Drone Relays for Battery-Free Networks
Battery-free sensors, such as RFIDs, are annually attached to RFly is based on a mirrored architecture made of two main
billions of items including pharmaceutical drugs, clothes, and
manufacturing parts. RFIDs are primarily used to identify and
track objects in warehouses, factories and supply chains. The
fundamental challenge with battery-free sensors is that they are
only reliable at short distances. To overcome this challenge, the
reviewed paper presents RFly, a system that uses drones as relays
for battery-free networks. The paper presents a hardware
prototype mounted on a Parrot Bebop drone, which demonstrates
that RFly enables communication with commercial RFIDs at
over 50 m.
Keywords - Drones, RFID, Relay, Full-Duplex, Localization,
Battery-free
I. INTRODUCTION
Passive RFIDs market is one of the largest and the
fastest growing market of networked devices. Over 5 billion
pieces where sold only in 2016 alone, the primary usage
consisting of tracking and identifying objects.
Passive RFIDs are battery-free stickers which are
attached to objects similar to barcodes. When queried by a
wireless device called a reader, RFIDs respond with their
unique IDs, enabling the reader identify them from a
distance.
The main problem with passive RFID technology is the
short distance reliability, caused by the inability to reliably
power RFIDs at longer distances. However, even if the
reader is in proper range, obstacles such as walls can occur
and interfere in communication, creating blind spots.
As a result, warehouse inventory control becomes a
time-consuming process when done by employees, taking up
to a month’s work.
Alternative solutions that consisted of vision-based
system have proven themselves to be limited, only operating
in line-of-sight. The reviewed paper presents RFly, a system
that combines the agility of drones with the sensing
capabilities of RF signals, enabling drones to detect and
localize objects over a wide area.
RFly is based on a new relay technology that can
integrate with already deployed RFID infrastructure. When a
reader transmits a query, the relay picks it up, forwards it to
an RFID then forwards the RFID’s reply back to the reader.
As the drone flies, it can scan an entire warehouse,
extending the range of an already deployed infrastructure
and eliminating blind spots.
The three main properties RFly had to satisfy were:
● bidirectionally full-duplex
● preserve the phase and timing characteristics of the
forwarded packets, because the phase is needed in
the localization algorithm
● compactness, so it can be mounted on an indoor
drone
components. The first component cancels the
self-interference. Unfortunately, the cancelation introduces
random phase and frequency offsets that can alter
localization. The second component compensates for these
offsets by correcting them on the complementary link.
Building a prototype in a custom PCB circuit and
mounting it on a Parrot Bebop Drone revealed the
following:
● RFly extends the range up to 10x over scenarios
where the relay is absent
● RFly’s median localization error is 19cm
RFly is amenable to immediate deployment because it is
transparent to RFID protocol, enabling scaling of already
deployed infrastructures of RFID readers.
One of RFly’s limitations is the drone navigation that
needs optical makers in order to avoid obstacles and
preserve stability. Also, localization range is still limited to
tens of meters.
II. RFLY OVERVIEW
Passive RFIDs are battery-free sensors which
communicate with a device called a reader. A reader
bootstraps communication by sending a query on the
downlink channel. The query serves both as a mechanism to
power up the device and a means to communicate with it. A
powered up RFID responds with its unique identifier
through ON-OFF modulation by switching its internal
impedance between two states: reflective and non-reflective.
The localization techniques work with the phase of the
received response, which encodes distance information.
While working on both line-of-sight and
non-line-of-sight environments, RFly enable us to
communicate with RFID-tagged objects in highly cluttered
environments by dispatching a drone throughout the area.
The drone carries an on-board RFID relay which acts as a
transparent intermediary between readers and tags. As the
drone flies, the relay continuously forwards uplink and
downlink traffic between the readers and RFIDs. The RFID
reader collects the unique IDs obtained through the relay
and uses a local database to map each ID to the object it is
attached.
