Professional Documents
Culture Documents
Frequency-Domain Contention and Polling MAC Protocols in IEEE 802.11 Wireless Networks A Survey
Frequency-Domain Contention and Polling MAC Protocols in IEEE 802.11 Wireless Networks A Survey
PII: S0140-3664(18)30012-4
DOI: 10.1016/j.comcom.2018.07.023
Reference: COMCOM 5742
Please cite this article as: Haithem Al-Mefleh, Osameh Al-Kofahi, Frequency-Domain Contention and
Polling MAC Protocols in IEEE 802.11 Wireless Networks: A Survey, Computer Communications
(2018), doi: 10.1016/j.comcom.2018.07.023
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service
to our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and
all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Abstract
T
Shifting from the time domain to the frequency domain is an interesting recent direction that has been exploited to
IP
improve the performance of wireless networks. It is shown that the frequency-domain based proposals are of obvious
benefits especially for higher rates and densities which are important attributes of the next-generation wireless networks.
In this survey paper, we provide an overview of the concept of frequency-domain based contention and polling MAC
CR
protocols, and we show how these schemes could enhance the performance of wireless networks. We also attempt to
provide a comprehensive survey of such schemes including related proposals for the next-generation IEEE 802.11ax
wireless networks. In addition, we compare the surveyed protocols based on factors like the evaluation methods, the
number of contention rounds, the conveyed signals, and the purpose of each protocol. We further address different
US
challenges and issues of frequency-domain schemes including the hidden terminal problem, the number of subcarriers,
TDMA-like transmissions, coexistence, the number of antennas, machine to machine networks, overlapping networks,
and synchronization. Finally, we highlight different possible directions for future work.
AN
Keywords: Wireless networks, IEEE 802.11, Medium Access Control (MAC) protocols, PHY
1. Introduction where users compete for the channel. In the backoff pro-
cedure, each user randomly picks a number of time slots
M
Today, IEEE 802.11-based transceivers are almost em- (during which the channel must be idle) to wait before at-
bedded everywhere; in computers, phones, TVs, routers, tempting to transmit. Moreover, the standard defines two
etc. [1]. In addition, there is a high dependency on WiFi modes of operation within DCF. First, the basic mode ad-
ED
(or wireless networks that are mostly implemented follow- vocates the exchange of data and ACK (acknowledgement)
ing the IEEE 802.11 standard) to access the Internet [1]. frames directly after the end of the contention period. Sec-
Hence, over the last few years, the IEEE 802.11 standard ond, RTS/CTS (Request to Send/Clear to Send) mode
([2, 3, 4, 5, 6, 7]) has evolved dramatically to support dif- demands the exchange of short control frames (RTS and
PT
ferent growing demands of users. The standard is adopting CTS) before the Data/ACK transmission. The RTS/CTS
new physical layers and technologies to achieve very high mode mitigates the performance penalty due to collisions
rates and support a higher number of users. Correspond- and the hidden terminal problem ([8, 9]).
ingly, vast research has been pushing for improving the QoS support was added with the IEEE 802.11e EDCA
CE
efficiency of the wireless networks specially those based on (Enhanced Distributed Channel Access) and HCF (Hybrid
the IEEE 802.11 standard. Coordination Function) ([3, 10]). Also, the IEEE 802.11e
The IEEE 802.11 standard provides the PHY (Physical) introduced the TXOP (Transmission Opportunity) con-
and MAC (Medium Access Control) specifications for wire- cept which specifies the maximum time for each trans-
AC
less networks. The standard initially provided low PHY mission. EDCA divides packets into classes based on
rates and included two basic MAC functions: a best ef- their type (video, audio, etc.). Then, each class follows
fort DCF (Distributed Coordination Function), and PCF DCF with different configurations (like different backoff
(Point Coordination Function) [7]. PCF is a contention- times) allowing service differentiation based on the pack-
free access scheme where a central controller (normally the ets’ classes. On the other hand, a polling method in HCF
access point, or AP) polls each user for data transmission. can be used to provide QoS ([11, 12, 13]). The AP sched-
On the other hand, DCF is based on a CSMA/CA (Carrier ules users’ transmissions based on their demands.
Sense Multiple Access with Collision Avoidance) scheme
Additional higher rates were also presented by the fol-
lowing generations of the IEEE 802.11 standard. Impor-
∗ Corresponding Author
tant features that allowed for the higher rates include
Email addresses: almehai@yu.edu.jo (Haithem Al-Mefleh), the use of OFDM (Orthogonal Frequency Division Mul-
osameh@yu.edu.jo (Osameh Al-Kofahi) tiplexing) modulation, MIMO (Multiple-Input Multiple-
Preprint submitted to Elsevier May 23, 2018
ACCEPTED MANUSCRIPT
Output) technology, frame aggregation, block acknowl- In this survey paper, we focus on the contention and
edgments, channel bonding [14], and TXOP sharing [15]. polling MAC schemes that are based on the frequency-
Moreover, a new generation, the IEEE 802.11ax draft domain. We attempt to provide a comprehensive survey
([16, 17, 18, 19, 20, 21, 22]), is being developed to of such schemes. Also, this survey includes a review of
further boost the efficiency of IEEE 802.11-based net- recent related proposals that are related to the new IEEE
works for high-dense scenarios. Features of the IEEE 802.11ax draft. First, we explain the concept of frequency-
802.11ax include the OFDAM (Orthogonal Frequency Di- based polling and contention, and we describe how it could
vision Multiple Access) access, MU-MIMO (Multi-User enhance the performance of wireless networks. Then, we
MIMO) method, and in-band full-duplex communications. summarize different frequency-domain based contention
Also, the IEEE 802.11ax draft is endorsing new MAC ac- and polling protocols and their various operations. There-
cess methods to benefit from the advanced physical amend- after, we further compare the surveyed protocols based on
ments. factors including the evaluation methods, number of con-
T
With the ongoing development of the IEEE 802.11 tention rounds, types of conveyed signals, and the purpose
standard, various work is proposed to increase the effi- of each protocol. Additionally, we explain some problems
IP
ciency of different operations of the standard ([19, 23]). and issues that are related to the frequency-domain based
Research identified different sources of overhead and is- schemes. We discuss the effect of diverse factors includ-
sues that would degrade the performance of the network ing the hidden terminal problem, the number of antennas,
CR
([18, 20]). Overheads include the contention time ([24]), the number of subcarriers and subchannels, TDMA-like
the inter-frame spacing ([25]), control packets and headers scheduling, synchronization, coexistence, overlapping net-
([25, 26, 27]), and collisions ([28]). Moreover, results show works, machine to machine networks, and support of QoS
that the performance declines as the PHY rate gets higher (Quality of Service). We also address some directions for
hancements of wireless networks. To reduce collisions and tively. Fifth, we summarize and discuss several features,
the number of time slots per contention, dynamic and challenges, and future directions in Section 6. Finally, we
adaptive algorithms are proposed ([31, 32]). Also, binary conclude the paper in Section 7.
count-down algorithms with jamming signals are utilized
ED
[36] and the possibilities of MPR (Multi-Packet Reception) The IEEE 802.11 DCF utilizes a CSMA/CA with a
([37, 38]) techniques. Also, research exploited pipelined BEB (Binary Exponential Backoff) mechanism as shown
procedures ([39, 40, 41, 42]) where the channel is divided in Fig. 1. Each node randomly chooses a number, say BO,
CE
into different subchannels, and each subchannel is used for within the range [0, CW ], where CW stands for contention
a part of the communication. For example, data exchange window. Then, the node must sense the channel to be idle
could take place in one channel while other nodes continue for a DIFS (DCF Interframe space) period. Then, the
contending on another channel. A similar area of research node must wait for a backoff period of BO time slots. The
AC
considers running backoff per subchannel ([43, 44]). To node decrements its backoff counter by one for each idle
remove (or reduce) the overhead of control frames, attach- slot, and it transmits only when its backoff counter reaches
ment of control information ([45, 46, 47, 48]) exploits some zero. Upon a successful reception, the receiver replies with
of the OFDM subcarriers of the data frames to convey an ACK frame after an SIFS (short IFS) period. On the
control signals (like RTS). Another approach is to replace other hand, collisions occur when two nodes have the same
the whole control frames with short bursts or jam signals backoff value. Each colliding node doubles its CW , picks
([23, 29]). Moreover, an interesting direction of research a new backoff value, and retries to send the packet. The
proposes shifting from time-domain to frequency-domain IEEE 802.11 DCF also supports RTS/CTS mode as shown
([49, 50]) in OFDM-based wireless networks. Results of Fig. 2; short control frames (RTS and CTS) are exchanged
such schemes show much of lessening of the contention by the transmitter and the receiver before the actual data
time and collisions. Also, OFDMA access is considered transmission. The RTS/CTS mode alleviates the time
to allow multiple users to transmit concurrently, and it is wasted by collisions of long frames and reduces the hid-
included in the IEEE 802.11ax draft ([18, 20, 51]). den terminal problem. Finally, the IEEE 802.11e EDCA
2
ACCEPTED MANUSCRIPT
adds service differentiation. Packets are classified based on the frequency domain instead of time-domain. Fig. 3 il-
the traffic type, and each class is given different contention lustrates the difference between both domains. In time-
parameters. For example, priority gets higher for smaller domain algorithms, a node arbitrarily selects the number
CW values. In addition, the TXOP (Transmission Oppor- of time slots to wait before transmitting. Winners are
tunity) configuration parameter is added to limit the time nodes with the least number of idle slots to wait. Colli-
a node can access the channel per contention. sions occur when more than one winner exist (i.e. more
than one node have the same backoff value). On the other
hand, in frequency-domain, a node sends a signal (which
can be a busy tone, a signature, or a normal message) on
a subcarrier (or a subchannel) that is randomly selected.
Figure 1: IEEE 802.11 DCF, basic operation Hence, winners are identified using some rule like nodes
selecting the lowest subcarrier ID. Collisions occur when
T
more than one winner exist following the contention rules.
