ACCEPTED FROM OPEN CALL

mmWave IEEE 802.11ay for

5G Fixed Wireless Access
Khalid Aldubaikhy, Wen Wu, Ning Zhang, Nan Cheng, and Xuemin (Sherman) Shen

Abstract Thanks to mmWave technology, the 5G FWA can

provide Internet access with enhanced quality
Fixed wireless access (FWA) utilizing both of service and seamless user experience due to
licensed and unlicensed millimeter wave the high speeds, low latency, low cost, and faster
(mmWave) spectrum is considered a key tech- time-to-market that it can provide. Due to these
nology that can lead to the early deployment of advantages, several service providers around the
the fifth-generation new-radio (5G-NR) networks. globe, e.g., Verizon and Facebook, have adopt-
5G FWA can provide easy installation of network ed 5G FWA or plan to exploit this technology.
infrastructure and ubiquitous high-speed Internet Even though 5G FWA can be adopted by utiliz-
access at low cost compared to the conventional ing the technologies available in the 5G (3GPP
broadband fixed access networks. In this article, standard) using licensed mmWave spectrum, the
we investigate the mmWave distribution network mmWave distribution network (mDN) use case of
(mDN) use case that has been standardized IEEE 802.11ay that operates on the unlicensed 60
recently by the IEEE 802.11ay standard as an alter- GHz band is an alternative low-cost solution for
native 5G FWA solution. Specifically, we provide a operators to provide similar 5G performance.
comprehensive tutorial view of the considered new The mDN allows operators to provide fiber-re-
protocol specifications and design elements of the placement broadband services, where it has the
mDN. We also highlight some challenging research potential to build denser urban networks with
issues in the field of the mDN. Finally, we provide a the support of various emerging applications. In
case study based on the investigation of the mDN addition to the unlicensed 60 GHz band benefit,
where a low-complexity concurrent transmission the enhanced technologies considered in the IEEE
protocol is proposed to enhance the network per- 802.11ay standard can provide a promising high
formance while mitigating the interference. network performance [1]–[3]. Moreover, the mDN
can provide a similar user experience as 5G in
Introduction terms of capacity, efficiency and flexibility at a frac-
In the emerging next-generation networks tion of the cost of equivalent 5G small cell infra-
era, such as fifth-generation (5G) and the IEEE structure. The primary purpose of the mDN is to
802.11ay, data-intensive applications and a wide realize both wireless backhauling and point-to-mul-
range of use cases have become feasible as a tipoint (P2MP) mmWave access, which is based on
result of the high speeds and low latency provid- the concept of mesh networking topology. There-
ed by millimeter wave (mmWave) technology. fore, the mDN can offer a multi-Gbps fiber-like
Due to the vast spectrum swath, mmWave tech- speed and low-cost solution for 5G FWA.
nology can dramatically improve wireless appli- This article aims to investigate the mDN in
cation experiences by providing multi-gigabit order to deliver a comprehensive tutorial view
data-rate services. Heterogeneous networks with of the new protocol specifications and design
very diverse requirements are supported in the elements that have been proposed recently by
5G where different technologies are employed IEEE 802.11ay, where 5G FWA and IEEE 802.11ay
with different use cases, such as enhanced mobile standard are first reviewed. Then, future research
broadband (eMBB). Fixed wireless access (FWA) directions to address the challenges associated
is a specific case of the eMBB, where it can with the mDN are highlighted. Finally, a case study
exceed the quality of service of most current fixed is conducted to elaborate on how to improve the
broadband networks when either licensed or unli- network performance by taking advantage of the
censed mmWave spectrum is exploited. The 5G spatial reuse in a dense network.
FWA is viewed as one of the potential services
that can be enabled in early 5G deployment.
Although the concept of FWA was previous-
An Overview of 5G FWA and IEEE 802.11ay
ly used in the past cellular generations and stan- 5G Fixed Wireless Access
dards derived from the Wi-Fi family, the speeds FWA is a broadband network that provides
and latency provided are not comparable to the Internet access to customers utilizing wireless
service provided by fiber broadband networks. technology instead of the wired network, such

K. Aldubaikhy is with the Department of Electrical Engineering, Qassim University, Buraydah, Saudi Arabia. W. Wu and X. Shen are with
the Department of Electrical and Computer Engineering, University of Waterloo, Canada. N. Zhang is with the Department of Computing
Digital Object Identifier: Sciences, Texas A&M University at Corpus Christi, TX, USA. N. Cheng is with the School of Telecommunications Engineering,
10.1109/MWC.001.1900174 and with the State Key Lab of ISN, Xidian University, Xian, China.

