Content uploaded by Daniele Tarchi
Author content
All content in this area was uploaded by Daniele Tarchi on Jan 02, 2019
Content may be subject to copyright.
TRANSACTIONS ON EMERGING TELECOMMUNICATIONS TECHNOLOGIES
Trans. Emerging Tel. Tech. 2014; 00:1–14
DOI: 10.1002/ett
RESEARCH ARTICLE
A M2M Cognitive MAC protocol for Overlaid OFDMA
Environments
D. Tarchi1∗, R. Fantacci2and D. Marabissi2
1Department of Electrical, Electronic and Information Engineering, University of Bologna, 40136 Bologna, Italy
2Department of Information Engineering, University of Florence, 50139 Firenze, Italy
ABSTRACT
Machine to Machine (M2M) communications have gained in the last years an increasing interest due to ever growing
number of machine type devices that are used in different application fields by allowing low cost and efficient
communications among devices mainly in an autonomous manner. Even if M2M protocols need of dedicated resources,
a new paradigm, called Cognitive M2M (CM2M) has been recently introduced in order to exploit cognitive/opportunistic
radio communications. The aim of this paper is to outline a CM2M mechanism where the primary network is based on
the OFDMA technique, while the M2M communication based secondary network uses a novel Medium Access Control
(MAC) technique, named Data Aided Cognitive Technique (DACT). The performance of the proposed DACT protocol
is derived by means of suitable analytical methods under different operational conditions. Analytical predictions are also
validated by comparisons with numerical results obtained through computer simulations, in order to show the effectiveness
of the proposed solution in terms of throughput, delay, resource wastage and CM2M devices queue length; to this aim the
proposed DACT protocol has been implemented with different alternatives. Among them, an adaptive approach allows to
optimize the system performance by increasing the overall throughput while keeping under control the system delay and
the resource wastage. Copyright c
2014 John Wiley & Sons, Ltd.
∗Correspondence
Department of Electrical, Electronic and Information Engineering, University of Bologna, 40136 Bologna, Italy.
1. INTRODUCTION
Nowadays the wireless networks frequency assignment is
regulated by fixed policies performed by governmental
agencies. Although the fixed spectrum assignment allows
an easier deployment for large scale communication
systems, the limited availability and the inefficient usage
require a new communication paradigm to allow a dynamic
access to the spectrum. For this reason, one of the key
enabling technologies for next generation networks is the
cognitive radio [1]. It provides the capability to use or share
the spectrum in an opportunistic way.
The cognitive radio technology is based on two main
characteristics: cognitive capability and reconfigurability.
Cognitive capability refers to the ability of sensing the
radio environment in order to identify those portions of
the spectrum that are unused at a specific time or location.
Reconfigurability enables the radio to be dynamically
adapted to the radio environment. More specifically, the
cognitive radio can be designed to transmit and receive on
a variety of frequencies and to use different transmission
access technologies supported by its hardware design.
A cognitive network environment is composed by at
least two network infrastructures: one primary network
with a fixed allocation in terms of spectrum resources,
and a secondary network, that is the effective cognitive
network, aiming to exploit the portion of the spectrum left
unused by the primary network. In a very general case,
we can refer to environments where multiple independent
secondary networks co-exist; however, in the following,
we will refer to the most common case with one secondary
network.
The scenario considered in this paper refers to a
cognitive environment where the primary network is
based on the Orthogonal Frequency Division Multiple
Access (OFDMA) principle. The secondary network,
instead, is based on a Machine to Machine (M2M)
protocol [2], aiming to exploit the primary network
communication protocol for implementing an independent
and autonomous cognitive system. In particular, due to
the OFDMA technique, and to the resulting subcarrier
allocation method, it is possible that some frequency
holes occur within the frame structure. Aim of the
secondary network is to discover and exploit these holes
for implementing an independent network.
Copyright c
2014 John Wiley & Sons, Ltd. 1
Prepared using ettauth.cls [Version: 2012/06/19 v2.10]
A M2M Cognitive MAC protocol for Overlaid OFDMA Environments D. Tarchi, R. Fantacci and D. Marabissi
M2M is a network paradigm that is receiving an
increasing interest in contexts where several nodes
(typically sensors, metering devices, actuators) need to
be connected among them. As a consequence, M2M
communications are characterized by a low data rate
information exchange, short messages, and low priority.
