An Empirical Analysis of IEEE 802.

11ax
Siraj Muhammad, Jiamiao Zhao, and Hazem H. Refai
Electrical and Computer Engineering
University of Oklahoma
Tulsa, OK, USA
{sirajmuhammad, jiamiaoz, hazem}@ou.edu

Abstract—An empirical analysis of the newly released standard latency, extremely high bandwidth, and massive density. In the
IEEE 802.11ax, widely known as Wi-Fi 6, is presented in this licensed spectrum domain, 5G New Radio (NR) is the answer
paper. Several tests were conducted to evaluate key performance to new requirements. Device-to-Device (D2D) communication
metrics, including throughput and jitter as a function of net-
work parameters (e.g., packet size and window size), as well is regarded as a potential LTE technology to address spectrum
as environment variables (e.g., SNR). Empirical models were scarcity in ultra-dense heterogeneous networks [5]. To address
developed using collected results to quantify the behavior of said demands in the unlicensed spectrum, domains such as Free-
metrics. Channel utilization of the system was also investigated Space Optical (FSO) communication are going through active
and compared to its precedent, 802.11ac. research to supplement high bit rate multi-user links [6].
Index Terms—Wi-Fi, 802.11ax, WLAN, throughput, jitter,
channel utilization To this end, the IEEE standardization body introduced a
new member to its 802.11 family, namely “802.11ax High-
Efficiency,” or Wi-Fi 6.
I. I NTRODUCTION
Khorov, et. al [7] described key features of the latest Wi-
Wireless connectivity is essential today and, in the future, Fi breakthroughs with particular focus on the draft D3.0 of
as it facilitates unplugging excessive wires from conventional the 802.11ax standard, which was released in May 2018.
instruments, adds mobility to others, and transforms appliances Orthogonal Frequency Division Multiple Access (OFDMA)
into smart gadgets. From smart phones to smart TVs, medical is the cornerstone advancement of the standard and is aimed
devices to connected cars and autonomous vehicles, and the to address the throughput bottleneck at the Medium Access
revolution of the internet of things (IoT), each wireless device Control (MAC) layer.
depends on its ability to reliably and quickly communi- Qu [8] analyzed the way in which OFDMA coordinated
cate massive amounts of data from one point to another. access point and devices, ensuring a node can simultaneously
For the past two decades, wireless data transport has been transmit or receive. The researchers developed a MAC proto-
achieved largely by the 802.11 standard family, specifically col, including new physical layer sensing, fast back-off pro-
the 802.11a/b/g/n variants. This standard has matured over cess, enhanced RTS/CTS mechanism, and frame structures, as
the years, improving achieved throughput, enhancing channel well as a mathematical model to predict maximum throughput
access schemes, and accommodating an increasing number by leveraging the new MAC protocol. Results proved that
of users. While these improvements have kept current with maximum throughput increased by 160%.
demand, society is being ever more inundated by a massive Naik [9] studied the uplink multi-user (MU) OFDMA and
number of connected devices. derived an analytical model using Bianchi’s Markov Chain
According to statistical studies, the number of connected to characterize system performance at the MAC layer. The
devices has been exponentially increasing for the past 10 authors also introduced a new metric, namely BSR (Buffer
years. Aruba Networks reported that 50% of all internet traffic Status Rate) delivery rate, and described the tradeoff between
was carried by Wi-Fi technology in 2018 [1]. The remaining offered network throughput and the capability to support new
traffic has been supported by LTE. According to an Ericsson users. Analysis were validated through simulations.
report, traffic generated by smartphones in 2022 is expected Some researchers [7] have detailed a number of challenges
to increase 10 times the amount reported in 2016 [2]. This facing 802.11ax implementation (e.g., OFDMA scheduler, dy-
translates into more than 60 exabytes of data per month. This namic adjustment of sensitivity threshold, and energy savings
extensive amount of data flow is attributed mainly to new as an optimization problem between energy consumption and
use cases of wireless connectivity projected for the next five throughput).
years (e.g., online gaming, virtual reality (VR), critical services In [10], Bellalta provided a review of expected applications
and infrastructure control, sensor networks, and smart trans- and scenarios that call for a new amendment to existing
portation). The automotive domain has experienced a surge in standards, such as 802.11n and 802.11ac. A number of studies
WLAN-equipped vehicles, and researchers have investigated have also evaluated those two protocols from a systems level,
the feasibility of Wi-Fi access points in such scenarios [3], providing insights on their physical-layer and MAC-layer
[4]. Such applications rely on three main characteristics of efficiency [11], [12].
next generation wireless communications, namely ultra-low- A Markov chain model was developed in [13] for estimating