88 1536-1284/20/$25.00 © 2020 IEEE IEEE Wireless Communications • April 2020

Authorized licensed use limited to: University of Waterloo. Downloaded on June 01,2020 at 12:29:45 UTC from IEEE Xplore. Restrictions apply.
as the fiber broadband network. In the past, transmitting data over channels simultaneously Another reason that
previous cellular generations and standards [1]–[3]. Channel bonding and aggregation are makes 5G FWA essen-
derived from the Wi-Fi family have been utilized utilized to create a wider channel from multiple tial is that many houses
for the FWA. However, the speeds and latency channels by combining two or more channels, around the globe
provided in these approaches cannot compete where there are up to six 2.16 GHz channels
with the fiber broadband networks. With the in the unlicensed 60 GHz frequency band. The do not have reliable
new advanced technologies, such as mmWave client-nodes (CNs) with enhanced direction- broadband access,
beamforming (BF) and small cell base stations al multi-gigabit (EDMG) ability can use channel where it can be expen-
[4]–[7], 5G FWA can deliver Internet access bandwidth containing 4.32, 6.48, or 8.64 GHz by sive and time-con-
to homes with fiber-like speeds. Another rea- combining two, three, or four 2.16 GHz channels, suming to deploy fiber
son that makes 5G FWA essential is that many respectively.
houses around the globe do not have reliable MIMO Channel Access: IEEE 802.11ay sup- broadband network
broadband access, where it can be expensive ports the downlink MU-MIMO to enable simul- everywhere. Therefore,
and time-consuming to deploy fiber broadband taneous transmissions for multiple CNs such that this raises an opportu-
network everywhere. Therefore, this raises an the spatial reuse and transmission robustness to nity for operators and
opportunity for operators and service providers link-outage can be enhanced [1]–[3]. To enable service providers to
to adopt the 5G FWA. MIMO channel access, physical carrier sensing,
5G FWA Deployments: Several 5G FWA virtual carrier sensing or clear channel assessment adopt the 5G FWA.
trials have been conducted by the industry to are implemented with specific MIMO backoff
demonstrate the potential of both licensed and procedure. Thus, a transmitter (TX) is permitted
unlicensed mmWave technologies. In 2017, to gain a transmission opportunity for a down-
Verizon started pre-standard 5G trials to pro- link MU-MIMO transmission when all the MIMO
vide 5G FWA and thereby launched this ser- beams are sensed to be idle while the backoff
vice, called 5G-Home, in limited locations in procedure is satisfied.
the United States. Another 5G FWA trial was Directional MAC: The IEEE 802.11ay direc-
launched recently by Huawei and TELUS in tional MAC protocol includes a beacon interval
Canada by using Huawei 5G products that are (BI) access time that has four main components,
based on entirely 3GPP standards. Facebook, where the functions of each of them are described
Deutsche Telekom, and Qualcomm are work- as follows. First, during the beacon transmission
ing together on Facebook’s Terragraph project interval (BTI), BF training can be initiated by a dis-
which is based on the mDN that operates in the tribution-node (DN) or an access point, where an
unlicensed 60 GHz band. Furthermore, SiKlu Announce frame or EDMG-Beacon is transmitted.
launched a P2MP mmWave system which is Second, during association beamforming training
another 5G FWA trail that is based on the unli- (A-BFT), BF training can be completed by CNs by
censed 60 GHz band. utilizing A-BFT slots. The number of A-BFT slots
5G FWA Advantages: The main advantages in IEEE 802.11ay is increased to up to 16 slots
of the 5G FWA based on either licensed or unli- using the extended A-BFT mechanism or up to 48
censed mmWave spectrum are summarized as slots utilizing the multi-channel A-BFT method [1],
follows: [3]. Third, during the announcement transmission
• High speeds and very low latency can be pro- interval (ATI), exchanges in the request-response-
vided due to the mmWave spectrum. based management and allocation information
• Reduced cost of network infrastructure can be can be announced. Finally, during the data trans-
realized compared to the fiber broadband net- fer interval (DTI), slot allocations are established
works. by using either contention-based access periods
• Faster time-to-market can be achieved, where or scheduled-service periods (SPs). Time division
quick and cheap Internet access can be duplex (TDD) channel access allocation (TDD-SP)
enabled. is another type of allocation that is utilized for the
• Ubiquitous connectivity to rural areas can be mDN use case during the DTI.
delivered, where wired access infrastructure is BF Training: BF training is utilized to estab-
unavailable. lish a communication link between a TX and a
receiver (RX) in order to overcome the challeng-
IEEE 802.11ay es that result from the directional transmission.
The IEEE 802.11ay standard is an advanced The BF training of IEEE 802.11ay is similar to the
communication system that operates on the legacy 802.11ad, except that it includes new BF
unlicensed 60 GHz band to enhance the user training stages to enhance the BF process. BF
experience, where point-to-point (P2P) and P2MP for multi-channel, BF for asymmetric links, BF for
are considered for indoor and outdoor commu- MIMO, and BF for the TDD channel access are
nications. The IEEE 802.11ay standard has been the new BF training protocols, where the basic BF
established to improve the legacy 802.11ad stan- training of IEEE 802.11ay is a sector-level sweep
dard while guaranteeing backward compatibility (SLS) BF stage. The SLS is described briefly in the
for legacy users. IEEE 802.11ay utilizes BF training following which is a necessary operation before
and specific directional medium access control communication can be established.
(MAC), where multi-user (MU) multiple-input mul- • First, initiator-transmit-sector-sweep (I-TXSS)
tiple-output (MIMO) and channel bonding and frames are transmitted by the DN in different
aggregation are supported to improve the spec- directions in the BTI whereas quasi-omni mode
trum efficiency and system throughput. is used by each CN to receive the frames.
Channel Bonding and Aggregation: In order • Second, each CN responds by transmitting a
to enhance the channel utilization and achieve responder-transmit-sector-sweep (R-TXSS)
significant throughput gain, the IEEE 802.11ay frame in the A-BFT whereas quasi-omni mode is
standard supports the multi-channel approach by used by the DN to receive the frames.