IP
CR
Figure 2: IEEE 802.11 DCF, RTS/CTS operation
into a number of orthogonal subcarriers, and each subcar- Frequency-domain MAC protocols mainly aim at reduc-
rier is labeled with an ID (identifier). In other words, each ing the time, number of slots, needed to resolve contention.
subcarrier (or a subchannel which is a group of subcarri- However, more slots may be required depending on the
number of contending nodes. Thus, for a high number of
ED
(like T2F [54]), the contention period is only 44.4us for antennas at least (i.e. could transmit and receive simul-
two rounds. In addition, while collision rates highly in- taneously). Of course, it is acceptable to assume having
crease with the number of nodes in DCF, they are stable more than one antenna due to the use of MIMO. In ad-
and low in frequency-domain contention protocols. For dition, the contention process can be of only one round
example, a collision probability of 0.03 is reported in a or multiple rounds. We elaborate on these features and
two-rounds T2F when 52 subcarriers are utilized for con- related issues in Section 6.
tention. Hence, much of the contention overhead is re-
duced when shifting contention into the frequency-domain.
Polling in the frequency domain is generally demon-
strated by Fig. 5. Initially, the AP transmits a polling re-
quest. Then, each node transmits a signal over its own sub-
carrier which is unique and is assigned a priori. The signal
T
can be as simple as a busy tone indicating that the node is
ready to transmit (or receive). Then, the AP could assign
IP
resources accordingly. More details of frequency-domain
polling schemes will be given in Section 5. Frequency-
domain polling aims at reducing the overhead of the inter-
CR
frame space times and polling control frames (especially
when a node is not ready to be polled) in time-domain
polling as shown in Fig. 6.
4.1.1. Bursts
3.2. Frequency-Domain Based Protocols
MCBC (Multi-Carrier Burst Contention) [55] is an elim-
As illustrated in Fig. 7, the surveyed protocols can be ination scheme of R rounds with two time slots (a con-
identified as contention or polling schemes. Moreover, we tention slot followed by a feedback slot) per round. Also,
organize different contention schemes based on the pur- frequency-domain contention is employed during each con-
pose of each protocol: contention, QoS support, and col- tention slot. MCBC operation with 3 rounds is illustrated
lision resolution. Also, each node applying one of these in Fig. 8, and it is summarized in the following points in
protocols would send a signal to contend for the chan- order:
nel (in contention schemes) or to request to be polled (in
polling schemes). The signal can be a message, a burst, or • After the channel is sensed idle for a DIFS period,
a signature. Moreover, while some protocols are designed each contending user becomes a nominee with a prob-
for nodes with a single antenna (i.e. can only receive or ability of p. In Fig. 8, nominees of Round 1 are n1 to
transmit at a time), others assume that each node has two n8.
4
ACCEPTED MANUSCRIPT
T
words, each user can transmit and receive at the same
time. For the first round, each user randomly selects one
IP
of the available 52 subcarriers (f 1 to f 52) and transmits
a burst on the selected subcarrier (say f m). At the same
time, the user receives all signals and decides the lowest
CR
Figure 8: MCBC operation subcarrier (fmin ) that is active. The user continues to
contend when its selected subcarrier is the same as fmin
(i.e. when fm = fmin ). Fig. 9 shows that fmin = f 2 in
• Round 1: slot 1, and thus the winners of the first round are n1 and
Figure 9: T2F
Round 1 in our example of Fig. 8.
• The operation of Round 1 is repeated for all subse- T2F allows all users who activate one of the least K
quent rounds. At end, a winner of the last round subcarriers during the first round to be winners of that
AC
starts to transmit an RTS frame. Accordingly, col- round. After the second round, these users transmit in
lisions occur when more than one user win the con- a TDMA-like schedule in the order of their bursts. To
tention, and a successful transmission completes oth- illustrate we provide an example in Fig. 10 with K = 2.
erwise. In Fig. 8, n1-n3 are winners of Round 1, n1 Then n2 and n3 are winners of the first round, and they
and n2 are winners of Round 2, and the winner of the would transmit in the order n2 and then n3.
last round is n1 with no collisions. Back2F (Back to Frequency) protocol [59] is another
frequency-domain contention scheme whose operation has
In [56], MCBC is evaluated for ad hoc networks in indoor much resemblance to that of T2F. An important difference
fading environments with no direct line of sight. Rayleigh is that Back2F approximately emulates the IEEE 802.11
fading is modeled, and energy detection (instead of pream- fairness by adjusting backoff values (selected subcarriers)
ble detection) is assumed for sensing. To enhance perfor- after losing a contention using deduction. For example,
mance under fading, the authors suggest to use shorter assuming the scenario in Fig. 10, the highest frequency of
frames, to enable RTS/CTS mode, and to activate two or winners in the first round is f 3. Then, a losing node like
5
ACCEPTED MANUSCRIPT
T
Figure 10: The TDMA-like transmission in T2F
IP
n1 running Back2F would burst f 1 (or f 4 − f 3) during
CR
the first round of the next contention period.
REPICK (REversed contention and PIggy-backed
ACK) [29] divides the subcarriers into identification and
contention subcarriers. Moreover, each user is assumed to
have two antennas with the ability to receive and transmit
concurrently. Also, each user is uniquely assigned one of
the identification subcarriers. The operation of REPICK
is summarized by the example given in Fig. 11 and also by
US
AN
the following:
carrier assigned to the sender. In addition, the re- cause it allows only one user to access the whole chan-
ceiver activates the senders next contention subcar- nel. Even for a small packet, the user must contend for
rier (fnext ). In other words, the receiver contends on the channel wasting the network time (the backoff time).
behalf of the sender to mitigate the hidden terminal The problem is augmented for higher PHY rates which are
CE
that is called transmission retreat. Each sender keeps K subcarriers per subchannel are reserved for contention.
track of the number of missing ACKs in a variable Then as shown in the example given in Fig. 12, users could
(say counter). When an ACK is not received, the contend for the channel concurrently, and later they would
sender increments the value of counter and defers ac- transmit data simultaneously to enhance the performance
cessing the channel for a number of rounds that is ran- of the network. Operations of FICA are summarized in
domly chosen from the range [0, counter]. Thus, hid- the following:
den nodes probabilistically contend at different times
decreasing the collision probability. To illustrate, as- • After the channel is idle for a period of DIFS, con-
sume n1 and n2 are two hidden nodes with the same tention of one round in the frequency domain is fol-
values of counter variables (say 3). Then, n1 and lowed during a period that is called M-RTS period.
n2 may randomly delay their contention for 3 and 0 Each user randomly chooses one subcarrier (say f m)
rounds, respectively (i.e. n2 does not refrain from from the K subcarriers within the subchannel to ac-
contention). Thus, n2 would have a higher chance to cess. Then the user sends a special signal (works as
6
ACCEPTED MANUSCRIPT
an RTS) on its selected f m subcarrier as shown in The paper in [62] is also based on FICA. The authors
the M-RTS period of Fig. 12. For each subchannel, a propose to adapt both the number (like with FICA) and
receiving node would reply by sending a special signal the width of accessible subchannels. Adapting the chan-
(operates as a CTS) on the highest frequency during nel width would reduce the time wasted as illustrated by
the M-CTS period signifying who can transmit for Fig. 13. The paper provides theoretical analysis and op-
that subchannel. timizations of the number and the width of subchannels.
On the other hand, Aileron ([63, 64]) extends FICA by
• In FICA, the special signals (those operating as RTS adding frame aggregation. In other words, each node can
and CTS) are spread over more than one subcarrier to transmit more than one frame following the contention
include additional information in contention. Further- period. Hence, the utilization is improved for scenarios of
more, different subcarriers are replicated four times non-homogeneous traffic and network status at different
for reliability. For example such information consists nodes.
T
of a tag identifying the type of the signal (32 sub-
carriers), a hash vector to identify the receiver (160
IP
subcarriers), a NAV (Network Allocation Value) value
(64 subcarriers), and guard subcarriers (32 subcarri-
ers). Thus, FICA works only for OFDM architectures
CR
with a very large number of subcarriers.
domain for one round after sensing the channel idle for
some time (like DIFS). Time-domain contention follows
only when multiple users are detected. The operation of
WiFi-BA is demonstrated in Fig. 14 and is summarized as
ED
follows.
In [61], the authors propose a scheme based on FICA listen to the channel as WiFi-BA assumes multiple
to increase the supported number of users in a WLAN antenna devices. Only eight subcarriers are used for
network. Subchannels are only used during data trans- contention (k = 8), and thus collisions are expected
missions. In other words, contention subcarriers are not to be high. Therefore, when a node detects collisions
AC
related to a specific subchannel. During M-RTS, a number by sensing high magnitudes on its inactive subcarri-
of subcarriers are allocated for contention over the whole ers, a binary-count down approach (in time-domain)
channel. Later, a successful user (a user with no collisions follows to resolve collisions.
in M-RTS) is assigned one of the available subchannels by
the AP. Obviously, the number of users who can trans- • Time-Domain: Starting from the MSB (Most Signif-
mit is limited by the number of subchannels. Another icant Bit) of the binary number, the user transmits
difference from FICA, the proposed scheme requires that its code (for a duration of one symbol) when the bit
each user contends for the channel with a probability of is ′ 1′ . Otherwise, the user listens to the channel (for
p. The authors also attempt to analytically find optimal a duration of one symbol), and it aborts contention
p and optimal number of contention subcarriers, and they when detecting any signals. Once no collisions are
suggest a scheme applied at the AP to penalize violating detected (users are capable of transmitting and lis-
users. Finally, the paper presents neither an evaluation tening at the same time), or all bits are checked, the
nor a comparison. user would start data transmission.
7
ACCEPTED MANUSCRIPT
ness algorithm.