Accepted for publication in proceedings of the International Conference on Communications, Signal Processing and their Applications 2020
energy efficiency when increasing the contention window size B. Trigger Frame and Multi-User Transmission
for an 802.11ax node. Researchers also compared energy effi- 802.11ax operates in either single- or multi-user fashion.
ciency relative to the number of spatial streams for transmitting The earlier 802.11ac standard previously supported downlink
frames. This study reported that leveraging MU-MIMO (multi- multi-user transmission (DL MU), leveraging multiple-input
user multiple-input multiple-output) optimized efficiency with multiple-output (MIMO) technology. 802.11ax supplements
four bandwidths offered by 802.11ax. Finally, researchers in functionality by multiplexing users in the uplink, as well.
[14] examined using 802.11ax for IoT, assuming that it is able Uplink multi-user transmission (UL MU) is realized via the
to resolve energy efficiency and range problems. inherited multi-user MIMO (MU-MIMO) in 802.11ac and
The work presented in this paper focuses on application borrowing from the technological advancement of cellular
layer performance (e.g., throughput, delay, and packet loss) technology in OFDMA. In both methods, the access point
of 802.11ax under various conditions and provides empirical (AP) controls and orchestrates transmissions among stations
models of network performance for future deployments. The within the network. Notably, MU-MIMO and OFDMA can be
balance of this paper is organized as follows. Section II combined in an 802.11ax network.
provides an overview of key features and enhancements of The main challenge of the new standard is embodied in
802.11ax. Section III describes the experimental setup and the UL MU transmission, namely synchronization among
software used for testing. Results and discussion are presented different stations when sending data simultaneously to the AP,
in Section IV, and the paper concludes in Section V. as their clocks might drift as a result of jitter. Fortunately,
II. IEEE 802.11 AX AT A G LANCE the specification defines a new type of control frame, namely
Trigger Frames, that AP sends to all users. This trigger frame
The new 802.11ax specification introduces significant initiates uplink transmission for all users for sending data or a
changes to the physical and MAC layers of the protocol. A response to multi-user block acknowledgment request (MU-
key change is the introduction of an Orthogonal Frequency- BAR). Trigger frames include information about upcoming
Division Multiple Access (OFDMA) approach, which is uplink transmission, such as duration, GI (identical for all
built on top of legacy CSMA/CA (Carrier Sense Multiple participating UL MU transmission [16]) OFDMA resource
Access with Collision Avoidance). The standard was de- allocation of RUs, per-station parameters (e.g., MCS, TX
signed with a careful consideration for legacy specifications, power, and others), and number of spatial streams.
802.11a/b/g/n/ac that although challenging, it allows backward
compatibility with earlier versions. C. OFDMA Random Access
Since AP controls and allocates RUs for stations transmit-
A. Physical Layer
ting in UL MU mode, there are occasions in which AP might
The 802.11ax amendment allows access to the 2.4 GHz and be unaware of an associated station with data to send or an
5.0 GHz bands; it also supports channel bandwidths up to 40 unassociated station desiring to join the Basic Service Set
MHz at 2.4 GHz and up to 160 MHz at 5.0 GHz. By virtue (BSS). Such short packet UL frames must be accounted for
of OFDMA, the channel subcarriers are grouped into units in a UL MU transmission. Because Distributed Coordination
in time and frequency domains (i.e., resource units [RU]). Function (DCF) scheme is otherwise inefficient and costly in
This approach facilitates simultaneous communication with terms of overhead, the 802.11ax specification was designed
multiple receivers by mapping various RUs to different stations to address this issue by leveraging the OFDMA Back-off
and avoids frequency selectivity by allocating a given station (OBO) procedure [17]. AP sends a special Trigger Frame,
