Secure optical communication enabled by a
wavelength-division-multiplexed quantum alarm
Amir Minoofar
Cisco Quantum Lab
Santa Monica, USA
aminoofa@cisco.com
Jiapeng Zhao
∗
Cisco Quantum Lab
Santa Monica, USA
penzhao2@cisco.com
Michael Kilzer
Cisco Quantum Lab
Santa Monica, California, USA
mkilzer@cisco.com
Eneet Kaur
Cisco Quantum Lab
Boston, USA
ekaur@cisco.com
Ramana Rao Kompella
Cisco Systems
San Jose, USA
rkompell@cisco.com
Reza Nejabati
CISCO Quantum Lab
Santa Monica, California, USA
rnejabat@cisco.com
Abstract
Quantum
technologies
are
explored
to
improve
the
security
of
classical communications and networking. Here, we propose and
experimentally demonstrate a quantum-assisted classical secure
communication system, which provides real-time eavesdropping
detection. We randomly multiplex the classical and quantum signal
in the wavelength domain based on the output of a quantum random
number generator, and monitor the quantum entanglement to detect
the presence of eavesdropping. Compared to previous studies, our
approach preserves the same level of security without sacrificing
the classical data rate.
Keywords
Secure communication, quantum security, DWDM, quantum ran-
dom number generator
ACM Reference Format:
Amir
Minoofar,
Jiapeng
Zhao,
Michael
Kilzer,
Eneet
Kaur,
Ramana
Rao
Kompella, and Reza Nejabati. 2025. Secure optical communication enabled
by a wavelength-division-multiplexed quantum alarm. In
Proceedings of the
2025 Workshop on Quantum-Resistant Cryptography and Security (QRSEC
’25), October 13–17, 2025, Taipei, Taiwan.
ACM, New York, NY, USA, 5 pages.
https://doi.org/10.1145/3733820.3764677
1
Introduction
Data security has become critical in the big data era especially for
optical communications. To protect optical communication against
eavesdropping, classical encryptions are implemented, which typi-
cally rely on mathematical complexity. However, these encryptions
can be deciphered via quantum computers in real time, leading to
potential security loopholes [3].
Quantum communication, or quantum cryptography, has been
proposed and implemented in the past couple of decades to provide
unconditional security. Due to the noncloning theorem, quantum
Permission to make digital or hard copies of all or part 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 components of this work owned by others than the
author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or
republish, to post on servers or to redistribute to lists, requires prior specific permission
and/or a fee. Request permissions from permissions@acm.org.
QRSEC ’25, Taipei, Taiwan
© 2025 Copyright held by the owner/author(s). Publication rights licensed to ACM.
ACM ISBN 979-8-4007-1908-0/25/10
https://doi.org/10.1145/3733820.3764677
information
cannot
be
deciphered
without
disturbing
the
quan-
tum entanglement. Therefore, any attacks by the eavesdropper will
break the entanglement, and by monitoring the quantum entangle-
ment, legitimate users can guarantee that the communication is
not being attacked. However, the inherent rate-distance trade-off
significantly limits the performance of quantum communication,
which cannot satisfy the demand for high-rate data exchange today
[6].
In practical applications, unconditional security is not always
necessary. Therefore, it is possible to use quantum photons to mon-
itor the channel security and use the classical light streams to trans-
fer
data
in
the
same
physical
layer.
Such
a
quantum
alarm
(QA)
system has been demonstrated in optical fibers via randomly time
multiplexing of a classical signal and a quantum beacon signal [
4
,
5
].
Users first prepare entangled photon pairs and randomly multiplex
one
of
the
photons
with
the
classical
data
stream
in
time.
Since
only authorized users have knowledge of the location of quantum
signals,
the
eavesdropper
cannot
identify
whether
the
zeros
are
quantum signal or classical zeros. The best strategy an eavesdropper
can take is to tap the entire communication link without bias, and
resend new zeros. Since the zeros sent by the eavesdropper have
no quantum correlation with the other photon in the entangled
pair, this disturbance will break the quantum entanglement, which
can be immediately noticed by users through real-time monitoring
of the quantum state. Hence, legitimate users can detect the pres-
ence of the eavesdropper by monitoring the entanglement of the
quantum beacon.