IEEE Wireless Communications • April 2020 89

Authorized licensed use limited to: University of Waterloo. Downloaded on June 01,2020 at 12:29:45 UTC from IEEE Xplore. Restrictions apply.
 a later section.) Various placement options, such
Fiber as rooftops, house-walls, and street poles, are
PoP supported for the DNs in the mDN. The mDN is
WTTH
connected to the provider network through point-

Fib
Provider of-presence (POP) using an optical fiber link. The

er
Network
WTTB distance between the DNs is up to 1000 m and
300 m for rooftops and street poles, respectively,
Fiber iber
PoP F while the distance between a DN and CN is up
Small-Cell
to 100 m.

mDN mDN Applications

Provider The primary purpose of the mDN is to achieve
Network wireless backhaul network and perform P2MP
mmWave access to CNs for the sake of realizing
Fiber
PoP the 5G fiber-replacement application scenarios.
These application scenarios include wireless-to-
the home (WTTH) access, wireless-to.the-building
(WTTB) access, WiFi DN backhauling, and small-
cell backhauling. For the WTTH scenario, the DNs
Fiber PoP
can deliver the Internet service to homes by using
to a window access point at the CN side. For the
Provider
Network WTTB scenario, an additional wireless or fixed
network inside the building is required to serve
the individual units. A further and optional indoor
wireless or wired network can be utilized in the
WTTH and WTTB scenarios. Different frequency
bands from the mmWave links for backhauling
DN node CN node mm-Wave link Fiber link can be used for the WTTH and WTTB scenari-
os. For WiFi and small-cell backhauling scenar-
FIGURE 1. The network topology of the mDN.
io, the mDN is used to aggregate and carry the
traffic of the mobile services to the mobile core.
• Finally, the decision for CN association is made, The capacity level in this scenario is less than the
and the best sector identification (Sid) is report- WTTH and WTTB scenarios.
ed by the DN when sector sweep (SSW)-Feed-
back is transmitted in the A-BFT. mDN Benefits
Data traffic in communication networks in cities
mmWave Distribution Network is growing exponentially which demands a tre-
IEEE 802.11ay supports the mDN use case, mendous amount of bandwidth per km 2 to sup-
which can provide a promising cost-efficient and port new smart city applications, such as smart
high-performance solution to fiber broadband net- utility meters, traffic cameras, a massive number
works. Recently, FWA has gained considerable of sensors and smart parking meters. Although
attention from both academia and service pro- optical fiber cable solutions can offer a tremen-
viders because of the reduced cost of network dous amount of bandwidth, the infrastructure cost
infrastructure and high speeds compared to the of the fiber network is expensive, and it can be
fiber-to-the x (FTTx) networks. The mDN has the slower to implement in all locations. Therefore, an
potential to build denser urban networks that are efficient solution, such as the mDN, is needed in
capable of connecting more people everywhere order to provide Internet access everywhere and
in the world with the support of various emerging improve quality of life. The main benefits of the
applications. The design elements and specific mDN can be summarized as follows:
details of the IEEE 802.11ay mDN are presented • High-speed access (fiber-like speed) can be
in the following subsections. achieved which is based on the 60 GHz of the
IEEE 802.11ay.
mDN Architecture • A significant lower-cost network infrastructure is
As illustrated in Fig. 1, the mDN is based on offered for fixed access networks.
the concept of mesh networking topology, • Faster time-to-market can be delivered, where
where each node not only sends and receives the small mDN access points (DNs) can be
its data (within its coverage area) but also acts implemented on pre-existing street poles or
as a relay node to have a direct and dynamic buildings without the need for time-consuming
connection with other nodes in the network [8]. permissions.
The mDN can be deployed by several outdoor • Ubiquitous coverage with high speeds can be
DNs, where redundancy can be attained. The realized, where the service can reach areas that
mDN can enable neighboring outdoor DNs over can be difficult for fiber network infrastructure
multi-hops to construct and preserve non-over- in urban or rural areas.
lapping data transmission and control allocations
dynamically. Furthermore, multi-hop communi- mDN Protocol
cation is supported to reach destinations in the The IEEE 802.11ay standard defines a distribut-
area coverage. Line-of-sight (LOS) communication ed scheduling protocol specifically for the mDNs
is required among the DNs by considering the when two neighboring DNs or more are partially
potential mDN challenges, such as interference overlapped in order to enhance spatial sharing
and beam tracking. (This is further discussed in and interference mitigation, where one or more