FC-MAC [75] is proposed for a one-hop OFDM wireless
network. Also, FC-MAC assumes multiple antennas and
a maximum of k subcarriers (f 1 to f k). At first, each
node is assigned a distinct signature which is a PN se-
quence number. When a node has data to transmit, it
would randomly generate a contention vector (say CV ) of
k bits. Then, the node would send (with one antenna)
its signature on a subcarrier when the corresponding bit
in the CV is ′ 1′ . To improve the reliability of signatures’
detection, the length of a PN sequence is set to 20 bytes.
Simultaneously, with all other antennas, the node would
T
listen to the channel receiving a composite of all transmit-
ted contention vectors. Because of the use of PN sequence
IP
numbers, match filters can be used to detect different sig-
Figure 14: WiFi-BA natures. A node abort contention in case it is unable to
detect any signature. Accordingly, the winner is the node
CR
who used the minimum contention vector. In [76], a very
A similar approach is also provided in [66] and [67]. FD-
similar approach with PN sequences is also introduced.
OOAT (Frequency-Domain On-Off Accumulative Trans-
mission) [68] is another similar approach but only one
4.1.3. Messages
frequency-domain round of contention is employed, the
transmitted signal is repeated over multiple subchannels,
an AP is assumed, and thus multi-user detection is possi-
ble.
US In Generalized CSMA/CA ([77, 78]), an OFDMA chan-
nel is divided into a number of subchannels. A node with
data to transmit would randomly select a time-domain
backoff value. Then, for each time slot, the user decre-
AN
In [69], the authors define a problem with WiFi-BA and
ments its backoff value by the number of idle subchan-
provide a solution. They state that WiFi-BA introduces
nels. When the backoff counter reaches zero, the node
a new type of collisions (collisions between data and the
arbitrarily selects one of the idle subchannels to start its
contention signals). To illustrate, assume nodes A and B
own transmission. Hence, the efficiency is affected because
select ”0110” and ”0010” respectively. Then both nodes
M
for each idle round. When the timer is zero, the station both domains (time and frequency) in contention. The
selects randomly one of the RUs during the reached round. time-domain backoff procedure is equivalent to that used
Then, the station would send a bandwidth request includ- in the IEEE 802.11. However, lower values of contention
ing at least the queue size. Collisions occur when multiple window are used (for both CWmin and, CWmax ). A node
stations transmit during the same RU. Thereafter, the AP also randomly selects the band (center frequency and the
transmits a TF (Trigger Frame) which contains a transmis- bandwidth) to be used for its data transmission. When col-
sion schedule (built by the AP based on the successfully lisions are detected, in addition to the time-domain back-
received requests) to be followed. Such schedule would in- off, a node also would backoff in the frequency-domain. In
dicate which stations are allowed to transmit and the RUs other words, the node changes the transmission band with
that should be used by each of these stations. Finally, high probability. On the other hand, following a successful
the AP sends an M-BA (Multiple Block Acknowledgment) transmission, the node would change its band with a small
frame. probability. Here, it is assumed that no other nodes are
T
using that band, and thus the successful node could keep
using the same band. Finally, to reduce waste of band-
IP
width, each node also halves its bandwidth with a small
probability after sensing another node’s transmission. In
other words, the node would free some bandwidth to be
CR
used by others.
FD-Backoff [85] is a one round frequency-domain backoff
scheme that follows the procedure shown in Fig. 17. With
Figure 15: IEEE 802.11ax new contention scheme
FD-Backoff, each node randomly picks one of the data
OMAX (OFDMA based Multiple Access for IEEE
802.11ax) [82] utilizes contention in both domains (time
and frequency) as illustrated in Fig. 16. The first round
US subcarriers and sends its RTS packet on to that subcar-
rier. The receiver, as identified in the RTS frame, replies
with a CTS packet. Then, data and ACK frames are ex-
changed. The authors point out that there is no much
AN
is of fixed size time-domain contention (the Backoff stage overhead added because of transmitting the RTS packet
in Fig.16). Like DCF, each node randomly selects a back- over only one subcarrier (the transmission time of the RTS
off time (a number of slots). However, the backoff time packet with one subcarrier is approximately the same as
is decremented by the number of available subchannels when using the whole channel with the IEEE 802.11a/g/n
M
for each time slot. When the backoff time becomes zero, standards). The work of [86] is similar to FD-Backoff but
the node transmits an RTS frame on a randomly selected assumes that each node has two antennas. Another differ-
subchannel. With OMAX, the whole channel is sensed. ence, the work of [86] provides a two-rounds version like
In other words, once a transmission is seen, others de- with T2F (see Fig. 9); the nodes choosing one of the least
ED
fer accessing the channel until the next contention period. L frequencies during the first round would contend in the
Accordingly, following the concurrent RTS transmissions, second round. In [87], the authors show how to detect and
the AP sends back a G-CTS (Group-CTS) frame indi- penalize violating nodes when the protocol of [86], and
cating which nodes can use which subchannels. Nodes thus FD-Backoff, is applied. Likewise, Multiband CSMA
PT
transmit their data concurrently, and the AP sends back with RTS/CTS [88] applies a similar approach to that of
a G-ACK (Group-ACK) frame. C-OFDMA (Concurrent FD-Backof. However, each node also waits for a random
OFDMA) [83] allows nodes to transmit concurrently like number of time slots following the DIFS period. Hence, a
with OMAX. However, individual subchannels are sensed
CE
At the same time, polled nodes use these special messages be resolved allowing a node to distinguish who is trans-
to inform the AP about their uplink data. Moreover, the mitting. However, a signature usually takes a longer time
un-polled nodes could use the unassigned subchannels to than a burst, or it requires many subcarriers. For example,
send similar messages when they have any uplink data. while the signature length in FC-MAC (a signature-based
Here, random access in frequency-domain is utilized. Sec- scheme) is about 42us [75], a burst is only 5.2us in T2F.
ond, concurrent downlink transmissions take place. Using Hence, the contention period could be longer in signature-
assigned subchannels, the AP sends downlink data to each based schemes. In message-based schemes, the contention
node which then replies with an ACK. Third, the AP polls period is longer as messages are used. Moreover, colli-
nodes for uplink data specifying which nodes can transmit sion rates are higher because backoff values are selected
and on which subchannels. Then, all uplink data trans- from subchannels instead of subcarriers. Therefore, most
missions occur concurrently. Finally, the AP sends a block of such protocols additionally adopt a time-domain con-
acknowledgement. tention scheme.
T
4.2. QoS Support
IP
4.2.1. Bursts
In [57], QoS features are suggested for MCBC (see sub-
CR
section 4.1). Basically, different subcarriers are assigned
for different priorities and emergency messages during the
contention slot of the first round. In addition, a user would
activate both contention and feedback subcarriers to as-
Figure 18: HMAC
the AP replies with multiple ACKs for successful trans- Medley [91] is an OFDMA-based MAC designed to pro-
missions. For the next random-access period, each node vide a delay-fair access to accommodate different service
with a successful reservation is assigned the same RU. On requirements in a WLAN. Medley is built on FICA (see
the other hand, contention is performed only with the un- subsection 4.1) as shown in Fig. 12. Medley makes changes
ED
for a high number of nodes. However, MCBC has a longer Fig. 19, subchannels in M-RTS are divided into groups,
contention time and allows only one winner per contention. and each group is further divided into n blocks (B1,
On the other hand, T2F allows TDMA-like schedules B2,...,Bn); where a block is a number of subcarriers. Each
CE
which would enhance the performance. REPICK works for group is associated with a different service class. Conse-
only a low number of nodes because it assigns a part of the quently, each node in a group represents its demand (the
subcarriers for identification purposes. On the other hand, number of required subchannels) as an n-bit binary num-
WiFi-BA uses time-domain once a collision is detected in ber. Then, starting from the MSB (most significant bit),
AC
the frequency-domain competition (the first round). Thus, if the bit is ′ 1′ , the node sends a burst over a randomly
the performance with WiFi-BA may suffer as collisions can selected subcarrier of Bn. The operation is repeated for
be severe in the first round as shown in [54]. FICA is all subsequent bits. Later in the M-CTS period, the AP
similar to T2F but also allows channelization for parallel resends the bursts seen during the M-RTS, and thus nodes
contention and transmissions. Accordingly, overheads of learn subchannels’ allocations. Finally, Medley utilizes an
interframe spacing and contention are reduced. However, AIMD (additive-increase, multiplicative-decrease) proce-
FICA assigns a part of the available subcarriers to convey dure to control the number of used subchannels in order
additional information in contention. Also, FICA follows to reduce collisions.
one round of contention. Hence, FICA may have a higher WT2F (Weighted T2F) [92] changes the first round of
collision rate when compared to T2F with a comparable T2F (see subsection 4.1) in order to provide service dif-
OFDM architecture. Compared to burst-based schemes, ferentiation. Different priority levels are assigned various
signature-based schemes are expected to provide higher ranges of subcarriers. Accordingly, each user randomly se-
successful rates. This is because collided signatures may lects the subcarrier to activate from the range related to
10
ACCEPTED MANUSCRIPT
T
Figure 19: Grouping of subcarriers in Medley
4.2.3. Messages
IP
its priority level. As illustrated in Fig. 20 while high pri- QoS-OFDMA [95] extends OMAX (see subsection 4.1)
ority stations select from the range [1, F ], the low priority to support QoS. After the backoff stage (in Fig. 16), a node
CR
users choose from the range [1, F max]. Finally, propor- with a higher priority sends more than one RTS frame.
tional fairness of the two-priority levels case is analyzed. When one of these RTS frames is received successfully by
the AP, the AP assigns the whole channel to the high
priority node.
