Mobi2Sense: Enabling Wireless Sensing Under Device Motions
Junqi Ma†∗ , Zhaoxin Chang§† , Fusang Zhang†∗ , Jie Xiong‡ , Beihong Jin†∗ Daqing Zhang§♯
†Institute of Software, Chinese Academy of Sciences, ∗ University of Chinese Academy of Sciences,
‡ University of Massachusetts Amherst, § Institut Polytechnique de Paris, France, ♯ Peking University,
ABSTRACT
Besides the communication function, various RF signals such as
WiFi and RFID have been actively exploited for sensing purposes
recently. However, a missing component of existing RF sensing is
sensing under device motions. This paper takes the first step to in-
volve device mobility into the ecosystem of RF sensing. Owning to
the miniaturization and low cost of ultra-wideband (UWB) chips in
recent years, we propose to integrate the accuracy of UWB sensing
with device mobility to support truly ubiquitous RF sensing. This is
a challenging task because the motion artifacts from RF devices can
easily overwhelm the target motion, such as subtle chest displace-
ment for respiration sensing. In this demo, we propose Mobi2 Sense
to support sensing under device motions. We propose novel signal
processing schemes to remove the effect of device motions on sens-
ing and prototype Mobi2 Sense using a commodity UWB module.
Comprehensive evaluation demonstrates that Mobi2 Sense is able
to “hear” music and “see” human respiration at high accuracy in
the presence of device motions.
CCS CONCEPTS
• Human-centered computing → Ubiquitous and mobile com-
puting systems and tools.
1 INTRODUCTION
Recent years have witnessed the rapid progress of contactless sens-
ing using various wireless signals including WiFi [1], RFID [3],
mmWave [6], ultrasound [4] and LoRa [7]. The contact-free and
sensor-free nature of wireless sensing is appealing in many real-life
scenarios. A large range of applications have been enabled by wire-
less sensing including activity tracking [8], gesture recognition [1],
respiration monitoring [4] and vibration sensing [5]. While promis-
ing progress has been achieved, one interesting observation about
existing RF sensing systems is that the transmitter and receiver
are always static (e.g., placed on a desk or mounted on the ceiling).
However, it is noted that in some real-life scenarios, the RF device
can be non-static. For example, a smartphone can be held in hand by
a user to sense a target. A moving robot equipped with RF sensing
modules would prefer to be able to sense the surroundings without
a need to stop in order for RF sensing to work. Therefore, a missing
part of RF sensing is sensing with a mobile RF device.
In this work, we consider device mobility in RF sensing for the
first time. To see the effect of device motions on sensing, we conduct
Permission to make digital or hard copies of part or all of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for profit or commercial advantage and that copies bear this notice and the full citation
on the first page. Copyrights for third-party components of this work must be honored.
For all other uses, contact the owner/author(s).
ACM MobiCom ’22, October 17–21, 2022, Sydney, NSW, Australia
© 2022 Copyright held by the owner/author(s).
ACM ISBN 978-1-4503-9181-8/22/10.
https://doi.org/10.1145/3495243.3558748
766
Ground truth signal Device is static
variation
Human respiration Device held Device on a moving
monitoring in hand robot
Figure 1: Signal of respiration under device motion.
a range of benchmark experiments. Surprisingly, we find that RF
sensing stops working when an RF device is placed on a moving
robot. Even under subtle hand movement when a smartphone is
held in hand, the sensing performance is significantly degraded, as
shown in Figure 1.
To handle the challenge of device motion, we propose Mobi2 Sense,
a system which is capable of removing random device motions to
enable fine-grained sensing under device motions. The basic idea
is to employ one static object in the environment as a reference
to cancel out the effect of device movement. The reference can be
commonly-seen static objects such as a wall, a desk, a laptop, or
even a water bottle. To achieve the goal, we model the effect of
device motion on signal variation. We cancel the signal variation
caused by device motion through a division operation between the
signal reflected from the target and the signal reflected from the
static reference. We demonstrate the correctness of the proposed
method both theoretically and experimentally. To tackle the incon-
sistency of viewing device motions at different locations, we convert
the signal variation caused by device motion at the reference to
that at the target to make them consistent for cancellation.
We implement the proposed contactless sensing system Mobi2 Sense
using commodity UWB module and conduct comprehensive exper-
iments under two typical motion scenarios, i.e., hand-held scenario
and robot-carried scenario. We further showcase two applications
of different target motion scales: i) employing the proposed sys-
tem to sense sub-millimeter level vibration of a speaker to “hear”
the music play; ii) holding the UWB device in hand to accurately
monitor the patients’ millimeter level respiration (rate and depth)
in a real hospital environment. For both applications, our system
achieves highly accurate sensing performance in the presence of
device motions (both held in hand and placed on a moving robot).
2 SYSTEM DESIGN
In this section, we introduce the design of our system. Figure 2
shows the workflow of Mobi2 Sense, which consists of three mod-
ules.
 Sensing scenario Reference Device motion Target movement
selection recovery Web-based user interface
Device with Target cancellation
motion
Range-profile Target
Q
Move to origin

Division I
Reference
UDP
Recovered
1 2 3
UWB movement
PC Static objects
module
Data collection Data processing Result demonstration

Figure 2: Workflow of Mobi2 Sense.