90 IEEE Wireless Communications • April 2020

Authorized licensed use limited to: University of Waterloo. Downloaded on June 01,2020 at 12:29:45 UTC from IEEE Xplore. Restrictions apply.
of the channels utilized in these DNs are similar Announce frame
[1]. The purpose of this protocol is to arrange
dynamic non-overlapping allocations and assure TDD-Slot- TDD-Slot-
fair resource scheduling among DNs, where a ... Structure- Schedule- ...
highly dynamic network is considered with delay Element Element
sensitive traffic. The main aspects of the mDN
scheduling protocol are summarized as follows.
•Distributed scheduling protocol can be
Beacon interval
established during an Announce frame in
order to initiate the TDD channel access. The Beacon Data transfer interval
Announce frame contains two subfields called header
TDD-Slot-Structure-Element and TDD-Slot-Sched- SP or
ule-Element; they are responsible for describing BTI A-BFT ATI SP 1 SP 2 TDD-SP 1 ...
TDD-SP
the structure of the TDD channel access and
addressing the access time assignments by using
TDD bitmap schedule, respectively. Therefore, in
order to enable the distributed scheduling proto- TDD-Interval 1 TDD-Interval 2 ... TDD-Interval Q
col and identify the upcoming transmission sched-
ules of neighboring DNs, each DN must listen
for Announce frames from neighboring DNs and
may need to listen in different channels.
•For a specific channel, each DN needs to GT2 GT2 GT2
divide its BI duration by the number of discovered GT1 GT3
neighboring DNs for the sake of distributing chan- TDD-Slot TDD-Slot TDD-Slot TDD-Slot TDD-Slot TDD-Slot
nel resources. To maintain fairness among DNs in ... ...
0 1 i i+1 i+2 M
the same coverage area with the same channel,
the maximum access time allocations of each DN TX TDD slots RX TDD slots
is (BI duration)/(number of discovered neighbor-
ing DNs + 1). FIGURE 2. The allocation structure of the TDD channel access.
TDD Channel Access: In order to achieve
fairness communication with no interference in iii) A TDD-SSW-Ack frame is sent by the DN after
the mDNs, a TDD-SP allocation is defined, where the reception of a TDD-SSW-Feedback in order
the TDD channel access rules are specified. As to acknowledge the received configuration.
shown in Fig. 2, a TDD-SP consists of consecutive This process can be repeated if needed unless
TDD-Intervals, where each TDD-Interval consists it is specified by the DN to end the TDD individu-
of consecutive TDD-Slots that are separated by al BF during the TDD-SSW-Ack.
new guard times (GTs). During both the TDD-SP TDD Group BF: In order to increase the TDD
and TDD-Interval, the allocation time may contain BF efficiency by reducing the BF training over-
one or more slots. The TDD bitmap schedule is head, TDD group BF is introduced by perform-
utilized to indicate the allocation type, e.g., TX or ing BF training for multiple CNs. Hence, the BF
RX allocations, and the access permissions. The training time is reduced since a certain number
bitmap information is transmitted by each DN to of CNs are trained simultaneously. As shown in
its neighboring DNs in unicast messages during Fig. 3(b), the procedure of the TDD group BF is
the Announce frame. Moreover, the allocation similar to the TDD individual BF with the following
category, such as a control, management, or modifications.
Data-only frame, is identified. i) A DN sends multiple TDD-SSW frames for mul-
tiple CNs by using broadcast the MAC address
mDN BF Training Protocol subfield in order to initiate the TDD group BF.
Regular BF training is not appropriate with the In the TDD-SSW frames, the number of CNs to
TDD channel access allocations because differ- be trained is specified.
ent interframes to separate the BF frames are uti- ii) TDD-SSW-Feedback frames are sent by each
lized. Therefore, a new TDD BF training is defined CN in different time allocations using the sector
for the mDNs by considering the new GT inter- with the best link quality.
frames. The TDD individual and group BF training iii) After receiving TDD-SSW-Feedback frames,
procedures are described in the following. the DN transmits several TDD-SSW-Ack frames
TDD Individual BF: To establish a communica- to each CN in different time allocations in
tion link between a CN and a DN, the TDD indi- order to transmit the sector used by the DN,
vidual BF is utilized by performing a sequential BF the sector used by the CN, and whether the BF
training. In this process, only one CN can perform training is completed or not.
BF training with a DN during assigned TDD-Slots
by using unicast TDD BF frames, and these frames mDN Challenges
are repeated for each CN in order to be trained. Despite the attractive benefits of the mDN, many
As illustrated in Fig. 3(a), the procedure of the open questions, problems, and limitations remain.
TDD individual BF is given as follows. In this section, we identify some future research
i) A DN sends several TDD-SSW frames while a directions based on the challenges that are associ-
CN uses all of its receive sectors during this ated with the mDN.
time. The TDD-SSW frame is sent repeatedly to Beam Tracking: The overhead generated from
span all of DN’s Sid. beam tracking is a challenge related to mmWave
ii) A CN transmits a TDD-SSW-Feedback frame by communication that arises because of the need
using the sector that has the best quality. for tracking the changes that occur in analog and