CBQO (Channel Bonding based QoS-aware OFDMA)
its access attempt and starts over when the primary chan-
4.2.2. Signatures nel is idle for a DIFS period. Once the backoff counter
QoS-Fi [93] extends D-Fi (see subsection 4.1) to sup- reaches zero, the node sends its RTS on a randomly se-
ED
port QoS like the IEEE 802.11e EDCA access mechanism. lected subchannel. For QoS traffic, the node would send
Merely, each station is assigned four signatures with dif- another RTS on another random subchannel. Moreover, a
ferent lengths. Accordingly, a station uses a longer sig- node also transmits a tone signal on the first primary sub-
nature for a higher priority packet. This is similar to the channel to preserve synchronization; i.e. prevent nodes
distributed prioritized-mechanism provided in the IEEE
PT
lute priority. In WFC, different ranges of subcarriers are The surveyed QoS schemes are mostly extensions to con-
allocated for different priority levels. WFC assumes that tention methods, and thus they inherit similar limitations.
each user has a unique signature, signatures are known by Basically, they assign different resources (like subcarriers)
all users, and each user can listen and receive simultane- to various QoS levels. With bursts and signatures, sepa-
AC
ously. Moreover, WFC follows two rounds of contention rate ranges of subcarriers or longer signatures are utilized
like that of T2F as shown in Fig. 9. The following sum- for different types of traffic respectively. Alternatively, in
marizes WFC and its differences from T2F: message-based schemes, priority can be assured by sending
more requests on different subchannels. Accordingly, per-
• Round 1: Each user randomly activates a subcar- formance of lower priority traffic may be affected as they
rier from a range that is based on priority. Assum- contend with a lower number of resources. This in turn
ing only two levels, the subcarrier ranges [1, S] and may affect the performance of all nodes in the network,
[F + 1, F max] are for high and low priority levels and hence more resources are required for scenarios with
respectively; where (F < S < F max) as shown in higher densities.
Fig. 21. At the same time, each user listens to the
channel to conclude who the winners are (those acti- 4.3. Collision Resolution
vated the minimum subcarrier). All winners proceed All surveyed collision resolution schemes use messages.
to the next round. Fast Retrial [97] proposes frequency backoff for collision
11
ACCEPTED MANUSCRIPT
resolution in OFDMA wireless networks; see Fig. 22 for delay performance especially for nodes not involved in the
an illustration. When a user’s request collides, the user re- collision resolution.
sends the request message on a randomly selected ranging
subchannel (a group of subcarriers) during the next slot.
5. Frequency-Domain Based Polling Schemes
Finally, when a maximum number of retries is reached,
the user switches back to time-domain contention. Note In this section, we review frequency-domain based
that Fast Retry assumes that users know about collisions polling schemes.
during the same slot used for transmitting requests. An-
other similar approach is OARSB (Opportunistic access 5.1. Bursts
with random subchannel Backoff) [98]. However, OARSB
In PAMAC (Physical layer Assisted MAC) [100], an AP
proposes an improvement by which nodes choose from the
broadcasts a request message. Then each user activates
best L subchannels (a subset of all subchnnels available).
its preassigned subcarrier signifying its status (no data,
T
Analysis and simulation studies are provided in both works
low priority, medium priority, or high priority). To con-
of Fast Retry and OARSB. In OFDM-RR (OFDM reser-
IP
vey the status information, the subcarriers are distributed
vation random access) [53], the authors propose a similar
over four non-contiguous OFDM symbols. The user sends
approach for OFDMA femtocells. However, only one node
the first bit on symbols 1 and 2, and the second bit on
utilizes the whole channel after resolving collisions of re-
CR
symbols 3 and 4. For example, sending ”11” means high
quest messages.
priority traffic. Based on the parallel received signals, the
EBF (Exponential Backoff in Frequency) [99] is pro-
AP broadcasts a single burst subcarrier to indicate which
posed for femtocells with small number of nodes. A node
users can transmit.
could transmit multiple packets (all are of equal and fixed
size) concurrently using different channels. The access
method simply starts by having a node transmit over
all channels. If a collision occurs (a collision in any of
the selected channels), the node divides channels into two
US HT-MIMO [101] is a MAC protocol that considers
a MIMO and OFDM based WLAN. HT-MAC applies
polling in both uplink and downlink as show in Fig. 23
and Fig. 24 respectively. Basically, for IEEE 802.11n with
AN
4 antennas at the AP, thirteen subcarriers are assigned for
groups. Then, the node probabilistically picks one of the
each antenna at the AP, and these subcarriers are divided
two groups. The process is repeated as long as there are
into: 8 subcarriers to identify a node, 2 subcarriers to spec-
collisions and the number of used channels is greater than
ify the antenna on the node, and 3 subcarriers to define the
one.
type of a control message. By activating or deactivating
M
Remarks:
AC
T
IP
CR
Figure 25: Rapid polling in DOMINO
slot time which consists of the data packet time and a Figure 28: Uplink Polling
synchronization time (2 µs). Finally, an aggregate ACK
is sent back by the AP. PFDP-MAC (Priority-aware Fre- In [105], a similar polling approach is proposed for up-
quency Domain Polling MAC) [103] boosts FDP-MAC to link transmissions in an OFDMA IEEE 802.11ax network
ED
support different priorities of data. Each user is assigned where messages are used to transmit queues feedback. In
a number of subcarriers; one subcarrier for every level of addition to the size information, conveyed feedback mes-
priority. Also, a heuristic approach is provided to mini- sages provide channel estimations. Finally, similar strate-
mize the probability that packets of a given station do no gies are found in ([106, 107, 108]) with throughput analy-
PT
Remarks:
CE
5.2. Messages
6. Summary, Discussion, and Future Directions
In [104], polling in the frequency-domain is proposed for
both uplink and downlink directions. In both cases, the In this section, we summarize and discuss several fea-
AP initially broadcasts a polling frame indicating selected tures and issues of the frequency-domain contention and
13
ACCEPTED MANUSCRIPT
polling MAC protocols in subsection 6.1. In addition, we parallel. Then, a receiving node uses a filtering mechanism
address some possibilities for future work in subsection 6.2. to recognize all signatures. Hence, winners of contention
or order of transmissions can be identified. For example,
6.1. Summary and Discussion in Fig 29, we assume the signatures ”0001” and ”1000” for
nodes A and B respectively, and only f 1...f 4 are the avail-
6.1.1. Summary able subcarriers. Then, A activates f 4 and B activates f 1.
In Table 1, we compare different protocols included in The AP receives the burst ”1001”, then applies filtering to
this survey paper. First, the table specifies the purpose recognize both original signatures of A and B. Conse-
of each protocol (contention, QoS, polling, and collision quently, the AP could assign resources to different nodes
resolution). Second, the table gives the type of used sig- based on the recognized signatures. On the other hand, a
nals (burst, signature, or message). Third, we indicate PN sequence number is a binary number that is transmit-
the underlying technology assumed (OFDM, or OFDMA). ted using modulation but with no preambles. Matching
T
Fourth, the number of rounds used in contention schemes filters are required to recognize different signatures trans-
is stated. Finally, the table shows what type of evalua- mitted on every subcarrier. For signature schemes, as-
IP
tion is used (simulation, analysis, and real implementa- suming a unique signature per node, collisions may happen
tion). Note more than one evaluation method may be when signatures cannot be recognized correctly. For exam-
available. Different protocols provide real implementations ple, each of two contending nodes may incorrectly assume
CR
using Sora software radio (like in FICA [60]), FPGA (like itself a winner because of detection errors, errors in filter-
in PAMAC [100]), or USRP/GNURadio (like in T2F [54]). ing, etc. Finally, messages are simply control (like RTS)
On the other hand, many protocols (specially OFDMA- or data frames transmitted based on the underlying PHY
based schemes) simply assume the feasibility of the im- layer. Collisions of messages occur when transmissions of
plementation based on previous works. Table 2 summa-
rizes different implementation platforms. Note that COTS
(Consumer Of The Shelf) devices are not suitable to imple-
ment these protocols as they deviate greatly from the stan-
US frames overlap in time and frequency. Sending messages
directly in contention is simpler but has the drawback that
their collisions cause larger overheads.
AN
dard 802.11 operations (e.g., D-Fi [72]). Flexible open-
source drivers like the Atheros ath9k enable the user to
fine tune and control almost every detail in the operation
of the 802.11 protocol, but they do not allow to change
M
In general, a node would send a burst, a signature, or a The following are some issues related to the burst and
message. A burst is a tone signal that is transmitted with signature types:
no information [23], and it can be as short as one OFDM
symbol. A receiver is required only to recognize a burst • There is a limit on the maximum number of parallel
AC
by energy detection. In contention, a collision occurs when signatures that can be recognized. This is due to lim-
at least two nodes activate the same subcarrier in the last itations of the filters used and the effects of parallel
contention round. On the other hand, in polling each node transmissions on the channel state.
is normally assigned a unique subcarrier. Hence, the node • Each node is assigned a unique signature. Hence,
simply activates its subcarrier informing the AP that it is there is a need to assign and maintain signatures
ready to transmit (or receive) as in FDP-MAC [50]. Alter- based on different activities (like joining and leaving
natively, a signature is a sequence of bits that can be used the network) specially for ad-hoc networks where no
to uniquely identify a node. The bits are transmitted ei- central control exists.
