NOMA Testbed on Wi-Fi

Abstract

Non-Orthogonal Multiple Access (NOMA) is currently considering as a promising technology for increasing the network efficiency in heterogeneous scenarios with different types of traffic and interference conditions. In spite of numerous theoretical works, there are few experimental studies of NOMA, most of which consider cellular networks. In this short paper, we present our pioneering NOMA testbed for Wi-Fi networks. Because the Wi-Fi technology is much simpler and cheaper than the cellular ones, we expect that the testbed can be used by the research community to validate various solutions designed for NOMA in real equipment.

Full text

NOMA Testbed on Wi-Fi
Evgeny Khorov*†, Aleksey Kureev*† , Ilya Levitsky†
*National Research University Higher School of Economics, Moscow, Russia
†Institute for Information Transmission Problems, Russian Academy of Sciences, Moscow, Russia
Email: {khorov, kureev, levitsky}@iitp.ru
©2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future
media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or
redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. DOI: 10.1109/PIMRC.2018.8580931
Abstract—Non-Orthogonal Multiple Access (NOMA) is
currently considering as a promising technology for in-
creasing the network efficiency in heterogeneous scenarios
with different types of traffic and interference conditions.
In spite of numerous theoretical works, there are few
experimental studies of NOMA, most of which consider
cellular networks. In this short paper, we present our
pioneering NOMA testbed for Wi-Fi networks. Because
the Wi-Fi technology is much simpler and cheaper than
the cellular ones, we expect that the testbed can be used
by the research community to validate various solutions
designed for NOMA in real equipment.
Index Terms—Non-Orthogonal Multiple Access
(NOMA), IEEE 802.11, NOMA performance, Successive
Interference Cancellation, Experimental Testbed, USRP
I. Introduction
Non-Orthogonal Multiple Access (NOMA) is cur-
rently considering as a promising technology for increas-
ing the network efficiency in heterogeneous scenarios
with different types of traffic and interference conditions.
Many theoretical works prove its efficiency for multiuser
uplink and downlink transmission, design resource and
power allocation algorithms, modulation and coding
schemes which in theory should be much more efficient
than the traditional ones (e.g. see [1]). At the same
time, implementation of NOMA in real equipment is a
challenging task. A solution which perfectly works under
some theoretical assumptions may be inefficient in prac-
tice. To validate and evaluate various developed solutions
it is necessary to design prototypes and testbeds. To the
best of our knowledge, there are few NOMA testbeds
described in literature. Most of them are designed for
cellular technologies and the corresponding papers have
been published very recently. For example, paper [2]
presents a NOMA-over-LTE testbed implemented on
outdated Universal Software Radio Peripheral (USRP)
B210. Authors use it for the downlink transmission in
5 and 10 MHz channels. Because of the limitations of
the used hardware they were needed to conduct many
real-time computations on CPU, and they could not
achieve any performance gain with respect to the legacy
operation. Experimental study for the uplink scenario is
considered is paper [3]. The authors propose a channel
estimation and synchronization techniques for the uplink
NOMA receiver and build an experimental testbed using
aforementioned USRP B210. With the testbed, they
The work was carried out at NRU HSE and supported by the Russian
Science Foundation (agreement 18-19-00580)
Fig. 1: TX Scheme with Superposition Coding
evaluated the performance of NOMA with Turbo codes
and SC-FDMA with BPSK modulation.
In this paper, we present a novel experimental NOMA
testbed. To the best of our knowledge, this is the first
testbed which implements NOMA over Wi-Fi.
II. Implementation and results
In our testbed, we use USRP-2944R [4], which allows
data transmission and processing in 20 MHz band with
the potential increase up to 160 MHz used in IEEE
802.11ac. It has rather powerful FPGA inside, which
allows running several instances of Wi-Fi PHY and MAC
at the same time. Thus, in our testbed, we do not need
to use CPUs for real-time computation such as signal
processing and decoding/encoding.
To implement NOMA functionality on USRP, we use
802.11 Application Framework, which contains a simpli-
fied version of the Wi-Fi protocol. At the first stage of the
project, we consider downlink power-domain NOMA as
described in [1]. To implement it, we need to design
a Wi-Fi-based NOMA transceiver with Superposition
Coding (SC) and Successive Interference Cancellation
(SIC) that can interact not only with the same devices
but with traditional off-the-shelf devices.
The simplified scheme of the transmitter with SC is
shown in Fig. 1. Two independent applications generate
data streams on the Host PC and put all data in DMA
FIFO queues. FPGA gets data from DMA FIFO queues
and processes them in the MPDU Generation Blocks
where data are wrapped up into MPDU. As data process-
ing in the upper and the lower branches of the scheme
may take different time, we make synchronization right
after MPDU generation block. After that, in TX IQ
Processing Block, we construct a packet with common
L-STF and L-LTF fields according to Wi-Fi standard