In order to satisfy these characteristics the concept of
capillary networks is gaining attention [3]. In this case
the M2M communications are supported by a hybrid
infrastructure where machines do not access directly to
the cellular network, but groups of machines communicate
using suitable short range technologies and are connected
to the cellular networks via gateways that perform data
aggregation.
The scenario considered in this paper is similar to that of
a capillary network [3], indeed, the idea is to create a M2M
network independent from the primary cellular network,
that allows the direct communications (i.e., Devices-to-
Devices, D2D) among machines that are close each other.
This network could be part of a larger M2M network
using gateways to aggregate M2M traffic and to access the
cellular network thus achieving a wider coverage [4].
Similarly to the capillary networks, this system
has the advantage that avoids competitions between
M2M and H2H (Human-to-Human) communications for
the resource usage at the cellular base station (BS),
which represents a bottleneck when a high number of
machines wants to access the channel [5]. However,
in general, capillary networks use access technologies
different from the cellular networks, and operate in
separate frequency bandwidths. Differently, the system
proposed here considers a cognitive radio environment
where the frequency holes of the primary network
are opportunistically exploited to create an independent
polling based network for M2M communications. Our
study is motivated by the awareness that the overall amount
of free resources in an OFDMA frame varies during
the day without reaching the maximum capacity of the
primary network; this has been highlighted in [6,7], where
the occupancy statistics of an IEEE 802.16 frame are
considered as an example. The same occupancy statistics
can be considered valid also for LTE and other OFDMA
based networks.
Based on in-band signaling exploitation, in this paper
we propose a novel M2M MAC technique, the Data
Aided Cognitive Technique (DACT), suitable to support
M2M communications, for allowing the multiple access to
secondary users while avoiding interferences to the already
planned primary network; the secondary CM2M devices
are supposed to have a legacy interface so that the in-band
signaling of the overlaid network can be exploited in a
suitable way. The effectiveness of the proposed technique
has been validated through both theoretical analysis and
numerical results obtained by computer simulations. In
order to assess the effectiveness of the proposed technique,
some different approaches have been considered. Among
them, the adaptive approach allows to optimize the system
performance by maximizing the system throughput while
keeping under control the system delay and the resource
wastage.
The remainder of this paper is organized as follows.
In Section 2, the most relevant works with respect to the
topics of this paper are discussed. In Section 3, the system
scenario is presented, by highlighting the specific behavior
of the OFDMA primary network. In Section 4the polling
based MAC technique is introduced, by focusing also on
the theoretical analysis. In Section 5, numerical results
obtained through computer simulations are introduced,
and, finally, in Section 6, conclusions are drawn.
2. RELEVANT WORKS
The exploitation of cognitive techniques for setting
up M2M communications is a promising trend [8].
The interested reader could refer to [9], a special
issue on cognitive radio technologies applied to M2M
environments.
Moreover, the idea of using OFDMA within a
cognitive environment has been already considered in the
past. Indeed, one of the most widely known cognitive
standards, the IEEE 802.22 [10], exploits the OFDMA
as the access scheme used by the cognitive system.
In [11] the authors proposed a cross-layer spectrum
sensing and scheduling technique able to operate in
secondary OFDMA systems. Also in [12] the problem
of detecting primary transmissions by considering a
secondary OFDMA system was studied.
Differently from these approaches, we focus here on the
case of an OFDMA scheme, used as the access scheme
for the primary network. This scenario has been previously
considered in [13], where the authors focused on a sensing
technique for OFDMA primary systems, and in [14],
where a cooperative scheme for downlink transmissions
in OFDMA environments is proposed. In particular, we
investigate here the performance of a MAC protocol that
allows the coexistence of a secondary M2M network with
a primary OFDMA based network. This problem was
previously considered in [15,16] where, differently from
our approach, a random access scheme for the secondary
MAC is proposed. Another interesting paper is [17], where
the authors introduced a MAC protocol for cognitive radio
synchronized environments. A general overview of M2M
MAC techniques is given in [18,19], where the authors
introduce the main challenges and properties of different
techniques.