IEEE Wireless Communications • April 2020 91

Authorized licensed use limited to: University of Waterloo. Downloaded on June 01,2020 at 12:29:45 UTC from IEEE Xplore. Restrictions apply.
 TX Sector ID (i, ..., j) RX Sector ID (i)
this challenge, power selection allocation or prior-
TX Sector ID (j) TX Sector ID (i)
TDD-slot TDD-slot TDD-slot TDD-slot ity-based resource allocation is required to handle
TX ... ... ... ... ... ... ... ... large numbers of users in the multi-hop scenario
DN
RX
and to reduce the energy consumption of the sys-
Received using RX Sector ID (m)
tem while keeping the complexity reasonably low
RX
Sector (m) ... ...
for practical utilization.
CN
TX ... TX Sector ID (m)
Case Study
Sweep receiver sectors TDD-slot In the foreseeable future, emerging data-hun-
through its sector IDs
gry applications and the proliferation of wireless
(a) devices will substantially increase wireless data
TX Sector ID (i, ..., j) RX Sector ID (i) TX Sector ID (j) TX Sector ID (i) traffic. To further improve spectrum efficiency of
TDD-slot TDD-slot TDD-slot TDD-slot mmWave systems with a large number of users,
TX ... ... ... ... ... ... ... ... massive MIMO can be exploited to utilize con-
DN current transmission. Due to the narrow beams of
RX
Received using RX Sector ID (m) mmWave directional antennas, concurrent trans-
RX
Sector (m) ... ... missions are possible based on the employment
CN of a spatial reuse mechanism. The IEEE 802.11ay
TX ... TX Sector ID (m)
Received using RX Sector ID (w) standard supports the MU-MIMO transmission to
...

RX
Sector (w)
... ...
TDD-slot achieve significant throughput gain [1].
CN N This section presents a case study where a
TX ... TX Sector ID (w)
low-complexity concurrent transmission protocol,
Sweep receiver sectors TDD-slot utilizing one channel frequency, is employed to
through its sector IDs
(b) mitigate the interference and support the mDN
use case by exploiting the spatial reuse character-
istic. The network performance can be affected
TDD-SSW TDD-SSW-Feedback TDD-SSW-Ack
by the CCI and adjacent channel interference,
FIGURE 3. TDD BF training procedure: (a) TDD individual BF (b) TDD group BF. in which multiple neighboring nodes are partially
overlapped in a dense scenario. This case study
focuses on exploiting spatial reuse while mitigat-
baseband beams, e.g., due to mobility, RX rota- ing the interference in the dense mDN by imple-
tion, or blockage. In order to track beam changes menting an efficient MAC protocol. Specifically,
between a TX and RX while keeping overhead we propose a low-complexity concurrent trans-
at an acceptable level, an efficient mechanism is mission protocol to attain concurrent downlink
required by employing analog tracking with limit- transmissions and mitigate the CCI in a wireless
ed digital baseband channel tracking. mDN without employing digital BF.
Blockage Effect: Blockage effect is one of the It is worth noting that in order to mitigate the
main challenges associated with beam alignment CCI using traditional lower frequency systems,
in mmWave communications which can be a numerous mechanisms have been proposed in
result of many causes, including low-emissivity the literature by mainly using digital BF methods,
glass and high-density tree foliage. To overcome such as zero-forcing BF [11]. Nevertheless, for
this challenge, new and efficient beam tracking mmWave communications, these methods intro-
approaches are needed in order to guarantee the duce significant overhead and extreme compu-
link robustness and quality of user experience. tational complexity [12] because the size of the
For example, machine learning approaches can channel state information feedback increases with
be utilized to find the optimal transmit-receive the number of antennas and the number of devic-
beam pair while considering the beam alignment es. Furthermore, extra antennas are needed on
latency. each TX node for digital BF to mitigate the CCI.
Interference Management: Maintaining efficient
utilization of the mDN concept with ultra-dense Considered Scenario
deployment requires spatial and spectrum reuse, As shown in Fig. 4(a), CNs are densely deployed in
which negatively affects the end-user experience an mDN, where they can communicate with DNs
at cell boundaries due to co-channel interference with directional antennas by performing BF train-
(CCI) [9], [10]. Furthermore, interference can be ing of the IEEE 802.11ay system. TDD BF training
a crucial issue and difficult to be managed espe- is considered in order to realize an appropriate
cially for IEEE 802.11ay implementation since it is link budget between a TX and RX before transmit-
a distributed network. New and efficient methods ting data, where antenna reciprocity is assumed for
with low complexity for managing interference both the TX and RX. Highly dynamic topologies
and radio resources must be developed for the are assumed for the mDN use case for the sake of
ultra-dense 5G FWA networks. plug and play mechanism to lower the operating
expense even though the DNs are fixed while the
Uplink MU-MIMO Transmission CNs are not mobile frequently, nearly static.
In the mmWave uplink MU-MIMO, a large num-
ber of users is expected to be handled in dense Low-Complexity Concurrent Transmission Protocol
deployment scenarios and applications, such as As depicted in Fig. 4(b), to mitigate the MU inter-
the Internet of Things (IoT) and vehicular commu- ference and the CCI, the low-complexity concur-
nications. Although uplink MU-MIMO transmis- rent transmission protocol divides the coverage
sion is a promising technology, the user selection area of each DN into virtual sectors. These virtual
algorithm needs to be considered in a mmWave sectors are divided based on the beam direction
system with a large user population. To overcome associated with the optimal Sid determined when

92 IEEE Wireless Communications • April 2020

Authorized licensed use limited to: University of Waterloo. Downloaded on June 01,2020 at 12:29:45 UTC from IEEE Xplore. Restrictions apply.
 In the foreseeable

Sid 033

Sid 032
future, emerging

Sid 034
... ... . ...
... ... ..