ther as bursts (like in D-Fi [73]) or PN sequence numbers
(like in FC-MAC [75]). First, with bursts, a subcarrier • When bursts are involved, nodes are required to gen-
is assigned for each bit in the signature. Thereafter, a erate and detect burst signals on individual subcarri-
node sends a burst on the subcarrier when its signatures ers. Also, the performance is affected by false alarms
corresponding bit is ′ 1′ . All signatures are transmitted in (like detecting noise as a burst, or sensing an idle state
14
ACCEPTED MANUSCRIPT
Table 1: Summary
Sim: Simulation, Imp: Implementation
Protocol Usage Request Rounds Technology Sim Analysis Imp
Fast Retrial [97] CR Message 1 OFDMA Yes Yes
OARSB [98] CR Message 1 OFDMA Yes Yes
EBF [99] CR Message 1 OFDMA Yes Yes
OFDM-RR [53] C, CR Message 1 OFDMA Yes Yes
MCBC [55, 56, 57] C Burst 3 OFDM Yes Yes Yes
T2F [54, 58] C Burst 2 OFDM Yes Yes
T
WT2F [92] C, QoS Burst 2 OFDM Yes
Back2F [59] C Burst 2 OFDM Yes Yes
IP
REPICK [29] C Burst 1 OFDM Yes Yes
FICA [60, 30] C Burst 1 OFDM Yes Yes
D-Fi [72, 73] C Signature 1 OFDM Yes Yes
CR
[61] C Burst 1 OFDM Yes
[62] C Burst 1 OFDMA Yes Yes
QoS-Fi [93] C, QoS Signature 1 OFDM Yes Yes
WiFi-BA [65] C Burst 2* OFDM Yes
[66]
[67]
FD-OOAT [68]
Medley [91]
C
C
C
C, Qos
US
Burst
Burst
Burst
Burst
2*
2*
1
1
OFDM
OFDM
OFDM
OFDMA
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
AN
TF-CSMA/CA [84] C Message 2* OFDM Yes Yes
FD-Backoff [85] C Message 1 OFDM Yes Yes
[88] C Message 2* OFDM Yes
FC-MAC [75] C Signature 1 OFDM Yes Yes
M
15
ACCEPTED MANUSCRIPT
T
less medium activities could be sufficient to synchronize
DiFuse [70] USRP/GNURadio
nodes. For example, nodes contend for the channel follow-
PAMAC [100] Virtex-IV FPGA/a front-end radio
IP
ing the reception of a trigger frame sent by the AP in an
DOMINO [102] USRP/GNURadio
IEEE 802.11ax network. In addition, for bursts, transmit-
ting for a somewhat longer duration is utilized to provide
CR
Table 3: Simulation Environment synchronization in schemes like T2F specially when no AP
Protocol Simulation Environment is assumed. In T2F, a node transmits a jamming signal on
EBF [99] NS-3 the selected subcarrier for a period of (TF F T + 2Pd ) con-
FC-MAC [75] Qualnet sisting of one FFT and twice the maximum propagation
[76]
OMAX [82]
QoS-OFDMA [95] NS-2
CBQO [96]
Qualnet
NS-2
NS2
US delay (Pd ).
Frequency synchronization is also essential to prevent
misdetections. This has been addressed by using subcar-
rier spacing. For example, only a small subset of subcarri-
AN
HMAC [89] Simulation programmed with C++ ers is used to have a high frequency spacing in MCBC; only
[80] Matlab 6 subcarriers for contention. Also, nodes may synchro-
OJRC-MAC [90] NS-2 nize following feedback from an AP. Finally, in [109], the
PAMAC [100] Qualnet authors argue that frequency synchronization is hard to
M
• Since any node could generate bursts, security should into subchannels. Then, all or a subset of the subcarri-
be considered for protocols that are based on using ers, or subchannels, are exploited for contention or polling.
bursts. A harmful node may simply send a burst on Assume that B is the channel bandwidth with k total sub-
the highest priority subcarrier all time preventing le-
CE
Some schemes, like T2F [54], assume that each node ers. Hence, the lower the value of m (or the higher the
could transmit and receive simultaneously. Such assump- value of c), the higher the delays when messages or signa-
tion is reasonable due to the implementation of MIMO tures are transmitted in a frequency-domain contention or
technology. Hence, each node could contend and deter- polling protocol. In addition, for contention protocols, the
mine the result of contention through the same slot. With higher the number of subcarriers (or subchannels) used
multiple antennas at each node, the transmitted signal in contention, the lower the collision rate. The lowest
could affect the detection of signals at the receiving an- collision rate could be achieved when all subchannels are
tenna. The self-transmitted signal is strong enough that it used in contention and m = 1. This is because there is a
could leak to adjacent subcarriers. Frequency spacing can lower probability that two nodes choose the same subcar-
be utilized to tackle this problem as in T2F. On the other rier (or subchannels). Finally, the number of subcarriers
hand, different schemes, like MCBC [55], are designed for utilized in frequency-domain have an impact on the relia-
nodes with a single antenna. In other words, each node bility of the system. Using a higher number of subcarriers
either transmits or receives at a time. In such scenarios, would increase the challenge of proper detection (due to
16
ACCEPTED MANUSCRIPT
Table 4: Analysis
Protocol Analysis
Fast Retrial [97] Delay, Throughput
OARSB [98] Delay, Throughput
OFDM-RR [53] Throughput
MCBC [55, 56, 57] Delay, Throughput
REPICK [29] Throughput
EBF [99] Delay
D-Fi [72, 73] Collision probability
[61] Optimal attempt probability/number of bands
[62] Throughput
[67] Collision probability
T
FD-OOAT [68] Robustness to delay
TF-CSMA/CA [84] Quick convergence, Staying in good states
IP
FD-Backoff [85] Efficiency gain
FC-MAC [75] Success/collision probability
CR
WFC [94] Throughput
OMAX [82] Throughput
C-OFDMA [83] Throughput
Generalized CSMA/CA [77, 78] Throughput
[79]
[80]
OJRC-MAC [90]
HT-MIMO [101]
US
Throughput
Throughput
Throughput
Throughput
AN
[105] Throughput
the increased sensitivity to frequency synchronization, and A protocol with feedback slots, like MCBC, solves the
power leakage into adjacent subcarriers), and it would in- problem to some level. After feedback, each node knows
M
crease the time to process all subcarriers [65]. about any hidden transmissions. In our example, the feed-
back slot conveys that f 1 and f 3 are activated. Thereafter,
B recognizes that it has lost the contention. However, B
6.1.6. Hidden Terminal Problem
ED
time because of undecodable packets. A node cannot tell Fig. 33, then the winning node (say n1) starts to transmit
when its turn to transmit when it is unable to decode one and the other node (n2) waits its turn. However, n2 lis-
of the transmitted packets. Another cause of unexpected tening (using CCA or Clear Channel Assessment) to ch2
problems is the hidden terminal problem. Let us consider may be unable to detect the transmission of n1 on a differ-
a T2F network with three nodes: A, B, and C. Assume ent channel (ch1). Thereafter, n2 could go into contention
A and C are hidden nodes, and A, B, and C burst f 1, period causing a collision with the ongoing transmission of
f 2, and f 3 respectively. Hence, A assumes that it should n1. Third, one of the aforementioned two scenarios would
transmit first, and B assumes it is the second to transmit. occur when one node chooses from the shared subcarriers,
However, C also assumes it is the second node to trans- and the other node selects from the unshared subcarriers.
mit as it cannot overhear the transmission of A. Then, A Accordingly, frequency-domain based schemes could fail
starts to transmit. At the same time, C would go into con- for OBSSs and would be unfair to nodes in the overlap-
tention as it may find the channel idle longer than DIFS ping regions.
T
period. Hence, the result is a collision at B, and B misses
its turn to transmit.
IP
6.1.8. QoS Support
The AP would provide QoS by building appropriate
CR
schedules when polling is activated. Alternatively, dif-
ferent methods are applied to achieve service differentia-
tion in some of the frequency-domain contention schemes. Figure 31: An example of two overlapping networks
First, QoS is realized by assigning different ranges of con-
tention frequencies per a QoS level like in WFC [94]. Sec-
US
ond, higher or longer signatures are applied for higher pri-
ority packets like in QoS-Fi [93]. Third, a node may trans-
mit more than once per contention emphasizing a higher
AN
priority. For example, a QoS-OFDMA [95] node trans-
mits more than one RTS frame for a higher priority traffic.
However, in all aforementioned methods, the range of con-
tention subcarriers (or subchannles) could be insufficient,
M
would be operating on overlapping channels. As a result, not backward compatible with the IEEE 802.11. Hence,
we expect that frequency-domain MAC schemes would fail they would cause severe problems for coexisting nodes.
in such scenarios. To illustrate consider the example of two For example, to compete for the channel, T2F nodes al-
OBSSs as shown in Fig. 31; AP 1 and n1 utilizing chan- ways send bursts after the channel is sensed idle for DIFS.
nel ch1 with (f 1...f 6) subcarriers, and AP 2 and n2 using Consequently, an IEEE 802.11 node may never be able
channel ch2 with (f 5...f 10) subcarriers. Here, ch1 and to access the channel. The IEEE 802.11ax protocol (and
ch2 overlap as they have shared subcarriers (f 5 and f 6). hence all related enhancements) is backward compatible as
Accordingly, the following scenarios are possible. First, the AP first contends for the channel to initiate the new
when both nodes select from the non-shared subcarriers random access scheme as illustrated in Fig. 15.
as illustrated in Fig. 32, then each of them is unaware of In addition, research and industry are considering the
the other node contention. As a result, both nodes start coexistence of WLANs and the LTE in the unlicensed 5G
to transmit resulting in collisions at APs. Second, when spectrum (LTE-U/LAA) ([22, 111, 112, 113, 114, 115, 116,
both nodes select from the shared subcarriers as shown in 117, 118, 119, 120]). Mainly, the LTE technology is the
18
ACCEPTED MANUSCRIPT
T
els. Accordingly, further improvements are required. We
also argue that integrated polling and contention in the
IP
frequency-domain should be analyzed to support mMTC
and cMTC networks.
Full-duplex technology also is expected to be an impor-
CR
Figure 33: An example: shared subcarriers tant part of next-generation wireless network [124], and
it would improve the delay performance and efficiency in-
cluding the case of short packets [125]. Recently, differ-
one invading the WiFi spectrum, so usually, it is the duty ent proposals exploit contention in the frequency-domain
of LTE systems designers to be concerned with the co-
existence with WiFi not the other way around. That is
why LTE and WiFi coexistence schemes are out of the
scope of our survey. In summary LTE-U/LAA introduce
US instead of time-domain for full-duplex MAC protocols
([125, 126, 127, 128]). For example, in [127], each node ac-
tivates a subcarrier for one round following a DIFS period.