In order to discover the unused spectrum parts of the
primary BS, several cognitive techniques based on the
spectrum sensing approach have been proposed in the
literature: an overview is provided in [20]. Among them, a
well known method is that based on the Energy Detection
(ED) approach. Alternative approaches are possible when
the signal to reveal satisfies the cyclostationary property.
2Trans. Emerging Tel. Tech. 2014; 00:1–14 c
2014 John Wiley & Sons, Ltd.
DOI: 10.1002/ett
Prepared using ettauth.cls
D. Tarchi, R. Fantacci and D. Marabissi A M2M Cognitive MAC protocol for Overlaid OFDMA Environments
Differently, here the idea at the base of the proposed
scheme is the exploitation of the in-band control signal
broadcast by each BS of the primary OFDMA system
for allowing the communications among the nodes of
the secondary M2M network. In the two most common
OFDMA based standards, i.e., IEEE 802.16 [21]and LTE
[22], two specific fields within the downlink subframe
are defined for notifying in broadcast to the users on the
spectrum usage. The in-band signaling exploitation has
been already introduced by the authors in [23], where
the analysis has been focused on the early stage of the
idea. A similar approach has been also considered in [24],
where differently from this approach the authors focused
on a femtocell scenario by using interference alignment
techniques.
A similar approach, named Cognitive Pilot Channel
(CPC) [25], have been also taken into account by the
European Telecommunication Standardization Institute
(ETSI) as a viable option for aiding the coexistence on the
same spectrum resources of heterogeneous systems. The
exploitation of the CPC has been taken into account also
in [26], where the CPC is supposed to be piggybacked on a
Terrestrial Digital Media Broadcasting (T-DMB) framing
structure.
3. SYSTEM SCENARIO
This paper considers the cognitive paradigm where two
different communication networks coexist without mutual
interference, avoiding interference among them and with
low performance degradations. In particular, the primary
is the network that pre-exists with respect to the cognitive
application. Hence, we assume that it is a wide area
network with lower adaptation capabilities. On the other
hand the secondary network operates in a cognitive way.
Its main aim is to exploit the holes in the communication
bandwidth of the primary network in order to set up
an independent communication network avoiding mutual
interference. Toward this end we focus on an approach that
exploits the in-band signaling of the primary network.
The considered primary network is an OFDMA-based
system where the bandwidth is divided into several
orthogonal subcarriers; moreover, the time axis is divided
in symbols, thus, each subcarrier and each time symbol
can be seen as an independent unit. Each combination
of subcarriers and time symbols forms an independent
resource that can be assigned to any communication. The
transmitted signal during any OFDMA symbol is:
s(t) = Re
ej2πfct
(Nused−1)/2
X
k=−(Nused−1)/2
ckej2πk∆f(t−Tg)
,
0< t < Ts,(1)
where fcis the carrier frequency, Nused is the number of
used subcarriers, ckis a complex value, corresponding to
the data to be transmitted on the k-th subcarrier, Tgis the
guard time, Tsis the OFDMA symbol duration, including
guard time, and ∆fis the subcarrier frequency spacing.
Let us focus the attention on the Nused value; it corresponds
to the total number of subcarriers that can be used in the
system for sending pilots or data subcarriers. The Nused
value can vary depending on the subcarrier allocation and
overall bandwidth, but its value does not influence the
system under consideration.
However, differently from a TDMA-OFDM transmis-
sion system where each communication occupies all the
subcarriers for a certain time symbol, in the OFDMA the
resource occupancy is scattered within the communication
bandwidth. This could result in the presence of spectrum
holes due to the fact that not all the available resources
are occupied. For example, during low traffic periods
(e.g., during the night), the OFDMA network could be
underused for the actual needs resulting in a lower resource
occupancy [6,7]. Aim of this paper is to design a proper
MAC technique able to exploit these holes for setting up
an independent (secondary) M2M network.
Let us consider the case where, for a certain frequency
offset ˆ
k, the value cˆ
kis equal to 0 because for that
subcarrier no pilot and no data have to be sent. In this case,
if we define as Ktthe set of subcarriers in which data or
pilot have to be sent, i.e.:
ck∈Kt⇔ck6= 0 (2)
we can modify (1) as:
su(t) = Re
ej2πfct
(Nused−1)/2
X
k=−(Nused−1)/2
ck∈Kt
ckej2πk∆f(t−Tg)
,
0< t < Ts.