Fig. 2: RX Scheme with Successive Interference Can-
cellation
and the weighted sum of IQ samples corresponding
to both streams. For simplicity, the weight of the first
stream is one, while the weight of the second one
is defined by the configurable online scaling factor.
The scaling factor defines the amplitude ratio of the
weaker and the stronger frames. To help the receiver to
understand which scaling factor is used for transmission,
we encode this information in the last 6 bits of the L-
SIG field of the weaker frame without modifying the
L-SIG of the stronger frame which can be received by
the legacy devices. After summarizing, we apply Fast
Fourier Transform (FFT) and then the digital signal goes
to the Radio Frequency (RF) module.
Fig. 2 shows the receiver part of our prototype with
SIC. After synchronization the data goes to RX IQ
Processing Block where IFFT and channel equalization
are calculated. After that, we buffer the received samples
in FPGA and process the strongest signal in the upper
RX Bit Processing and MAC RX Blocks. The user which
is intended receiver of this data stops in this stage and
transmits the received data from MAC RX to the Host
PC. Other users rebuild IQ samples from the higher
branch and subtract them from the stored samples. Then
the resulting samples go to the lower branch of the
scheme to the Bit Processing and MAC RX Blocks and
then the received data are transmitted to the Host PC.
Note that the legacy devices can be intended receivers
of the data, corresponding to the higher branch since
receiving these data does not need implemented SIC
functionality.
To validate our testbed, we carried out a set of ex-
periments in which one device transmits pairs of frames
with NOMA while another one receives it. We set the
TX power of the strongest frame 5db higher than the
level when more than 95% of frames are successfully
received at MCS3. Then we vary the scaling factor,
which determines the power of the weaker frame and
measure Frame Reception Ratio (FRR) for the weaker
frame for different Modulation and Coding Schemes
(MCS) for the weaker and the stronger frames. Fig. 3
shows how the FRR depends on the Signal to Noise
Fig. 3: FRR from SNR of subtracted sample.
Ratio (SNR) of the weaker frame for different pairs of
MCSs (the first value in the legend corresponds to the
MCS of the stronger frame). While calculating SNR of
the weaker frame, we substract the receive power of the
strongest one.
III. Conclusion
In this paper, we have designed a NOMA-over-Wi-
Fi testbed compatible with legacy devices, which can
be used to validate and evaluate the performance of
numerous NOMA-related approaches improving in the-
ory the performance of wireless networks. In contrast to
cellular technology, Wi-Fi is much simpler and cheaper.
So it is much easier to use it for academic and in-
dustrial research. Naturally, Wi-Fi uses random access
and operates in unlicensed spectrum. So, from one side,
while carrying out experiments, it is necessary to control
interference and establish tight synchronization needed
for NOMA. From the other side, the compatibility of
the designed testbed with numerous cheap off-the-shelf
devices makes it possible to evaluate the performance
of NOMA in dense networks with huge interference.
Moreover, NOMA can follow MU-MIMO and OFDMA
that have already been included in the Wi-Fi technology.
Thus the designed testbed is useful for the development
of future Wi-Fi.
References
[1] S. R. Islam, N. Avazov, O. A. Dobre, and K.-S. Kwak, “Power-
domain non-orthogonal multiple access (noma) in 5g systems:
Potentials and challenges,” IEEE Communications Surveys &
Tutorials, vol. 19, no. 2, pp. 721–742, 2017.
[2] X. Wei, Z. Geng, H. Liu, K. Zheng, and R. Xu, “A portable
sdr non-orthogonal multiple access testbed for 5g networks,” in
Vehicular Technology Conference (VTC Spring), 2017 IEEE 85th.
IEEE, 2017, pp. 1–5.
[3] S. Abeywickrama, L. Liu, Y. Chi, and C. Yuen, “Over-the-air im-
plementation of uplink noma,” arXiv preprint arXiv:1801.05541,
2018.
[4] DEVICE SPECIFICATIONS NI USRP-2944R, National Instru-
ments, 2016.