...
... ... ...
data-hungry appli-
...
...

...
...

8
01
S id

7
cations and the pro-

01
Sid
S

04
id
6
Sid 049 01

S
id
S id
liferation of wireless

...
... ... 05

...
0

...

...

...
devices will substan-

...

...
... ..
3 2 tially increase wireless

... ...
Spatial-interference
problem

. ...
Sid 064 4 1 Sid 002 data traffic. To further
Sid 001
improve spectrum
Sid 065
Sid 066
5 8 Sid 128
efficiency of mmWave

. ...
6 7 systems with a large

... ...

... ..
number of users,

...

...
Sid

...

...
0 S 11
11 4
massive MIMO

...
08

id
3
Sid
can be exploited to

08 1

Sid
08
2
Spatial-

11
...

Sid
... utilize concurrent

2
interference ...

Sid
...
problem ...

Sid 098
... ..

Sid 097
Sid 096
. ... ... ... transmission.

(a) Considered case study scenario. (b) Virtual segmentation of the predefined antenna
directional Sid (total Sid = 128, as an example).

FIGURE 4. The illustration of the concurrent transmission protocol.

the CNs report their Sid during the TDD BF train-

ing. Then, these virtual sectors are divided into A
two orthogonal groups (G1 and G2) so that con-
current transmission can be achieved in separate Sector angle ( )
time slots for each group [13]–[15]. For the sake
of canceling interference among CNs that overlap
in the same active virtual sector, a Spatial-Reuse x1
condition is proposed as an essential pre-requisite
criterion to meet before transmission can be grant- r
ed where the spatial-interference problem can be
avoided. As illustrated in Fig. 4(a), the spatial-inter-
ference problem happens when more than one
CN is reported in the same virtual sector within DN 1 x2
a single hop or when the used virtual sector will
incur CCI to any active CNs in other DNs. Specifi-
cally, the concurrent transmission can start when DN 2
the Spatial-Reuse condition is fulfilled, and other-
wise, it is terminated from the concurrent transmis-
sion and schedule the intended CN sequentially in
a different time slot if necessary. The details of the FIGURE 5. Example of the concurrent transmission
proposed low-complexity concurrent transmission scenario in an mDN.
protocol are outlined as follows.
1) A DN initiates an Announce frame to estab- b) If Spatial-Reuse condition is not satisfied
lish a TDD channel access while each DN must when the used virtual sector will incur CCI
listen to identify the upcoming transmission to any active CNs in partially overlapped
schedules of neighboring DNs, where the max- DNs, the intended concurrent transmis-
imum permitted BI duration and bitmap infor- sion in this specific virtual sector for the
mation can be exchanged. corresponding CN will be terminated and
2) After the BF training is completed, all DNs then be rescheduled in a different time slot
check and announce the best Sid information sequentially.
that is exchanged from all the associated CNs. 4) Finally, Data-only TDD-Slots are utilized for
3) Both the MU interference and CCI will be concurrent and sequential data transmissions
examined by a DN upon reception of the best during the TDD-Interval allocations, where
Sid information before granting concurrent data these allocations are announced.
transmission allocations as follows.
a) If Spatial-Reuse condition is not satisfied Performance Evaluation
when more than one CN are reported in We evaluate the performance of the low-com-
the same virtual sector within a single hop, plexity concurrent transmission protocol by ana-
one CN with the highest signal-to-noise ratio lyzing the spatial multiplexing gain. The spatial
(SNR) will be picked for concurrent transmis- multiplexing gain represents the number of con-
sion. During the TDD-Interval, the selected current transmissions allowed over the same
CN will be allocated in time slot t or t +1 if mmWave channel in a given area from the per-
it is located in G1 or G 2, respectively. The spective of spatial reuse. Moreover, the probabili-
un-selected CNs will be allocated in a differ- ty of achieving concurrent transmissions depends
ent time slot sequentially. on the coverage area and the virtual sector angle,

IEEE Wireless Communications • April 2020 93

Authorized licensed use limited to: University of Waterloo. Downloaded on June 01,2020 at 12:29:45 UTC from IEEE Xplore. Restrictions apply.
 Concurrent Transmission Probability

Average mDN Throughput (Gbps)

Proposed Protocol (single-hop performance)
Proposed Protocol (one interfering sector is eliminated)
Proposed Protocol (both interfering sectors are eliminated)
Single transmission (TDMA)

Number of DNs Number of DNs

(a) Probability Q versus the number of DNs (r = 1600m). (b) Average network throughput.