Data collection. We use a commercial Single-Input Single-

Reference
Output (SISO) UWB module as the sensing front-end. We connect
the UWB module to the Raspberry Pi and package it into a 3D- Reference
Target
printed box with a size of 13.9 cm × 9.4 cm as a hand-held sensing Target
device. After the Raspberry Pi receives the raw signal samples from UWB module
the UWB module, it forwards the raw data to a computer through
WiFi connection for further processing. UWB module

Data processing. Mobi2 Sense has a three-stage data processing

pipeline, including reference selection, device motion cancellation,
and target movement recovery. In reference selection, we measure
(a) Hand-held scenario (b) Robot-carried scenario
the quality of reflection signal from different static objects, and pick
the object with the highest quality as the reference to ensure the
Figure 3: Demo 1: Music recovery.
best cancellation performance. After the reference is selected, we
perform a delicate motion effect transfer to mitigate the difference of
device motion on target and reference. Then, we propose a division- respiration (millimeter level) in a contact-free manner. Vital sign
based cancellation algorithm to eliminate the device motion from sensing plays a key role in health care. As shown in Figure 4a,
the target. Finally, we employ a circle fitting algorithm to remove we show that our proposed system can enable the nurse to hold
the static component and only keep the target motion component the device to sense the patient’s respiration. The UWB device can
in the signal for sensing. also be placed on a robot to patrol around to monitor the human
Result demonstration. We develop a GUI interface to visualize vital signs as shown in Figure 4b. The demo video can be found at
the sensing performance of Mobi2 Sense, including the recovered https://youtu.be/apGZD_MRj2I.
real-time target movements and the ground-truth video.

3 DEMO SETUP
We implement Mobi2 Sense on a COTS UWB module (XETHRU
X4M05) which has an integrated X4 system-on-chip [2]. X4M05 Reference

chip works on the 7.25–10.2 GHz frequency band with a bandwidth UWB module
Reference
of 1.5 GHz. We use two applications, i.e., audio recovery and respi-
ration monitoring, to demonstrate the performance of Mobi2 Sense. UWB module
In each application, we consider the device motion of hand-held
and robot-carried scenarios.
“Listen" to music through fine-grained UWB sensing. In (a) Hand-held scenario (b) Robot-carried scenario
this application, we use our system to sense the sub-millimeter level
vibration of a commodity speaker (Philips SPA20) when it is playing Figure 4: Demo 2: Respiration monitoring.
music, as shown in Figure 3a and Figure 3b. The sampling rate of
the UWB system is set to 1200 samples/s. The speaker is placed
on the desk. We test our system’s performance in both hand-held
and robot-carried scenarios. The experiment result shows that our ACKNOWLEDGMENTS
system can accurately recover the music content in both scenarios. This work is partially supported by the National Natural Science
The demo video can be found at https://youtu.be/0IfiomQ9Jnw. Foundation of China (No. 62172394, No. 62072450) and the Youth
Respiration depth and rate monitoring. Next, we show the Innovation Promotion Association, Chinese Academy of Sciences
feasibility of using Mobi2 Sense to monitor fine-grained human (No. 2020109).
2

767
REFERENCES
[1] 2013. Whole-Home Gesture Recognition Using Wireless Signals. In Proceedings
of the 19th Annual International Conference on Mobile Computing and Networking
(MobiCom ’13). Association for Computing Machinery, 27–38.
[2] Novelda AS. 2021. X4 - Datasheet. https://www.radartutorial.eu/19.kartei/13.labs/
pubs/x4_datasheet_revE_restricted.pdf.
[3] Ziyang Chen, Panlong Yang, Jie Xiong, Yuanhao Feng, and Xiang-Yang Li. 2020.
TagRay: Contactless Sensing and Tracking of Mobile Objects using COTS RFID
Devices. In IEEE INFOCOM 2020 - IEEE Conference on Computer Communications.
307–316.
[4] Adib Fadel, Mao Hongzi, Kabelac Zachary, Katabi Dina, and Miller Robert C.
2015. Smart Homes That Monitor Breathing and Heart Rate. In Proceedings of the
33rd Annual ACM Conference on Human Factors in Computing Systems (CHI ’15).
Association for Computing Machinery, 837–846.
[5] Chengkun Jiang, Junchen Guo, Yuan He, Meng Jin, Shuai Li, and Yunhao Liu.
2020. MmVib: Micrometer-Level Vibration Measurement with Mmwave Radar. In
768
Proceedings of the 26th Annual International Conference on Mobile Computing and
Networking (London, United Kingdom) (MobiCom ’20). Association for Computing
Machinery, New York, NY, USA, Article 45, 13 pages. https://doi.org/10.1145/
3372224.3419202
[6] Teng Wei and Xinyu Zhang. 2015. MTrack: High-Precision Passive Tracking Using
Millimeter Wave Radios. In Proceedings of the 21st Annual International Conference
on Mobile Computing and Networking (MobiCom ’15). Association for Computing
Machinery, 117–129.
[7] Fusang Zhang, Zhaoxin Chang, Kai Niu, Jie Xiong, Beihong Jin, Qin Lv, and Daqing
Zhang. 2020. Exploring LoRa for Long-Range Through-Wall Sensing. Proc. ACM
Interact. Mob. Wearable Ubiquitous Technol. 4, 2, Article 68 (June 2020), 27 pages.
[8] Fusang Zhang, Kai Niu, Jie Xiong, Beihong Jin, Tao Gu, Yuhang Jiang, and Daqing
Zhang. 2019. Towards a Diffraction-Based Sensing Approach on Human Activity
Recognition. Proc. ACM Interact. Mob. Wearable Ubiquitous Technol. 3, 1, Article
33 (March 2019), 25 pages.