bandwidth as low as 2.5 MHz or 26 subcarriers. When com- including information about random access RUs for stations
pared with 64 subcarriers supported by the earlier 802.11ac without allocated resource elements, in which to transmit and
standard, 802.11ax increases the number of subcarriers to 256, contend for once awarded. A given station selects a random
from which a maximum of 242 can be used. The remainder number in the range [0, CWO] where CWO is the OFDMA
are reserved to decrease inter-symbol interference (ISI) and contention window. Given that a station’s OBO value is less
leakage from adjacent tones. The increased number of tones than or equal to the number of random-access RUs declared
is accompanied by increased OFDM symbol duration to 12.8 in a Trigger Frame, one RU is randomly selected to transmit
µs and selectable Guard Interval (GI) of 0.8, 1.6, or 3.2 µs, its frame. Otherwise, the OBO value is decreased by the
which account for more robust outdoor operation. number of declared RUs in the Trigger Frame and awaits the
The new amendment also introduces a higher modulation subsequent one. Given that the transmission fails, the station
rate of 1024-QAM, in addition to those supported by 802.11ac; doubles its CWO and initializes its OBO value with a new
BPSK, 16-QAM, 64-QAM, and 256-QAM [15]. Moreover, random number from within the new range until CWO reaches
the updated standard adds support for higher forward error CWOmax . Following successful transmission, the station resets
correction rates of 1/2, 2/3, 3/4, and 5/6. The new physical its CWO to CWOmin .
layer offers an increased data rate of 25% from 802.11ac,
achieving 9.6 Gbps at a high Modulation and Coding Scheme D. BSS Coloring
(MCS) transmitted over 160 MHz or 80+80 MHz channel with Dense deployments are at the core of 802.11ax target
eight spatial streams and 0.8 µs GI. scenarios. Sites with highly dense APs can experience co-
channel interference from neighboring BSS. 802.11ax allows status. Netspot [22] and Acrylic WiFi [23] were utilized to
stations to determine whether a detected frame is originating assess the wireless connection and to collect measurements
from within its network or from its neighboring networks. By on RSSI and noise. MTR [24] is a network diagnosis tool
examining the BSS color bit in the frame header, a station that combines the functionalities of ‘traceroute’ and ‘ping’
can identify overlapping BSS (OBSS) and make decisions on and provides statistics on each route hop between host and
medium interference management. According to the amend- destination address, in addition to information about channel
ment, stations are allowed to adjust parameters related to state, connection state, and intermediate host responsiveness.
Clear Channel Assessment (CCA) procedure to differentiate
between intra- and inter-BSS frames without having to decode IV. 802.11 AX P ERFORMANCE A NALYSIS
the entire frame. Moreover, the standard defines two Network
Allocation Vectors (NAV) – one for the intra-BSS and another Results of conducted tests are presented and discussed in
for inter-BSS stations. Hence, NAVs from OBSS cannot affect this section. Measurements were repeated 10 times to ensure
NAVs from within a station’s BSS. Therefore, 802.11ax ex- repeatability and minimum mean square error (MMSE). It
hibits improved spatial reuse behavior and spectrum resource is important to note that results could vary given alternative
management. hardware implementation for the testing setup. That said, the
results reported herein should be regarded as representative
III. E MPIRICAL S ETUP of available commercial devices implementing 802.11ax stan-
dard.
An ASUS 802.11ax-enabled 4x4 MIMO router (i.e. access
point [AP]) [18] was used in conjunction with a laptop A. Throughput vs. SNR
equipped with an Intel AX200 2x2 MIMO wireless card [19]
supporting 802.11ax implementation. AP was connected to Throughput is expected to increase as bandwidth and SNR
another laptop over ethernet for generating traffic. The setup increase. This was confirmed and is demonstrated in Fig. 2
layout is depicted in Fig. 1. and Fig. 3, which show TCP protocol throughput in downlink
(Fig. 2) and uplink (Fig. 3).