Thereafter, the AP specifies the primary and secondary
AN
coexistence solutions like LBT (Listen Before Talk), CCA transmitters. Also, a one round similar to T2F is uti-
(Clear Channel Assessment), and CSAT (Carrier Sensing lized to resolve contentions in the full-duplex MAC proto-
Adaptive Transmission: following on/off periods based on col proposed in [126]. Moreover, cognitive radio solutions
channel status). On the other hand, frequency-domain are crucial to fairly share the wireless bandwidth in future
M
operations are more aggressive than in the legacy IEEE networks [129]. Frequency-domain contention can be used
802.11. With frequency-contention, nodes sense the chan- to reduce the time required by different operations in cog-
nel to be idle for almost a fixed and shorter time, then nitive wireless networks. In each of ([130, 131]), contention
they send concurrent signals over the channel to compete, and cooperative sensing are performed concurrently in the
ED
and finally they send data. Hence, this could impact the frequency-domain. In [132], the idea of frequency-domain
performance of LTE network. As a result, LTE and WiFi contention is applied to accelerate the cooperative sens-
coexistence approaches should be reevaluated in the case ing and reporting in cognitive industrial wireless networks.
of frequency-domain schemes and IEEE 802.11ax. Further analysis and improvements are possible, and real
PT
number of devices and cMTC (critical Machine-Type • Contention and polling in frequency-domain could be
Communication) with ultra-reliability and low latency re- integrated. The result would be hybrid protocols that
quirements ([26, 121, 122, 123]). Mostly, packets are short could further enhance the performance as they should
in an MTC, and thus it is essential that overheads are limit contention to only new nodes. Hence, collisions
made minimal [26]. This requirement can be achieved are much reduced, and backoff periods might be short-
with the surveyed frequency-domain proposals as they do ened with proper designs of these protocols.
minimize overheads including contention time, collisions,
interframe spaces, and control frames. In addition, for • Almost all frequency-domain backoff schemes do not
downlink there is a challenge due to different statuses of properly consider the hidden terminal problem. Many
nodes that can be sleeping or active. It is difficult to en- of such protocols simply assume that RTS/CTS and
tirely predict which nodes are active due to sleeping cy- reattempting transmissions are sufficient to tackle the
cles and what type and size of traffic is available at each problem. We believe that this is not the case specially
node. Frequency-domain polling might provide a quick for dense networks like in an IEEE 802.11ax scenario.
19
ACCEPTED MANUSCRIPT
Moreover, there is a need to carefully investigate the frequency-domain based contention and polling protocols.
frequency-domain contention in multi-hop ad hoc and Next, we compared the surveyed protocols based on many
overlapping networks. features including the evaluation methods, the number of
contention rounds, the underlying PHY technology, and
• Many frequency-domain protocols require modifica-
the types of conveyed signals. We also discussed differ-
tions to the IEEE 802.11 standard PHY layer, and
ent features and issues that are related to the frequency-
they increase the complexity of the wireless devices.
domain protocols like the number of antennas, the number
The additional circuitry could result in faster deple-
of subcarriers, QoS support, TDMA-like transmissions,
tion of energy. For example, nodes are transmit-
synchronization, coexistence, overlapping networks, and
ting most of the time in frequency-domain backoff
the hidden terminal problem. Finally, we proposed differ-
schemes. To burst the channel for every contention
ent directions for future work.
round may impact the energy consumption at each
node. This could be of high significance when con-
T
sidering different types of nodes like in IoT (Inter- References
IP
net of Things) and mobile devices. Furthermore,
[1] Zhang, R., Cai, L., Pan, J.: MAC protocols for high data-rate
new proposals are generally compared to the IEEE wireless networks. Resource Management for Multimedia Ser-
802.11 with basic configurations. However, there have vices in High Data Rate Wireless Networks, pp. 9–42 (2017)
CR
been numerous proposals to optimize the IEEE 802.11 [2] IEEE standard for information technology- telecommunica-
without any changes to the PHY layer. Hence, we tions and information exchange between systems- local and
metropolitan area networks- specific requirements part ii:
believe that it is necessary to evaluate such proposals Wireless LAN medium access control (MAC) and physi-
against the OFDM frequency-domain contention and cal layer (PHY) specifications. IEEE Std 802.11g-2003
polling schemes while considering different trade-offs
(energy, complexity, throughput, delay, etc.).
• For wireless networks, there have been interesting
works to find different estimates specially the num-
US (Amendment to IEEE Std 802.11, 1999 Edn. (Reaff 2003) as
amended by IEEE Stds 802.11a-1999, 802.11b-1999, 802.11b-
1999/Cor 1-2001, and 802.11d-2001) pp. i–67 (2003). DOI
10.1109/IEEESTD.2003.94282
[3] IEEE standard for information technology–local and metropoli-
AN
tan area networks–specific requirements–part 11: Wireless LAN
ber of active nodes [23]. In [133], the authors propose medium access control (MAC) and physical layer (PHY) specifi-
exploiting the frequency-domain to estimate the num- cations - amendment 8: Medium access control (mac) quality of
service enhancements. IEEE Std 802.11e-2005 (Amendment to
ber of neighbor nodes in a vehicular network. Such
IEEE Std 802.11, 1999 Edition (Reaff 2003) pp. 1–212 (2005).
estimates could be utilized to improve the perfor- DOI 10.1109/IEEESTD.2005.97890
M
mance of the network. We believe that the same con- [4] IEEE standard for information technology– local and metropoli-
tention signals in the frequency-domain could also be tan area networks– specific requirements– part 11: Wireless
LAN medium access control (MAC)and physical layer (PHY)
used to estimate the number of nodes in the network. specifications amendment 5: Enhancements for higher through-
Subsequently, contention parameters (like number of
ED
domain operations in a well-known network simulator cess Control (MAC) and Physical Layer (PHY) specifications
(like NS-2, OPNET, and NS-3). Amendment 1: High-speed Physical Layer in the 5 GHz band
[7] IEEE Std 802.11b-1999: Wireless LAN Medium Access Control
(MAC) and Physical Layer (PHY) Specifications: Higher-Speed
7. Conclusion Physical Layer Extension in the 2.4 GHz Band
[8] Al-Mefleh, H., Chang, J.M.: Turning hidden nodes into helper
In this paper, we attempted to provide a comprehensive nodes in IEEE 802.11 wireless LAN networks. In: International
survey of frequency-domain based contention and polling Conference on Research in Networking, pp. 824–835. Springer
(2008)
MAC schemes proposed for wireless networks. The survey [9] Kosek-Szott, K.: A survey of MAC layer solutions to the hidden
included a review of recent proposals related to the next- node problem in ad-hoc networks. Ad Hoc Networks 10(3), 635–
generation IEEE 802.11ax. First, we clarified the mean- 660 (2012)
[10] Malik, A., Qadir, J., Ahmad, B., Yau, K.L.A., Ullah, U.: Qos
ing of the frequency-based polling and contention opera- in IEEE 802.11-based wireless networks: a contemporary re-
tions, and we discussed how they could improve the perfor- view. Journal of Network and Computer Applications 55, 24–46
mance of wireless networks. Then, we reviewed different (2015)
20
ACCEPTED MANUSCRIPT
[11] Wan, Z., Xiong, N., Ghani, N., Vasilakos, A.V., Zhou, L.: Adap- Zhang, Y., Tan, Z.: Fine-grained channel access in wireless
tive unequal protection for wireless video transmission over LAN. IEEE/ACM Transactions on Networking (TON) 21(3),
IEEE 802.11 e networks. Multimedia Tools and Applications 772–787 (2013)
72(1), 541–571 (2014) [31] Zhang, C., Chen, P., Ren, J., Wang, X., Vasilakos, A.V.: A
[12] Viegas, R., Guedes, L.A., Vasques, F., Portugal, P., Moraes, backoff algorithm based on self-adaptive contention window up-
R.: A new MAC scheme specifically suited for real-time indus- date factor for IEEE 802.11 DCF. Wireless networks 23(3),
trial communication based on IEEE 802.11 e. Computers & 749–758 (2017)
Electrical Engineering 39(6), 1684–1704 (2013) [32] Al-Mefleh, H.: SEMP: Self-elimination mac protocol for IEEE
[13] Yu, X., Navaratnam, P., Moessner, K.: Resource reservation 802.11 wireless networks. Wireless Personal Communications
schemes for IEEE 802.11-based wireless networks: A survey. 94(3), 755–776 (2017)
IEEE Communications Surveys & Tutorials 15(3), 1042–1061 [33] Kedžo, I., Ožegović, J., Kristić, A.: BPC–a binary priority
(2013) countdown protocol. Ad Hoc Networks 11(3), 747–764 (2013)
[14] Mammeri, S., Yazid, M., Bouallouche-Medjkoune, L., Mazouz, [34] Fang, M., Malone, D., Duffy, K.R., Leith, D.J.: Decentralised
A.: Performance study and enhancement of multichannel ac- learning MACs for collision-free access in WLANs. Wireless
cess methods in the future generation VHT WLAN. Future networks 19(1), 83–98 (2013)
T
Generation Computer Systems (2017) [35] Sanabria-Russo, L., Barcelo, J., Bellalta, B., Gringoli, F.: A
[15] Charfi, E., Chaari, L., Kamoun, L.: PHY/MAC enhancements high efficiency MAC protocol for WLANs: Providing fairness
IP
and QoS mechanisms for very high throughput WLANs: A sur- in dense scenarios. IEEE/ACM Transactions on Networking
vey. IEEE Communications Surveys & Tutorials 15(4), 1714– (TON) 25(1), 492–505 (2017)
1735 (2013) [36] Jung, D., Lim, H.: Opportunistic MAC protocol for coordi-
[16] Asai, Y.: Advanced progress in IEEE 802.11 WLAN standard- nating simultaneous transmissions in multi-user MIMO based
CR
ization. In: Microwave Conference (APMC), 2014 Asia-Pacific, WLANs. IEEE Communications Letters 15(8), 902–904 (2011)
pp. 911–913. IEEE (2014) [37] Sachdeva, G., Sharvani, G.: Design and implementation of an
[17] Bellalta, B.: IEEE 802.11ax: High-efficiency WLANs. IEEE efficient and adaptive asynchronous MPR MAC protocol for
Wireless Communications 23(1), 38–46 (2016) WLAN. In: Advanced Computing & Communication Tech-
[18] Bellalta, B., Bononi, L., Bruno, R., Kassler, A.: Next generation nologies (ACCT), 2015 Fifth International Conference on, pp.