However, due to the nature of time division multiplexing, QA
systems still require the classical channel to be turned off, which
is
not
only
limited
in
data
rate,
but
is
also
not
compatible
with
industrial standard communication protocols [
4
,
5
]. To solve these
two problems, we propose a dense wavelength division multiplex-
ing (DWDM) based quantum alarm, which multiplex the classical
and quantum signal, and randomly switch the wavelength of both
signals
to
different
DWDM
ITU
channels.
The
random
channel
selection sequence is generated via a Cisco quantum random num-
ber generator (QRNG), which guarantees that no bias exists during
the communication. Our approach aligns well with the industrial
standard DWDM protocol, and provides enhanced security without
sacrificing classical data rates.
43
QRSEC ’25, October 13–17, 2025, Taipei, Taiwan
Amir Minoofar, Jiapeng Zhao, Michael Kilzer, Eneet Kaur, Ramana Rao Kompella, and Reza Nejabati
Figure 1: Experimental setup of quantum alarm. PC: polarization controller; FBG: fiber Bragg grating; PBS: polarization beam
splitter.
2
Experimental Setup
The schematic of the experimental system is shown in Fig. 1. The
system consists of an entanglement source, QRNG, and Cisco small
form factor pluggable transceivers (SFPs). We first introduce the
functionality of each subsystem, and then illustrate the concept of
our DWDM-based QA.
2.1
Micro-ring Resonator Entanglement Source
The entanglement source is a critical device in quantum applica-
tions. Our system generates time-energy entanglement from a quan-
tum photonic micro-ring resonator based on AlGaAs on-isolator
(AlGaAsOI). Compared to other on-chip entanglement source, Al-
GaAsOI
has
a
strong
𝜒
(
3
)
nonlinearity,
and
a
low
on-chip
loss,
resulting in a high quality factor over 3
×
10
6
[
7
]. The nonlinear
process for the generation of entanglement pairs is spontaneous
four-wave
mixing
(SFWM),
in
which
two
pump
photons
(wave-
length
𝜆
𝑝
) are annihilated to create one signal photon (wavelength
𝜆
𝑠
) and one idler photon (wavelength
𝜆
𝑖
). The free spectral range
(FSR) of the ring is carefully designed so that all three wavelengths
are on standard DWDM ITU grids. Therefore, this entanglement
source
is
naturally
compatible
with
industrial
standard
DWDM
infrastructure. The chip is designed by Cisco and manufactured by
UC Santa Barbara.
A tunable continuous-wave (CW) laser is used to set the wave-
length
and
power
of
the
input
light
coupled
to
the
photonic
in-
tegrated
circuit
(PIC)
entanglement
source.
To
reduce
amplified
spontaneous
emission
(ASE)
noise
from
the
input
CW
laser,
we
cascade two narrow fiber Bragg grating (FBG) filters with 30 dB
isolation and a full width at half maximum (FWHM) of 0.17 nm. A
polarization controller (PC) and a polarization beam splitter (PBS)
are used to maximize the coupling efficiency into the PIC and to en-
sure the polarization matches the transverse electric (TE) mode of
the PIC edge couplers. The temperature of the ring resonator is set
to 23
◦
C using a temperature controller. Two monitoring couplers
are used to identify the optimal resonance frequency of the PIC. The
first monitoring coupler is placed before the PIC, and the monitor-
ing tap is coupled to a delay-line interferometer (DLI) with a fixed
delay. The output is coupled to a photodiode (PD) and measured
by
an
electrical
oscilloscope
(EOS),
which
assists
in
fine-tuning
the frequency of the tunable laser source. The second monitoring
coupler is placed after the PIC, and the monitoring tap is coupled to
a PD and EOS to characterize the microring performance, measure
the FSR, and find the optimal resonance frequency, as shown in
Fig. 2.
We first characterize the resonance of our source as shown in
Fig. 2. There are 8 micro-ring resonators on the same chip: 4 with
FSR equal to 100 GHz, 2 with FSR equal to 200 GHz, and 2 with
FSR equal to 400 GHz. Resonances are measured with an arbitrary
unit. By pumping the ring resonator at the wavelength in the red
resonance box, we can generate entanglement photons in the signal
and idler box. In order to reduce the noise photon from the classical
signal to the quantum signal, the spectral separation between signal
and idler needs to be large enough. Therefore, we choose the 400
GHz FSR ring as shown in the bottom. With an on-chip power of
9.33 uW, we can measure an off-chip entanglement rate at about 29
K counts per second (cps), which can be converted to an on-chip
pair rate at about 1.32 M cps.