Variable Distance
600
20 MHz
802.11ax Station 802.11ax 40 MHz
(Client) AP 500 80 MHz
160 MHz
Throughput [Mbps]

Ethernet
400

300

Server
200

Fig. 1. Experimentation setup.

100

Tests were carried out in an office setting to resemble a

0
real-world scenario. The station and AP were not connected 0 10 20 30 40 50 60 70
to the Internet in an effort to prevent background traffic and SNR [dB]
limit interference that could affect measurements.
Signal-to-Noise Ratio (SNR) was varied by changing the Fig. 2. Downlink throughput of 802.11ax node via TCP protocol.
separating distance between the station and the AP, since
transmission power control was not accessible in the AP. Although downlink and uplink throughputs reach almost
Furthermore, to avoid interference from surrounding devices, the same saturation limits using similar channel bandwidths,
all tests were conducted in the evenings or on weekends. the uplink traffic under 160 MHz channel bandwidth acts
Hence, it can be assumed the background noise was both differently. At 160 MHz, uplink traffic does not benefit from
minimal and constant throughout testing. doubling the channel bandwidth, and saturation throughput is
Various software tools were utilized for testing and data nearly the same for 80 MHz bandwidth, and sometimes lower.
collection (e.g., delay, jitter, throughput, and packet loss This can be attributed to differences in 802.11ax implemen-
rate, among others). iPerf3 [20] is a tool for investigating tation, since client and server utilize different hardware. The
IP networks and taking active measurements on maximum relationship between throughput and SNR for downlink and
achievable bandwidth. iPerf3 supports protocols, such as TCP, uplink traffic can be characterized using an exponential model:
UDP, and SCTP, with IPv4 and IPv6; it also aids in tuning
network timing parameters. Wireshark [21] monitored traffic θ = αeβx + γeδx (1)
 600 700
20 MHz 20 MHz
40 MHz 600 40 MHz
500 80 MHz 80 MHz
160 MHz 160 MHz
Throughput [Mbps]

Throughput [Mbps]
500
400
400
300
300
200
200

100 100

0 0
0 10 20 30 40 50 60 70 0 500 1000 1500 2000
SNR [dB] MSS [Bytes]

Fig. 3. Uplink throughput of 802.11ax node via TCP protocol. Fig. 4. TCP packet size effect on throughput.

TABLE I TABLE II
TCP DOWNLINK AND UPLINK MODELS COEFFICIENTS . M ODEL COEFFICIENTS OF THROUGHPUT VS . PACKET SEGMENT SIZE MSS.

α β γ δ R2 α β γ δ R2
20 MHz 1.365e+07 -0.01963 -1.365e+07 -0.01963 0.9815 20 MHz 264.6 -0.0001552 -151.2 -0.001554 0.9434
40 MHz 5.81e+06 -0.01767 -5.81e+06 -0.01767 0.9564 40 MHz 373.9 -0.0001047 -311 -0.003068 0.9486
DL

80 MHz -1.383e+07 -0.01662 1.383e+07 -0.01662 0.9545 80 MHz 787.6 -0.0003363 -760.3 -0.00168 0.9593
160 MHz -2.459e+07 -0.01557 2.459e+07 -0.01557 0.9314 160 MHz 1.472e+06 -0.0006497 -1.472e+06 -0.0006503 0.9569
20 MHz -9.112e+06 -0.01123 9.112e+06 -0.01123 0.9750
40 MHz 9.237e+06 -0.006891 -9.237e+06 -0.006892 0.9589
UL

