LeakyTrack, a novel system that enables non-coherent and training-free motion sensing with a single antenna. The key idea is to create unique spectrally coded signals at different spatial directions so that geometric properties of the receiving node, as well as any potential objects in the environment, leave spectral footprints on the collected signal. To do so, we exploit a single leaky wave antenna and create a THz Rainbow transmission that has distinct signals with unique spectral characteristics across the angular domain. Leveraging the spatial-spectral signatures in the THz Rainbow, all receivers can correlate the measured signal with the known transmission signatures to discover the sender’s path directions in one-shot.
Fundamentally, radio frequency sensing relies on the principle that changes in the characteristics of electromagnetic waves, as they travel in the wireless medium, can provide rich information about a receiver’s properties. That information contains critical details such as receiver’s distance, velocity, shape, size, and orientation as well as the objects that have interacted with the waves on their route. It is typically viewed that such properties (i) cannot be inferred with a single antenna, as a single antenna provides only one spatial observation point; and (ii) cannot be inferred solely based on detecting incident power, but rather require careful timing analysis of the phase of the incoming signal, i.e., coherent reception. In this work, we challenge that notion and show that neither of these common beliefs is true. In particular, we present the design and experimental evaluation of LeakyTrack, a novel non-coherent and training-free system that identifies nodal and environmental motion using a single antenna.
First, we transmit unique spectral codes in the spatial domain by introducing a novel radio frequency sensing architecture. Equipping the transmitter and receiver with LWA, we exploit the properties of LWA and develop the first LWA-based nodal and environmental sensing system. We show how to simultaneously scan a wide range of spatial angles by injecting a time-domain THz pulse, unlike naive LWA implementation that would create a spectral code at each direction by varying the frequency of the input signal. As a result, different frequencies fill the entire angular space with signals that have unique spectral content or color, color-coded scan.
Second, we demonstrate that LeakyTrack can sense nodal and environmental mobility by analyzing the spectral profile of the collected signals. In particular, we first devise a model that allows a LWA-equipped receiving node to locally predict its angular location and orientation relative to the TX. Color-coded scan allows the receiver to only capture certain signals with unique spectral codes. We show that the location-specific spectral codes experience a further non-uniform and frequency-dependent loss at the receiver LWA related to the receiver’s orientation. LeakyTrack leverages these insights to jointly extract location and orientation from the spectral information that was harvested rapidly, via a single pulse of broadband emission
Finally, we implement LeakyTrack on experimental testbed and perform extensive over-the-air experiments. Our testbed consists of a THz pulse generator, a broadband receiver, and custom LWAs configured in numerous topologies including with obstructing objects. Our key findings are summarized below.
(i) We experimentally realized a color-coded scan by injecting a THz pulse into an LWA and demonstrated that LeakyTrack can successfully sense and track the receiver’s parameters, even in complex real-world trajectories, with an average estimation error of less than 1 degree for angular position and less than 2 degrees for rotation angle.
(ii) We find that such motion inferences are feasible only within certain regimes. While characterizing these detection zones, we discover that LeakyTrack is prone to higher errors in regimes where spectrum is a slow-varying or irregular function of angle. Further, rotating a device causes power attenuation due to frequency misalignment, which ultimately yields negligible spectral information being received such that spatial inferences cannot be made. Nonetheless, we show that a surprisingly large range of rotation angles can be accurately tracked.
(iii) We measure the frequency-selective blockage footprints of objects at different locations and sizes. Within the field-of-view of the color-coded scan, we find that LeakyTrack accurately estimates the object’s geometric properties such as spatial position and size. A relatively larger object leaves more pronounced spectral footprints making the detection easier; nonetheless, under full blockage, the object positioning would be ambiguous.
(iv) We experimentally explore the tradeoff between estimation accuracy, airtime overhead, and computation complexity and quantify how improved computational efficiency costs lower estimation accuracy or higher airtime channel use.
Publications:
Ghasempour, Y., Yeh, C.Y., Shrestha, R., Amarasinghe, Y., Mittleman, D. and Knightly, E.W., 2020, November. “LeakyTrack: non-coherent single-antenna nodal and environmental mobility tracking with a leaky-wave antenna,” in Proceedings of ACM SenSys 2020, November 2020.