FIGURE 6. Low-complexity concurrent transmission protocol performance.

where interference occurs when the active virtual transmissions in the mDN has demonstrated the
sectors are overlapped. potential of concurrent transmission on enhancing
For simplicity, we consider a traceable case as network performance while mitigating the inter-
shown in Fig. 5, where A is a circle area with radi- ference.
us r. There are two or more DNs that have cover-
age area with radius xi, where they are randomly References
[1] IEEE Draft Standard for Information Technology–Telecommu-
distributed in A. Let a1 and a2 denote the areas of nications and Information Exchange Between Systems Local
the active virtual sectors of DN1 and DN2, respec- and Metropolitan Area Networks–Specific Requirements
tively. Then, the probability (P1) that a2 does not Part 11: Wireless LAN Medium Access Control (MAC) and
overlap with a1 is 1–(x21/2r2) while the probability Physical Layer (PHY) Specifications–Amendment: Enhanced
Throughput for Operation in License-Exempt Bands Above
(P2) that a1 does not overlap with a2 is 1–(x22/2r2). 45 GHz, IEEE P802.11ay/D2.0, July 2018, Standard.
Thus, the probability that both DN1 and DN2 do [2] Y. Ghasempour et al., “IEEE 802.11ay: Next-generation 60
not interfere with each other is Q = P1P2. Fig. 6(a) GHz Communication for 100 Gb/s Wi-Fi,” IEEE Commun.
shows the numerical results of the probability Q Mag., vol. 55, no. 12, Dec. 2017, pp. 186–192.
[3] P. Zhou et al., “IEEE 802.11ay-based mmWave WLANs:
versus the number of DNs when different values Design Challenges and Solutions,” IEEE Commun. Sur-
of virtual sector angle () and radius coverage (x) veys & Tutorials, vol. 20, no. 3, Third Quarter 2018, pp.
are considered, assuming all DNs have the same 1654–81.
x in every time. Fig. 6(a) shows how the concept [4] T. S. Rappaport et al., “Millimeter Wave Mobile Communica-
tions for 5G Cellular: It Will Work!” IEEE Access, vol. 1, May
of the proposed protocol by separating the cov- 2013, pp. 335–49.
erage zone into virtual sectors can increase the [5] J. G. Andrews et al., “What Will 5G Be?” IEEE J. Sel. Areas
channel utilization efficiency and interference mit- Commun., vol. 32, no. 6, Jun. 2014, pp. 1065–82.
igation as long as the coverage area of each DN [6] W. Wu et al., “Beef Up mmWave Dense Cellular Networks
with D2D-assisted Cooperative Edge Caching,” IEEE Trans.
is small enough. Vehicular Technology, vol. 68, no. 4, Apr. 2019, pp. 3890–
Fig. 6(b) shows how the unique spatial reuse 3904.
of the low-complexity concurrent transmission [7] J. Qiao, Y. He, and X. S. Shen, “Improving Video Streaming
protocol can be exploited to realize concurrent Quality in 5G Enabled Vehicular Networks,” IEEE Wireless
Commun., vol. 25, no. 2, Apr. 2018, pp. 133–139.
downlink transmission compared to the time-divi- [8] M. Grigat et al., “mmWave Distribution Network Usage
sion multiple access (TDMA) method. We simu- Model,” IEEE 802.11-17/1019r2, Tech. Rep., July 2017.
late nearly two-million BIs taking different network [9] E. Hossain et al., “Evolution toward 5G Multi-tier Cellular
topologies into account to show the average Wireless Networks: An Interference Management Perspec-
tive,” IEEE Wireless Commun., vol. 21, no. 3, Jun. 2015, pp.
mDN throughput performance per data time slot. 118–127.
As shown in Fig. 6(b), significant average network [10] A. Padaki et al., “Network Simulations for FWA Mesh Dis-
throughput can be achieved due to the two-time tribution Networks,” IEEE 802.11.18/1086r0, Tech. Rep.,
slots of G 1 and G 2 that can realize concurrent July 2018.
[11] M. A. Maddah-Ali, A. S. Motahari, and A. K. Khandani,
transmission. Simulation results demonstrate that “Communication over MIMO X Channels: Interference
both interference mitigation and simultaneous Alignment, Decomposition, and Performance Analysis,” IEEE
transmission can be realized without the need for Trans. Inf. Theory, vol. 54, no. 8, Aug. 2008, pp. 3457–70.
complex precoding methods. [12] A. Alkhateeb et al., “MIMO Precoding and Combining
Solutions for Millimeter-wave Systems,” IEEE Commun. Mag.,
Conclusions vol. 52, no. 12, Dec. 2014, pp. 122–131.
[13] X. Shen et al., “Methods and Systems for Multi-user
This article has investigated the new protocol Beamforming,” UNITED STATES: Patent Application No.
specifications and design elements of the mDN 15/360852, Nov. 23, 2016.
[14] K. Aldubaikhy and X. Shen, “Simultaneous DL Transmission
by providing a comprehensive picture to adopt an in mmWave Ultra Dense Networks: Inter-BSS Interference
alternative solution to conventional fixed networks Prospective,” in Proc. IEEE International Conference on Com-
that can be a strong business case to provide munications Workshops (ICC Workshops), May 2018, pp.
Internet access to households worldwide. Some 1–6.
[15] K. Aldubaikhy, W. Wu, and X. Shen, “HBF-PDVG: Hybrid
of the mDN challenges have also been highlight- Beamforming and User Selection for UL MU-MIMO
ed for future research and improvement. Finally, mmWave Systems,” in Proc. IEEE Global Communications
a case study on realizing concurrent downlink Conference (Globecom), Dec. 2018, pp. 1–6.