80 MHz -2.654e+08 -0.00672 2.654e+08 -0.00672 0.9569

160 MHz -2.005e+07 -0.01111 2.005e+07 -0.01111 0.9462 relationship can be quantified using the following exponential
model:
where θ is the downlink or uplink throughput; x is SNR; and
θ = αeβm + γeδm (2)
α, β, γ, and δ are coefficients listed in Table I. Goodness-of-
fit (R2 ) is higher than 0.92 for those models, which indicates where θ is the throughput; m is the segment size; α, β, γ,
high correlation between measurement points. and δ are coefficients listed in Table II.
B. Throughput vs. Packet Size C. Throughput vs. Window Size
Data is transmitted via packet chunks, the largest being When a TCP connection is established between two nodes,
Maximum Transmission Unit (MTU). For most network in- namely a client and a server, both allocate a buffer to store
terfaces and systems, MTU has a default value of 1500 bytes. data for reception and transmission. Operating systems control
Data packets larger than the defined MTU must be fragmented buffer size (i.e. windows) in a dynamic fashion to optimize
into smaller units before transmission. TCP and IPv4 protocols network performance. A small window might underutilize the
encapsulate user data in their protocol-specific headers when connection bandwidth, while a large window can block a huge
packets reach their respective network layer. Therefore, a TCP amount of host memory, which, in turn, could degrade running
packet can carry up to 1460 bytes of user data. This is defined application performance. Systems define default, minimum-,
as TCP MSS (Maximum Segment Size), which is equal to and maximum-allowed window sizes and employ algorithms,
MTU minus 40 bytes and accounts for TCP and IPv4 frame such as CUBIC [25], to dynamically control the window
headers. In the testing scenario detailed in this paper, MSS was according to network congestion. iPerf3 allows modifying
varied on the server side, and the client was running Wireshark TCP connection window size. In the test scenario detailed
to validate received payload length. SNR was ensured to be herein, transmitter window/buffer size was increased from 10
at least 60 dB during testing to eliminate noise effect on bytes to 2 MB. Achieved throughput was measured while
measured throughput. Fig. 4 shows the relationship between receiver window size remained fixed. Measurements were
packet segment size and throughput. repeated for a fixed transmitter and a variable receiver window
As MSS increased, throughput increased until saturation size. Fig. 5 and Fig. 6 illustrate results for various channel
throughput was achieved. It is interesting that given a 160 bandwidths. SNR was set high to ensure noise did not play a
MHz channel, saturation throughput was reached when packet role in reported measurements.
size was at least 1200 bytes. However, for all other channel Fig. 5 and Fig. 6 demonstrate the effect on throughput when
bandwidths, critical MSS was approximately 800 bytes. This changing transmitter and receiver window size, respectively.
 against achieved throughput in TCP downlink transmission for
600 four channel bandwidths. As the channel bandwidth increased,
20 MHz
40 MHz the saturated throughput (indicated by maximum channel
500 80 MHz occupancy) increased as discussed earlier in Section IV-A.
160 MHz At a given throughput measurement, the utilization decreased
Throughput [Mbps]

400 across the increasing channel bandwidths. This is an expected

behavior since the bandwidth is doubled. Moreover, Fig. 7
300 demonstrates that 802.11ax exhibits better channel utilization
than 802.11ac. DC of 802.11ax was nearly the same for 160
200 MHz and 80 MHz channel bandwidths up until 450 Mbps
achieved throughput. After which, the utilization of 160 MHz
100 channel improves over that of 80 MHz channel until saturation
is achieved at 600 Mbps.
0
100 102 104 106
Window Size [Bytes] 100

Fig. 5. Throughput performance as a function of transmitter’s TCP window

size. 80

Duty Cycle [%]

60
1000
20 MHz
40 MHz
800 80 MHz 40
160 MHz 802.11ax (20 MHz)
Throughput [Mbps]

802.11ax (40 MHz)

600 20 802.11ax (80 MHz)
802.11ax (160 MHz)
802.11ac (20 MHz)
400 0
0 100 200 300 400 500 600
Throughput [Mbps]
200
Fig. 7. Duty cycle of 802.11ax (four channels) and 802.11ac.