2.2
Quantum Random Number Generator
(QRNG)
QRNG
is
a
promising
candidate
for
true
unpredictable
random
number generation based on the inherent random nature of quan-
tum
mechanics.
In
quantum
optics,
quantum
vacuum
noise
can
be measured via homodyne detection that either detects the
𝑄
or
𝐼
quadrature
of
the
vacuum
state
|
0
⟩
,
and
this
noise
follows
the
Gaussian distribution. This Gaussian distributed random variable
becomes the quantum entropy source in Cisco QRNG. The Gaussian
raw bits from this vacuum noise measurement contain noises in-
duced by various classical correlations. In order to obtain quantum
44
Secure optical communication enabled by a wavelength-division-multiplexed quantum alarm
QRSEC ’25, October 13–17, 2025, Taipei, Taiwan
Figure 2: Cisco on-chip entanglement source (top) and the resonance measurement of the 7th ring (bottom).
secure random numbers, quantum random number extractors are
implemented, which generate true quantum random numbers at
the end.
The experimental realization of Cisco QRNG is shown in Fig. 3.
A fiber-coupled CW 1550nm laser (from INPHENIX) is split, by a
50:50 beam splitter (BS1), into two copies, which work as the local
oscillator (LO) for
𝐼
and
𝑄
quadrature measurements. The phase of
the LO laser in the lower arm is delayed by
𝜋
/
2 for
𝑄
measurement,
while the phase of the LO in the upper arm remains the same for
𝐼
measurement. The quantum vacuum state interferes with the LO
laser at BS2 and BS3 correspondingly, and the signals are measured
by two homodyne detectors (HOM1 and HOM2).
The high-speed realization of our concept is achieved by an 1.6
GHz bandwidth integrated homodyne detector module (Optoplex
90 degree hybrid with balanced receiver) and an AMD RFSoC de-
velopment board (ZCU111). To measure quantum vacuum noise,
the homodyne detector is operated in the shot noise limit regime.
After the detector, the analog signals are passed to the amplifiers
to bring the analog voltage range as close to the full-scale range of
the ADCs in the ZCU111 as possible [
1
,
2
]. The amplified signals are
then filtered with low-pass anti-aliasing filters. The amplified and
filtered voltage outputs are sampled by two ADCs on the RFSoC on
the ZCU111 development board at a rate of 2 G samples per second.
We implemented a two-stage randomness extraction strategy
in
our
QRNG
to
balance
both
rigorousness
and
speed.
A
strong
secure seed is extracted via Dodis extractor, and the seed is used for
high-speed randomness extraction via a Toeplitz extractor. Toeplitz
randomness extraction block is implemented with an extraction
efficiency of 66.67%, which produces a uniform random number
rate of 42.67 Gbps. The secure random bits have successfully passed
NIST and Dieharder tests.
2.3
Small Form-Factor Pluggable (SFP)
Transceivers
To
emulate
the
classical
communication
link,
we
use
two
small-
form-factor pluggable (SFP) modules (DP01QS28-E20) from Cisco.
One
SFP
serves
as
the
transmitter
and
the
other
as
the
receiver.
These SFPs operate at a 100 Gb/s line rate and allow wavelength
tunability across the C band (1530–1565 nm), while maintaining a
transmitted optical power of approximately -8 dBm and a receiver
sensitivity of approximately -30 dBm. These SFPs enable adding
or
upgrading
high-speed
links
on
a
network
chassis
(NC)
with-
out changing the entire setup, simply by inserting the appropriate
module into the corresponding port. Their main advantages are
45
QRSEC ’25, October 13–17, 2025, Taipei, Taiwan
Amir Minoofar, Jiapeng Zhao, Michael Kilzer, Eneet Kaur, Ramana Rao Kompella, and Reza Nejabati
Figure 3: Experimental configuration of Cisco QRNG.
modularity and improved link performance when on-board pro-
cessing is available. We use Cisco Network Convergence System
(NCS)
modules
(N540X-16Z4G8Q2C-A)
to
operate
the
SFPs.
To-
gether, these devices perform digital signal processing (DSP) and
clock recovery and implement stronger forward error correction
(FEC). This improves receiver sensitivity and allows the link to tol-
erate more loss and noise, enabling longer reach and more reliable
100 Gb/s operation. The SFPs are connected through a 10 km optical
fiber link FCA-SMF28-ULL with a total insertion loss of 1.9 dB.
2.4
DWDM-Based Quantum Alarm
Figure 4: Wavelength division multiplexing of classcial and
quantum channels based on QRNG output.
