TY - GEN
T1 - LiquID
T2 - 16th ACM International Conference on Mobile Systems, Applications, and Services,MobiSys 2018
AU - Dhekne, Ashutosh
AU - Gowda, Mahanth
AU - Zhao, Yixuan
AU - Hassanieh, Haitham
AU - Choudhury, Romit Roy
N1 - Funding Information:
Fitting LiquID into a mobile form factor entails incorporating the space, energy, and computation constraints of such platforms. An UWB chip occupies a 4mm × 4mm area and weighs 0.105дm [4]. Whereas we have used half-wavelength dipole antennas in our experiments, much smaller antennas are available [2]. UWB is a low-power protocol [22] and the decawave chip is rated to consume a maximum of 120mAh [4]. The signal processing blocks required by LiquID are already present on mobile devices [36]. Pipelining the signal processing with fetching of the CIR data can allow LiquID to run at near-realtime. Finally, we envision a mobile device with an antenna connected to an extensible wire. A liquid container is placed between the mobile device and this antenna to identify the liquid. 8 CONCLUSION This paper shows the feasibility of identifying liquids by analyzing UWB signals passing through it. We measure the time of flight of the signal and combine with its phase and RSSI to ultimately model the permittivity of the liquid. Given permittivity serves as a signature, it is now possible to identify the liquid without inserting probes into it. Our solution is realtime (sub-second latency), cheap (≈ $150), and lightweight (few pounds), underpinning a variety of applications. Our next step is focussed on analyzing more complex liquid mixtures, such as impure drinking water, blood, saliva, and other biologically relevant liquids. ACKNOWLEDGMENTS We would like to acknowledge Prof. José Schutt-Ainé for enabling the use of a vector network analyzer and dielectric probe required to obtain the baseline permittivity. We would also like to thank Davy Davidson of Boda Acrylic for creating the experimental acrylic box at short notice. We sincerely thank our shepherd, Dr. Tam Vu, and the anonymous reviewers for their valuable feedback. We are grateful to NSF (CNS - 1719337) for partially funding this research.
PY - 2018/6/10
Y1 - 2018/6/10
N2 - This paper shows the feasibility of identifying liquids by shining ultra-wideband (UWB) wireless signals through them. The core opportunity arises from the fact that wireless signals experience distinct slow-down and attenuation when passing through a liquid, manifesting in the phase, strength, and propagation delay of the outgoing signal. While this intuition is simple, building a robust system entails numerous challenges, including (1) pico-second scale time of flight estimation, (2) coping with integer ambiguity due to phase wraps, (3) pollution from hardware noise and multipath, and (4) compensating for the liquid-container’s impact on the measurements. We address these challenges through multiple stages of signal processing without relying on any feature extraction or machine learning. Instead, we model the behavior of radio signals inside liquids (using principles of physics), and estimate the liquid’s permittivity, which in turn identifies the liquid. Experiments across 33 different liquids (spread over the whole permittivity spectrum) show median permittivity error of 9%. This implies that coke can be discriminated from diet coke or pepsi, whole milk from 2% milk, and distilled water from saline water. Our end system, LiquID, is cheap, non-invasive, and amenable to real-world applications.
AB - This paper shows the feasibility of identifying liquids by shining ultra-wideband (UWB) wireless signals through them. The core opportunity arises from the fact that wireless signals experience distinct slow-down and attenuation when passing through a liquid, manifesting in the phase, strength, and propagation delay of the outgoing signal. While this intuition is simple, building a robust system entails numerous challenges, including (1) pico-second scale time of flight estimation, (2) coping with integer ambiguity due to phase wraps, (3) pollution from hardware noise and multipath, and (4) compensating for the liquid-container’s impact on the measurements. We address these challenges through multiple stages of signal processing without relying on any feature extraction or machine learning. Instead, we model the behavior of radio signals inside liquids (using principles of physics), and estimate the liquid’s permittivity, which in turn identifies the liquid. Experiments across 33 different liquids (spread over the whole permittivity spectrum) show median permittivity error of 9%. This implies that coke can be discriminated from diet coke or pepsi, whole milk from 2% milk, and distilled water from saline water. Our end system, LiquID, is cheap, non-invasive, and amenable to real-world applications.
UR - http://www.scopus.com/inward/record.url?scp=85051535387&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85051535387&partnerID=8YFLogxK
U2 - 10.1145/3210240.3210345
DO - 10.1145/3210240.3210345
M3 - Conference contribution
AN - SCOPUS:85051535387
T3 - MobiSys 2018 - Proceedings of the 16th ACM International Conference on Mobile Systems, Applications, and Services
SP - 442
EP - 454
BT - MobiSys 2018 - Proceedings of the 16th ACM International Conference on Mobile Systems, Applications, and Services
PB - Association for Computing Machinery, Inc
Y2 - 10 June 2018 through 15 June 2018
ER -