0 E. Jitter vs. SNR

100 102 104 106
Jitter can be conceptualized as the difference in received
Window Size [Bytes]
packet delays. At the sending side, packets are transmitted in
Fig. 6. Throughput performance as a function of receiver’s TCP window size. a continuous stream with evenly distributed delays (i.e., time
gaps) between packets. Network congestion (or queuing delay)
in network hops causes the time difference between packets
Both cases indicate that throughput under four channel band- to vary. If jitter is too high in time-sensitive applications, such
widths remains under 50 Mbps until buffer size reaches 10 KB. as voice or video stream, the jitter buffer is made large to
Subsequently, throughput increased proportionally with win- compensate. Fig. 8 shows the relationship between receiver
dow size until reaching saturation of approximately 1 MB. It SNR and jitter measured using UDP link. The figure also
is worth noting that given a 1 MB receiver window, maximum indicates standard deviation for each measurement.
downlink throughput was 880 Mbps under a Linux operating Overall, jitter in four bandwidths decreased as SNR in-
system, while the same setting under a Windows operating creased. Two phenomena can be observed from the jitter-SNR
system was 500 Mbps. This finding could be attributed to relationship. First, jitter remains lower than 5 ms when SNR
different TCP congestion control mechanisms and parameters increases from 20 dB to 60 dB, although jitter is typically
implemented in the two operating systems. shorter than 1 ms. Second, the jitter curve can be characterized
by an exponential behavior, described in (3) and Table III.
D. Channel Occupancy Based on the setup for this investigation, the router served
Duty cycle (DC) is a temporal measurement which is only two laptops. Therefore, one can surmise that jitter is
defined as the ratio of time the system signal exceeds an mainly due to packets lost as a result of low SNR.
activity threshold over the entire observation time. In this test
the channel utilization efficiency of 802.11ax wireless network τ = αeβx (3)
is investigated. Fig. 7 plots the duty cycle of 802.11ax pair τ is jitter; x is SNR; and α and β are constant coefficients.
 [3] A. Mourad, S. Muhammad, M. O. Al Kalaa, P. A. Hoeher, and H. Refai,
35 “Bluetooth and IEEE 802.11n system coexistence in the automotive do-
20 MHz main,” in IEEE Wireless Communications and Networking Conference,
30 40 MHz WCNC, 2017.
80 MHz [4] A. Mourad, S. Muhammad, M. O. Al Kalaa, H. H. Refai, and P. A.
160 MHz Hoeher, “On the performance of WLAN and Bluetooth for in-car
25 infotainment systems,” Vehicular Communications, vol. 10, no. August,
pp. 1–12, oct 2017.
Jitter [ms]