[21] Cheng, N., Shen, X.S.: Next-generation high-efficiency WLAN. based cooperative multi-channel MAC protocol for the next gen-
In: 5G Mobile Communications, pp. 651–675. Springer (2017) eration WLAN. Wireless Networks (2017). URL https://doi-
[22] Afaqui, M.S., Garcia-Villegas, E., Lopez-Aguilera, E.: IEEE org.sdl.idm.oclc.org/10.1007/s11276-016-1355-3
802.11 ax: Challenges and requirements for future high effi- [41] Xu, F., Zhao, Q., Zeng, Y., Yang, J., Dai, H., Dang, P.: A
ciency WiFi. IEEE Wireless Communications 24(3), 130–137 novel OFDM-based MAC protocol for wireless LANs. In: Com-
ED
jee, “WiFi-Nano: Reclaiming WiFi efficiency through 800 ns IEEE Access 5, 2534–2549 (2017)
slots,” in Proceedings of the 17th annual international confer- [43] Wang, X., Wang, H.: A novel random access mechanism for
ence on Mobile computing and networking. ACM, 2011, pp. OFDMA wireless networks. In: Global Telecommunications
37–48. Conference (GLOBECOM 2010), 2010 IEEE, pp. 1–5. IEEE
[25] C. Chen, H. Zhao, T. Qiu, R. Hou, and A. K. San- (2010)
CE
gaiah, “A multi-station block acknowledgment scheme [44] LEE, G., KIM, C.: Centralized contention based MAC for
in dense IoT networks,” Computer Communications, OFDMA WLAN. IEICE Transactions on Information and Sys-
vol. 119, pp. 179 – 190, 2018. [Online]. Available: tems 100(9), 2219–2223 (2017)
http://www.sciencedirect.com/science/article/pii/S0140366417 [45] Cidon, A., Nagaraj, K., Katti, S., Viswanath, P.: Flashback:
AC
21
ACCEPTED MANUSCRIPT
tomation, Control and Information Engineering 10(1), 72–76 channels. IEEE Transactions on Wireless Communications
(2015) 12(10), 5162–5171 (2013)
[50] Lin, J., Liang, W., Yu, H., Xiao, Y.: Polling in the frequency [69] Zeng, Y., Zhao, Q.: A novel collision analysis for multiple-
domain: a new MAC protocol for industrial wireless network subcarrier frequency-domain contention. In: International Con-
for factory automation. International Journal of Ad Hoc and ference on Wireless Algorithms, Systems, and Applications, pp.
Ubiquitous Computing 20(4), 211–222 (2015) 37–46. Springer (2017)
[51] Li, B., Qu, Q., Yan, Z., Yang, M.: Survey on OFDMA based [70] Lee, K.h., Yoo, J., Kim, C.k.: DiFuse: distributed frequency
MAC protocols for the next generation WLAN. In: Wireless domain user selection for multi-user MIMO networks. Wireless
Communications and Networking Conference Workshops (WC- Networks pp. 1–18 (2016)
NCW), 2015 IEEE, pp. 131–135. IEEE (2015) [71] Zhou, A., Wei, T., Zhang, X., Liu, M., Li, Z.: Signpost: Scalable
[52] Deng, D.J., Chen, K.C., Cheng, R.S.: IEEE 802.11 ax: MU-MIMO signaling with zero CSI feedback. In: Proceedings
Next generation wireless local area networks. In: Heteroge- of the 16th ACM International Symposium on Mobile Ad Hoc
neous networking for quality, reliability, security and robust- Networking and Computing, pp. 327–336. ACM (2015)
ness (QShine), 2014 10th international conference on, pp. 77–82. [72] Lee, S., Kim, C.K.: D-fi: A diversity-aware Wi-Fi using an
IEEE (2014) OFDM-based Bloom filter. In: Network Protocols (ICNP), 2012
T
[53] Mutairi, A., Roy, S.: An OFDM-aware reservation random ac- 20th IEEE International Conference on, pp. 1–10. IEEE (2012)
cess protocol for interference mitigation in OFDMA femtocells. [73] Lee, S., Choi, J., Yoo, J., Kim, C.K.: Frequency diversity-aware
IP
IEEE Transactions on Communications 63(1), 301–310 (2015) Wi-Fi using OFDM-based Bloom filters. IEEE Transactions on
[54] Sen, S., Choudhury, R.R., Nelakuditi, S.: Listen (on the fre- Mobile Computing 14(3), 525–537 (2015)
quency domain) before you talk. In: Proceedings of the 9th [74] Broder, A., Mitzenmacher, M.: Network applications of Bloom
ACM SIGCOMM Workshop on Hot Topics in Networks, p. 16. filters: A survey. Internet mathematics 1(4), 485–509 (2004)
CR
ACM (2010) [75] Lv, S., Zhang, Y., Li, W., Lu, Y., Dong, X., Wang, X., Zhou,
[55] Roman, B., Stajano, F., Wassell, I., Cottingham, D.: Multi- X.: Frequency-code domain contention in multi-antenna multi-
carrier burst contention (MCBC): Scalable medium access con- carrier wireless networks. Journal of Communications and Net-
trol for wireless networks. In: Wireless Communications and works 18(2), 218–226 (2016)
Networking Conference, 2008. WCNC 2008. IEEE, pp. 1667– [76] Lv, S., Dong, X., Shen, H., Wang, X., Zhou, X.: Signature-
[56]
1672. IEEE (2008)
Roman, B.A., Chatzigeorgiou, I., Wassell, I.J., Stajano, F.:
Evaluation of multi-carrier burst contention and IEEE 802.11
with fading during channel sensing. In: Personal, Indoor and
Mobile Radio Communications, 2009 IEEE 20th International
US [77]
based fast access coordination in multicarrier wireless networks.
In: General Assembly and Scientific Symposium (URSI GASS),
2014 XXXIth URSI, pp. 1–4. IEEE (2014)
Kwon, H., Seo, H., Kim, S., Lee, B.G.: Generalized CSMA/CA
protocol for OFDMA systems. In: Global Telecommunications
AN
Symposium on, pp. 57–61. IEEE (2009) Conference, 2008. IEEE GLOBECOM 2008. IEEE, pp. 1–6.
[57] Roman, B., Wassell, I., Chatzigeorgiou, I.: Scalable cross-layer IEEE (2008)
wireless access control using multi-carrier burst contention. [78] Kwon, H., Seo, H., Kim, S., Lee, B.G.: Generalized CSMA/CA
IEEE Journal on Selected Areas in Communications 29(1), 113– for OFDMA systems: protocol design, throughput analysis, and
128 (2011) implementation issues. IEEE Transactions on Wireless Commu-
M
[58] Sen, S., Choudhury, R.R., Nelakuditi, S.: Listen before you nications 8(8) (2009)
talk, but on the frequency domain. ACM SIGMOBILE Mobile [79] Wang, J., Kang, G., Zhang, P.: A two-dimensional medium
Computing and Communications Review 14(4), 7–9 (2011) access control protocol based on OFDMA and CSMA/CA. In:
[59] Sen, S., Roy Choudhury, R., Nelakuditi, S.: No time to count- Wireless Telecommunications Symposium (WTS), 2011, pp. 1–
down: Migrating backoff to the frequency domain. In: Pro- 5. IEEE (2011)
ED
ceedings of the 17th annual international conference on Mobile [80] Wang, Y., Luo, T., Li, J.: A hybrid multi-channel based
computing and networking, pp. 241–252. ACM (2011) medium access control mechanism for WLAN (2013)
[60] Tan, K., Fang, J., Zhang, Y., Chen, S., Shi, L., Zhang, J., [81] Deng, D.J., Lien, S.Y., Lee, J., Chen, K.C.: On quality-of-
Zhang, Y.: Fine-grained channel access in wireless LAN. In: service provisioning in IEEE 802.11 ax WLANs. IEEE Access
ACM SIGCOMM Computer Communication Review, vol. 40, 4, 6086–6104 (2016)
PT
pp. 147–158. ACM (2010) [82] Qu, Q., Li, B., Yang, M., Yan, Z.: An OFDMA based concur-
[61] Sarkar, T.K., Bodas, T.: A FICA based contention scheme for rent multiuser MAC for upcoming IEEE 802.11 ax. In: Wireless
WLAN with non cooperation. In: Communications (NCC), Communications and Networking Conference Workshops (WC-
2013 National Conference on, pp. 1–5. IEEE (2013) NCW), 2015 IEEE, pp. 136–141. IEEE (2015)
[62] Yin, J., Mao, Y., Leng, S., Jiang, Y., Khan, M.A.: Access gran- [83] Haile, G., Lim, J.: C-OFDMA: Improved throughput for next
CE
ularity control of multichannel random access in next-generation generation WLAN systems based on OFDMA and CSMA/CA.
wireless LANs. Computer Networks 91, 135–150 (2015) In: Intelligent Systems Modelling & Simulation (ISMS), 2013
[63] Chai, E., Shin, K.G.: Low overhead control channels in wireless 4th International Conference on, pp. 497–502. IEEE (2013)
networks [84] Herzen, J., Banchs, A., Shneer, V., Thiran, P.: CSMA/CA in
AC
[64] Chai, E., Shin, K.G.: Low-overhead control channels in wireless time and frequency domains. In: Network Protocols (ICNP),
networks. IEEE Transactions on Mobile Computing 14(11), 2015 IEEE 23rd International Conference on, pp. 256–266.