94 IEEE Wireless Communications • April 2020

Authorized licensed use limited to: University of Waterloo. Downloaded on June 01,2020 at 12:29:45 UTC from IEEE Xplore. Restrictions apply.
 Biographies sity of Waterloo in 2016, and the B.E. and M.S. degrees from
the Department of Electronics and Information Engineering,
Khalid Aldubaikhy (S’10-M’19) is currently an assistant professor
Tongji University, Shanghai, China, in 2009 and 2012, respec-
at Qassim University, Buraydah, Al-Qassim, Saudi Arabia. He
tively. He is currently a professor with the School of Telecom-
received the B.E. degree from Qassim University, Saudi Ara-
munication Engineering, and with the State Key Lab of ISN,
bia, in 2008, the M.A.Sc. degree in electrical and computer
Xidian University, Shaanxi, China. He worked as a post-doc-
engineering from Dalhousie University, Halifax, NS, Canada, in
toral fellow with the Department of Electrical and Computer
2012, and the Ph.D. degree in electrical and computer engineer-
Engineering, University of Toronto, from 2017 to 2018. His
ing from the University of Waterloo, Waterloo, ON, Canada, in
current research focuses on space-air-ground integrated sys-
2019. His research interests include millimeter-wave wireless
tems, big data in vehicular networks, and self-driving systems.
networks, medium access control, impulse radio ultra-wideband,
His research interests also include performance analysis, MAC,
and millimeter-wave 5G cellular networks.
opportunistic communication, and the application of AI for
Wen Wu (S’13-M’19) earned the Ph.D. degree in electrical vehicular networks. He is also interested in space-air-ground
and computer engineering from the University of Waterloo, integrated networks.
Waterloo, ON, Canada, in 2019. He received the B.E. degree
in information engineering from South China University of Tech- Xuemin (Sherman) Shen (M’97-SM’02-F’09) received the Ph.D.
nology, Guangzhou, China, and the M.E. degree in electrical degree in electrical engineering from Rutgers University, New
engineering from the University of Science and Technology of Brunswick, NJ, USA, in 1990. He is currently a University
China, Hefei, China, in 2012 and 2015, respectively. Starting Professor with the Department of Electrical and Computer
from 2019, he has been working as a post-doctoral fellow with Engineering, University of Waterloo, Canada. His research
the Department of Electrical and Computer Engineering, Univer- focuses on resource management in interconnected wireless/
sity of Waterloo. His research interests include millimeter-wave wired networks, wireless network security, social networks,
networks and AI-empowered wireless networks. smart grid, and vehicular ad hoc and sensor networks. He
is a registered Professional Engineer of Ontario, Canada, an
Ning Zhang (M’15-SM’18) received the Ph.D. degree from the Engineering Institute of Canada Fellow, a Canadian Academy
University of Waterloo, Canada, in 2015. After that, he was a of Engineering Fellow, a Royal Society of Canada Fellow, and a
postdoc research fellow at the University of Waterloo and the Distinguished Lecturer of the IEEE Vehicular Technology Soci-
University of Toronto, Canada, respectively. Since 2017, he has ety and Communications Society. Dr. Shen received the James
been an assistant professor at Texas A&M University-Corpus Evans Avant Garde Award in 2018 from the IEEE Vehicular
Christi, USA. He serves as an associate editor for IEEE Internet Technology Society, the Joseph LoCicero Award in 2015 and
of Things Journal, IEEE Transactions on Cognitive Communica- the Education Award in 2017 from the IEEE Communications
tions and Networking, IEEE Access, and IET Communications, Society. He has also received the Excellent Graduate Supervi-
and Vehicular Communications (Elsevier); and a guest editor for sion Award in 2006 and the Outstanding Performance Award
several international journals such as IEEE Wireless Communi- in 2004, 2007, 2010, and 2014 from the University of Water-
cations. He also served as a workshop chair for MobiEdge’18 loo and the Premier’s Research Excellence Award (PREA) in
(in conjunction with IEEE WiMob 2018) and CoopEdge’18 (in 2003 from the Province of Ontario, Canada. He served as the
conjunction with IEEE EDGE 2018), and 5G&NTN’19 (in con- Technical Program Committee Chair/Co-Chair for the IEEE
junction with IEEE ICCC 2019). He has received six Best Paper Globecom’16, the IEEE Infocom’14, the IEEE VTC’10 Fall, the
Awards from international conferences and journals, such as IEEE Globecom’07, the Symposia Chair for the IEEE ICC’10,
IEEE Globecom and IEEE ICC. His current research interests the Tutorial Chair for the IEEE VTC’11 Spring, the Chair for
include wireless communication and networking, mobile edge the IEEE Communications Society Technical Committee on
computing, and security. Wireless Communications, and P2P Communications and Net-
working. He is the Editor-in-Chief of the IEEE Internet of Things
Nan Cheng (S’12-M’16) received the Ph.D. degree from the Journal and the Vice President of Technical and Educational
Department of Electrical and Computer Engineering, Univer- Activities of the IEEE Communications Society.

IEEE Wireless Communications • April 2020 95

Authorized licensed use limited to: University of Waterloo. Downloaded on June 01,2020 at 12:29:45 UTC from IEEE Xplore. Restrictions apply.