Quantum alarm can be demonstrated using different multiplex-
ing schemes. Among these, we multiplex in the wavelength domain
and
select
standard
100
GHz
DWDM
ITU-grid
channels
in
the
C-band for the classical and quantum channels. We choose these
channels based on the QRNG output, as shown in Fig. 4. We quantize
the QRNG output: a 2-bit random number generates four different
states. Four ITU channels, L1, L2, R1, and R2, are predefined, and
the QRNG output specifies the locations of the classical and quan-
tum channels at different time instances. To minimize the effect of
Raman scattering and crosstalk between the classical and quantum
channels, we ensure that the wavelength separation between the
left channels (L1 or L2) and the right channels (R1 or R2) is at least
20 nm. The first bit specifies the quantum channel: if it is one, the
quantum channel is placed on the right side at R2; if it is zero, it is
placed on the left side at L2. The second bit determines the location
of the classical channel: if it is one, it is placed on L2 or R2; if it
is
zero,
it
is
placed
on
L1
or
R1.
To
enable
this
scheme,
we
use
tunable SFPs and 1
×
2 optical switches to redirect the classical and
quantum channels at different time instances. The first QRNG bit
governs the switch direction, based on which quantum channel is
selected. The second QRNG bit tunes the wavelength of the SFPs
for the classical transmission. In this experimental demonstration,
the
entanglement
source
is
used
for
the
generation
of
quantum
signals. We herald either the signal or the idler depending on the
QRNG
output
at
different
time
instances,
and
we
consider
only
single-photon transmission for the quantum channel.
The security analysis of our quantum alarm scheme is consistent
with the conventional time-multiplexed quantum alarm. When an
eavesdropper attacks the channel, she cannot identify whether a
classical or quantum signal is in the current channel. Similarly in
the time domain case, the best strategy she can take is to measure
46
Secure optical communication enabled by a wavelength-division-multiplexed quantum alarm
QRSEC ’25, October 13–17, 2025, Taipei, Taiwan
the signal without bias, which will inevitably introduce decoher-
ence to the quantum entanglement. The disturbance in the quantum
entanglement can be immediately measured by Alice and Bob, lead-
ing to the awareness of existence of eavesdropper. Therefore, in our
current paper, we mainly focus on the improvement of the physical
system to gain a better data rate for classical channels.
Acknowledgments
We appreciate the support from Gary Wang at Optoplex Corpora-
tion for his support of the balanced detection module.
References
[1]
Cédric Bruynsteen, Tobias Gehring, Cosmo Lupo, Johan Bauwelinck, and Xin Yin.
2023.
100-Gbit/s integrated quantum random number generator based on vacuum
fluctuations.
PRX quantum
4, 1 (2023), 010330.
[2]
Tobias Gehring, Cosmo Lupo, Arne Kordts, Dino Solar Nikolic, Nitin Jain, Tobias
Rydberg,
Thomas
B
Pedersen,
Stefano
Pirandola,
and
Ulrik
L
Andersen.
2021.
Homodyne-based quantum random number generator at 2.9 Gbps secure against
quantum side-information.
Nature Communications
12, 1 (2021), 605.
[3]
Craig Gidney and Martin Ekerå. 2021.
How to factor 2048 bit RSA integers in 8
hours using 20 million noisy qubits.
Quantum
5 (2021), 433.
[4]
Yupeng Gong, Rupesh Kumar, Adrian Wonfor, Shengjun Ren, Richard V Penty,
and Ian H White. 2020.
Secure optical communication using a quantum alarm.
Light: Science & Applications
9, 1 (2020), 170.
[5]
Yupeng Gong, Adrian Wonfor, Jeffrey H Hunt, Ian H White, and Richard V Penty.
2021.
Experimental demonstration of confidential communication with quantum
security monitoring.
Scientific Reports
11, 1 (2021), 21686.
[6]
Marco Lucamarini, Zhiliang L Yuan, James F Dynes, and Andrew J Shields. 2018.
Overcoming the rate–distance limit of quantum key distribution without quantum
repeaters.
Nature
557, 7705 (2018), 400–403.
[7]
Trevor J Steiner, Joshua E Castro, Lin Chang, Quynh Dang, Weiqiang Xie, Justin
Norman, John E Bowers, and Galan Moody. 2021.
Ultrabright entangled-photon-
pair generation from an al ga as-on-insulator microring resonator.
Prx Quantum
2, 1 (2021), 010337.
47