2303–2315 (2015) IEEE (2015)
[65] Huang, P., Yang, X., Xiao, L.: WiFi-BA: Choosing arbitra- [85] Alvi, S.A., Baig, A., Sattar, K.: Frequency-domain backoff
tion over backoff in high speed multicarrier wireless networks. mechanism for OFDM-based wireless LANs. Arabian Journal
In: INFOCOM, 2013 Proceedings IEEE, pp. 1375–1383. IEEE for Science and Engineering 41(12), 4995–5008 (2016)
(2013) [86] Alvi, S.A., Baig, A.: Contention resolution in wireless LANs
[66] Huang, P., Yang, X., Xiao, L.: Adaptive channel bonding in using frequency-domain backoff. In: Wireless and Mobile Com-
multicarrier wireless networks. In: Proceedings of the fourteenth puting, Networking and Communications (WiMob), 2012 IEEE
ACM international symposium on Mobile ad hoc networking 8th International Conference on, pp. 84–91. IEEE (2012)
and computing, pp. 297–300. ACM (2013) [87] Alvi, S.A., Baig, A.: No more hidden backoff: Advertise backoff
[67] Huang, P., Yang, X., Xiao, L.: Dynamic channel bonding: en- in frequency domain. In: Telecommunication Networks and
abling flexible spectrum aggregation. IEEE Transactions on Applications Conference (ATNAC), 2014 Australasian, pp. 217–
Mobile Computing 15(12), 3042–3056 (2016) 222. IEEE (2014)
[68] Wang, G., Wu, J., Zhou, G., Li, G.Y.: Collision-tolerant media [88] Mawlawi, B., DORe, J.B., Lebedev, N., Gorce, J.M.: Multiband
access control for asynchronous users over frequency-selective CSMA/CA with RTS-CTS strategy. In: Wireless and Mobile
22
ACCEPTED MANUSCRIPT
T
2015, pp. 75–82. Springer (2015) LTE and Wi-Fi in unlicensed bands,” IEEE Communications
[93] Lee, S., Choi, J., Yoo, J., Kim, C.K.: Providing quality-of- Magazine, vol. 52, no. 11, pp. 54–61, 2014.
IP
service for frequency-aware Wi-Fi using OFDM-based variable- [112] S. Yun and L. Qiu, “Supporting WiFi and LTE co-existence,”
length Bloom filters. EURASIP Journal on Wireless Commu- in Computer Communications (INFOCOM), 2015 IEEE Con-
nications and Networking 2014(1), 152 (2014) ference on. IEEE, 2015, pp. 810–818.
[94] Zhang, H., Zhao, Q., Ma, Z., Xu, F.: Design and analysis of [113] H. Zhang, X. Chu, W. Guo, and S. Wang, “Coexistence of
CR
weighted frequency-domain contention in wireless LANs. IEEE Wi-Fi and heterogeneous small cell networks sharing unlicensed
Access 5, 1639–1648 (2017) spectrum,” IEEE Communications Magazine, vol. 53, no. 3, pp.
[95] Zhou, H., Li, B., Yan, Z., Yang, M., Qu, Q.: An OFDMA based 158–164, 2015.
multiple access protocol with QoS guarantee for next generation [114] Y. Song, K. W. Sung, and Y. Han, “Coexistence of Wi-Fi and
WLAN. In: Signal Processing, Communications and Comput- cellular with listen-before-talk in unlicensed spectrum,” IEEE
ing (ICSPCC), 2015 IEEE International Conference on, pp. 1–6.
IEEE (2015)
[96] Zhou, H., Li, B., Yan, Z., Yang, M.: A channel bonding based
QoS-aware OFDMA MAC protocol for the next generation
WLAN. Mobile Networks and Applications 22(1), 19–29 (2017)
US Communications Letters, vol. 20, no. 1, pp. 161–164, 2016.
[115] Z. Guan and T. Melodia, “CU-LTE: Spectrally-efficient and
fair coexistence between LTE and Wi-Fi in unlicensed bands,”
in INFOCOM 2016-The 35th Annual IEEE International Con-
ference on Computer Communications, IEEE. IEEE, 2016,
AN
[97] Choi, Y.J., Park, S., Bahk, S.: Multichannel random access in pp. 1–9.
OFDMA wireless networks. IEEE Journal on Selected Areas in [116] B. Chen, J. Chen, Y. Gao, and J. Zhang, “Coexistence of LTE-
Communications 24(3), 603–613 (2006) LAA and Wi-Fi on 5 GHz with corresponding deployment sce-
[98] Chang, Y.J., Chien, F.T., Kuo, C.C.J.: Opportunistic access narios: A survey,” IEEE Communications Surveys & Tutorials,
with random subchannel backoff (OARSB) for OFDMA uplink. vol. 19, no. 1, pp. 7–32, 2017.
M
In: Global Telecommunications Conference, 2007. GLOBE- [117] X. Wang, S. Mao, and M. X. Gong, “A survey of LTE Wi-Fi
COM’07. IEEE, pp. 3240–3244. IEEE (2007) coexistence in unlicensed bands,” GetMobile: Mobile Comput-
[99] Mutairi, A., Roy, S.: Exponential backoff in frequency-domain ing and Communications, vol. 20, no. 3, pp. 17–23, 2017.
for random access in OFDMA femtocells. In: Wireless Com- [118] P. Gawlowicz, A. Zubow, and A. Wolisz, “Enabling cross-
munications and Networking Conference (WCNC), 2013 IEEE, technology communication between LTE unlicensed and WiFi,”
ED
current access control using subcarrier signature in heteroge- [120] S. Zinno, G. Di Stasi, S. Avallone, and G. Ventre, “On a fair
neous MIMO-based WLAN. Multiple Access Communications coexistence of LTE and Wi-Fi in the unlicensed spectrum: A
pp. 109–121 (2012) survey,” Computer Communications, vol. 115, pp. 35–50, 2018.
[102] Zhou, W., Li, D., Srinivasan, K., Sinha, P.: Domino: Rela- [121] S. Chen, R. Ma, H.-H. Chen, H. Zhang, W. Meng, and J. Liu,
tive scheduling in enterprise wireless LANs. In: Proceedings of “Machine-to-machine communications in ultra-dense networks–
CE
the ninth ACM conference on Emerging networking experiments A survey,” IEEE Communications Surveys & Tutorials, vol. 19,
and technologies, pp. 381–392. ACM (2013) no. 3, pp. 1478–1503, 2017.
[103] Zheng, M., Lin, J., Liang, W., Yu, H.: A priority-aware fre- [122] H. Ji, S. Park, J. Yeo, Y. Kim, J. Lee, and B. Shim, “Introduc-
quency domain polling mac protocol for OFDMA-based net- tion to ultra reliable and low latency communications in 5G,”
AC
works in cyber-physical systems. IEEE/CAA Journal of Auto- arXiv preprint arXiv:1704.05565, 2017.
matica Sinica 2(4), 412–421 (2015) [123] C. Bockelmann, N. Pratas, H. Nikopour, K. Au, T. Svens-
[104] H. Lou, X. Wang, J. Fang, M. Ghosh, G. Zhang, and R. Olesen, son, C. Stefanovic, P. Popovski, and A. Dekorsy, “Massive
“Multi-user parallel channel access for high efficiency carrier machine-type communications in 5G: Physical and MAC-layer
grade wireless LANs,” in Communications (ICC), 2014 IEEE solutions,” IEEE Communications Magazine, vol. 54, no. 9, pp.
International Conference on. IEEE, 2014, pp. 3868–3870. 59–65, 2016.
[105] Karaca, M., Bastani, S., Priyanto, B.E., Safavi, M., Landfeldt, [124] S. K. Sharma, T. E. Bogale, L. B. Le, S. Chatzinotas, X. Wang,
B.: Resource management for OFDMA based next generation and B. Ottersten, “Dynamic spectrum sharing in 5G wireless
802.11 WLANs. In: Wireless and Mobile Networking Confer- networks with full-duplex technology: Recent advances and re-
ence (WMNC), 2016 9th IFIP, pp. 57–64. IEEE (2016) search challenges,” IEEE Communications Surveys Tutorials,
[106] B. Bellalta and K. Kosek-Szott, “Ap-initiated multi-user vol. 20, no. 1, pp. 674–707, Firstquarter 2018.
transmissions in IEEE 802.11 ax WLANs,” arXiv preprint [125] M. Luvisotto, A. Sadeghi, F. Lahouti, S. Vitturi, and M. Zorzi,
arXiv:1702.05397, 2017. “RCFD: A novel channel access scheme for full-duplex wireless
[107] O. Sharon and Y. Alpert, “Scheduling strategies and through- networks based on contention in time and frequency domains,”
put optimization for the downlink for IEEE 802.11 ax and IEEE IEEE Transactions on Mobile Computing, pp. 1–1, 2018.
802.11 ac based networks,” arXiv preprint arXiv:1709.04818, [126] Luvisotto, M., Sadeghi, A., Lahouti, F., Vitturi, S., Zorzi, M.:
23
ACCEPTED MANUSCRIPT
T
CRAHNs. In: Global Communications Conference (GLOBE-
COM), 2012 IEEE, pp. 422–426. IEEE (2012)
IP
[131] Wang, L., Wu, K., Xiao, J., Hamdi, M.: Harnessing frequency
domain for cooperative sensing and multi-channel contention
in CRAHNs. IEEE Transactions on Wireless Communications
13(1), 440–449 (2014)
CR
[132] Zheng, M., Liang, W., Yu, H., Song, M.: SMCSS: A quick
and reliable cooperative spectrum sensing scheme for cognitive
industrial wireless networks. IEEE Access 4, 9308–9319 (2016)
[133] Kong, L., Chen, X., Liu, X., Rao, L.: FINE: Frequency-
divided instantaneous neighbors estimation system in vehicular
networks. In: Pervasive Computing and Communications (Per-
Com), 2015 IEEE International Conference on, pp. 172–177.
IEEE (2015) US
AN
M
ED
PT
CE
AC
24