III. THE CHALLENGE OF SELF-INTERFERENCE
The key challenge in designing RFly’s relay is
self-interference.
To overcome this challenge, RFly must address two
types of self-interference: (1) Inter-link self-interference,
which refers to the interference between the uplink and
downlink forwarding paths; (2) Intra-link self-interference,
which refers to the interference between the transmit and
receive antennas on each of the uplink and downlink
processing paths separately.
RFly’s solution to Inter-link self-interference challenge
exploits the signal structure of RFID communications in the
frequency domain. Specifically, it exploits the fact that the
reader’s query and the RFID’s response occupy different
subbands around the center frequency used for
communication.
RFly’s solution to Intra-link self-interference challenge
is an out-of-band full-duplex design: the relay transmits at a
signal whose frequency is different from the one it receives
at. For example, on the downlink channel, it downconverts
with the reader’s center frequency but upconverts with
another frequency, effectively achieving frequency division
between the reader-relay half-link and the relay-RFID
half-link.
bandpass filter only allows the RFID tag’s response to pass
on the uplink.
The relay is powered by the drone's battery and has a total
power consumption of 5.8 Watts, drawing 0.49 Amps from
the battery. Since the battery is designed to support up to
21.6 Amps, the relay consumes less than 3% of the drone’s
battery.
VI. RESULTS
Several experimental locations were used to evaluate
RFly’s accuracy of localizing RFIDs. The experiments
resulted in zero read rate at a range of 10 meters in the
absence of a relay, while the read rate stays at 100% even if
the relay is at more than 50m in line-of-sight. In
non-line-of-sight, the read rate drops to 75% at a distance of
55m.

VII. LIMITATIONS
IV. LOCALIZATION ALGORITHM The current prototype has few limitations that are left for
RFLy uses the fact that the drone’s movement emulates future work:
an antenna array and applies standard array equations to
(1) The current evaluation relies on vision-based systems for
perform localization.
navigation, obstacle avoidance, and stability. It is desirable
For simplicity, the 2D space localization is presented. to enable the drone to operate in more challenging
Intuitively, this formulation leverages the fact that every environments.
point (x,y) in 2D space can be described by a set of
(2) The current system’s range is limited to 55m due to the
distances from the different points along the drone’s
extent of self-interference cancellation. To cover a larger
trajectory. Hence, to localize, it applies a matched filter on
area, one could improve the design of cancellation circuits
all possible locations and chooses the highest peak.
and algorithms.
RFly’s estimated location of the RFID is very accurate
(3) RFly currently operates with a single drone. It is
relative to the RFID’s actual location, with an error less than
desirable to extend the system’s operation to swarms of
7 cm.
drones. Future research could explore designs that
Aside from the ability to accurately localize RFIDs, daisychain the drone-based relays to extend the range of
RFly’s approach of leveraging drones does not suffer from operation.
blind spots which are typically very challenging with
Despite these limitations, RFly marks an important step
stationary RFID readers (due to destructive interference or
toward enabling long-range communication and localization
orientation mismatch). This is because the drone can capture
in battery-free networks.
each RFID’s response from different perspectives

V. IMPLEMENTATION
The overall system consists of RFly’s relay mounted on a
low-cost commercial drone ($500), an RFID reader and
off-the-shelf low-cost passive RFID tags ($0.12).
The RFly's relay was built into a custom designed PCB
(10cm x 7.5cm, weight 35g) and consists of two paths: the
downlink forwarding path and the uplink forwarding path.
Each path employs two RF mixers. The first mixer
downconverts the signal to baseband while the second mixer
upconverts the baseband signal back to passband.
The relay eliminates inter-link interference through
baseband filtering: a low-pass filter with cut-off frequency at
100kHz and a bandpass filter with center frequency at
500kHz are implemented in the downlink and uplink
respectively. As a result, the baseband filter only allows the
reader’s query command to pass on the downlink; and, the