20 [5] A. Algedir and H. H. Refai, “Adaptive d2d resources allocation un-

derlaying (2-tier) heterogeneous cellular networks,” in 2017 IEEE 28th
Annual International Symposium on Personal, Indoor, and Mobile Radio
15 Communications (PIMRC), Oct 2017, pp. 1–6.
[6] F. Aveta, H. H. Refai, and P. LoPresti, “Multi-user FSO communication
10 link,” in 2017 Cognitive Communications for Aerospace Applications
Workshop, CCAA 2017. IEEE Inc., aug 2017.
5 [7] E. Khorov, A. Kiryanov, A. Lyakhov, and G. Bianchi, “A Tutorial
on IEEE 802.11ax High Efficiency WLANs,” IEEE Communications
Surveys & Tutorials, vol. 21, no. 1, pp. 197–216, 2019.
0 [8] Q. Qu, B. Li, M. Yang, and Z. Yan, “An OFDMA based concurrent
0 10 20 30 40 50 60 multiuser MAC for upcoming IEEE 802.11ax,” in 2015 IEEE Wireless
SNR [dB] Communications and Networking Conference Workshops (WCNCW).
IEEE, mar 2015, pp. 136–141.
Fig. 8. 802.11ax jitter with various SNR using UDP link. [9] G. Naik, S. Bhattarai, and J.-M. Park, “Performance Analysis of Uplink
Multi-User OFDMA in IEEE 802.11ax,” in 2018 IEEE International
Conference on Communications (ICC), vol. 2018-May. IEEE, may
TABLE III 2018, pp. 1–6.
C OEFFICIENTS AND GOODNESS OF FIT OF JITTER VS . SNR MODEL . [10] B. Bellalta, “IEEE 802.11ax: High-efficiency WLANS,” IEEE Wireless
Communications, vol. 23, no. 1, pp. 38–46, feb 2016.
α β R2 [11] S. A. Rajab, W. Balid, and H. H. Refai, “Comprehensive study of
20 MHz 33.03 -0.1108 0.9705 spectrum occupancy for 802.11b/g/n homogeneous networks,” in Con-
40 MHz 685 -0.3319 0.9982 ference Record - IEEE Instrumentation and Measurement Technology
80 MHz 93.36 -0.2663 0.9982 Conference, vol. 2015-July. IEEE Inc., jul 2015, pp. 1741–1746.
160 MHz 1028 -0.4213 0.9973 [12] N. Mostahinic and H. Refai, “Spectrum occupancy for 802.11a/n/ac
homogeneous and heterogeneous networks,” in 2019 15th International
Wireless Communications and Mobile Computing Conference, IWCMC
2019. IEEE Inc., jun 2019, pp. 1690–1695.
V. C ONCLUSION [13] Z. Machrouh and A. Najid, “High efficiency IEEE 802.11ax MU-MIMO
and frame aggregation analysis,” in 2018 International Conference on
Prospective technologies pose challenging requirements on Advanced Communication Technologies and Networking (CommNet).
next-generation wireless communication systems and must be IEEE, apr 2018, pp. 1–5.
quickly addressed to ensure a successful roll out. Several [14] M. S. Afaqui, E. Garcia-Villegas, and E. Lopez-Aguilera, “IEEE
802.11ax: Challenges and Requirements for Future High Efficiency
global standardization governing bodies and consortiums are WiFi,” IEEE Wireless Communications, vol. 24, no. 3, pp. 130–137,
tirelessly releasing novel features and enhancements to current jun 2017.
protocols or proposing innovative protocols in an attempt to [15] E. Park, J. Choi, J. Chun, D. Lim, J. Kim, S. Kim, H. Choi, and H. Cho,
“1024 QAM Proposal,” 2015. [Online]. Available: https://mentor.ieee.
swiftly adapt to future demands. 802.11ax or Wi-Fi 6 is one org/802.11/dcn/15/11-15-1070-03-00ax-1024-qam-proposal.ppt
such solution proposed by the IEEE and Wi-Fi Alliance for [16] Z. Rong, Y. Yang, P. Loc, L. Liu, J. Luo, Y. Luo,
meeting needs in the unlicensed band. and Y. Lin, “CP Indication for UL MU Transmission,”
2015. [Online]. Available: https://mentor.ieee.org/802.11/dcn/15/
This work highlights an empirical investigation of new 11-15-0813-00-00ax-cp-indication-for-ul-mu-transmission.pptx
amendment performance from an application layer perspective. [17] C. Ghosh, R. Stacey, E. Perahia, S. Azizi, P.-K. Huang, Q. Li, L. Cariou,
A multitude of tests were conducted to investigate a number of and X. Chen, “UL OFDMA-based Random Access Procedure,”
2015. [Online]. Available: https://mentor.ieee.org/802.11/dcn/15/
metrics (e.g., throughput and jitter) and their relationship with 11-15-1105-00-00ax-ul-ofdma-based-random-access-procedure.pptx
parameters (e.g., payload length and environment variables, [18] Asus, “ASUS RT-AX88U Router.” [Online]. Available: https://www.
such as SNR). Empirical models were developed using test asus.com/us/Networking/RT-AX88U/
[19] Intel, “Intel Wi-Fi 6 AX200.” [Online]. Available: https://ark.intel.com/
results to quantify the behavior of said metrics. Fitting accu- content/www/us/en/ark/products/189347/intel-wi-fi-6-ax200.html
racy was 0.95 on average. Channel occupancy of the system [20] “iPerf.” [Online]. Available: https://iperf.fr/
was investigated as well. Results show that not only 802.11ax [21] “Wireshark.” [Online]. Available: https://www.wireshark.org/
[22] “Netspot.” [Online]. Available: https://www.netspotapp.com/
achieves higher throughput than its precedent, 802.11ac, but [23] “Acrylic WiFi.” [Online]. Available: https://www.acrylicwifi.com/en/
also exhibits a better channel utilization by virtue of its higher [24] “MTR.” [Online]. Available: http://www.bitwizard.nl/mtr/
MCS. [25] S. Ha, I. Rhee, and L. Xu, “CUBIC: A new TCP-friendly high-speed
TCP variant,” in Operating Systems Review (ACM), vol. 42, no. 5, jul
2008, pp. 64–74.
R EFERENCES
[1] Aruba, “The Wi-Fi Market and the genesis of 802.11ax,” 2018.
[Online]. Available: https://www.arubanetworks.com/assets/wp/WP{ }
802.11AX.pdf
[2] Ericsson, “Future mobile data usage and traffic growth,”
2019. [Online]. Available: https://www.ericsson.com/en/mobility-report/
future-mobile-data-usage-and-traffic-growth