(3)
Eq. (3) defines the signal sent by the primary system.
Aim of the cognitive secondary network is to exploit all
the ˆ
kfrequencies for a certain symbol time for sending
data. Differently from common sensing approaches, our
aim is to exploit the in-band signaling within the framing
structure for getting knowledge of free resources.
Indeed, based on the data to be exchanged with the
User Equipments (UEs) in the coverage area, the BS needs
to schedule the data traffic. It is out of the scope of the
paper to know the policies used for the scheduling [27–29],
but it is important to focus the attention on the fact that
the scheduling decision has to be broadcast by the BS.
In particular, OFDMA based system, e.g., LTE/LTE-A,
IEEE 802.16x, are usually exploiting a frame oriented
resources management. Each frame is divided into two
subframes: the downlink and the uplink subframes. In the
downlink subframe, two specific fields are sent on each
frame, defining the resource allocation on the following
downlink and uplink subframes. The in-band signaling can
be seen as a field carrying all the information for having
knowledge about the occupancy of the following frame.
Trans. Emerging Tel. Tech. 2014; 00:1–14 c
2014 John Wiley & Sons, Ltd. 3
DOI: 10.1002/ett
Prepared using ettauth.cls
A M2M Cognitive MAC protocol for Overlaid OFDMA Environments D. Tarchi, R. Fantacci and D. Marabissi
The in-band signaling is present in the majority of the
recently introduced broadband wireless access standards,
e.g., IEEE 802.16 family [30] and LTE family [31]. Even if
their name is different in the two standards its task is quite
the same. The in-band signaling is referred as DL-MAP
and UL-MAP in the IEEE 802.16 family, and as PDCCH in
the LTE family. While the DL-MAP and UL-MAP carries
separately the information about resource management in
the UL and DL subframes, the PDCCH is used as the only
physical field for sending control information about the
resource assignment in both DL and UL.
The IEEE 802.16 MAPs contain information on DL/UL
burst allocation and physical layer control message. The
PDCCH carries scheduling assignments and other control
information. It is clear that both act in a similar way.
Moreover, both are sent in the first OFDMA symbols
(the number of symbols depends on the considered
implementation), and can be seen as a preamble.
Each UE is able to understand which subcarriers are
used and, among these, to understand if and where data
are sent to it. This implies that some subcarriers within the
frame are reserved by the BS to be used by the UEs (either
in uplink or in downlink), while some other subcarriers
could remain empty.
The secondary system is an OFDMA-based M2M
network that operates in an independent manner thus
avoiding congestion in the access control channel and
signaling overhead on the primary network. It is composed
by multiple scattered devices (e.g., metering stations or
sensor nodes) that communicate in an autonomous way
among them. Transmitted data are characterized by low
data rate, delay tolerance and are intermittent. The M2M
network uses the same physical interface of the primary
thus it is able to exploit the in-band control signals to
discover the empty subcarriers of the primary network. In
particular, each device, after connecting to the network,
will be able to read the in-band control signals sent by
the primary BS, acquiring the knowledge of the primary
subcarriers occupancy on a frame-by-frame basis.
In Fig. 1, the working scenario is represented, where the
coverage areas for the primary OFDMA and the secondary
cognitive M2M networks are overlapped.
The proposed method allows to set up an autonomous
network without the intervention of the BS, by exploiting
the in-band signaling; moreover, due to the absence of
sensing and the direct exploitation of unused holes, the
transmission delays are reduced with respect to other
approaches and there is not waste of resources used for
sensing.
Another important advantage introduced by the pro-
posed mechanism is that the interference towards those
primary users appearing during the secondary users com-
munications is avoided; this is because the in-band sig-
naling works in such a way that the primary users can
send/receive only if allowed by the BS. On the other
side the secondary users would exploit only those unused
resources not assigned to any primary user. Hence, in
case a new primary user appears, its transmission would
be reported in the in-band signaling fields, so that no
secondary users could send data in such resources.
4. THE DATA AIDED COGNITIVE M2M
MAC PROTOCOL
In this section we will introduce our proposal of a novel
cognitive MAC technique, named Data Aided Cognitive
Technique (DACT), that allows the communication of a
secondary M2M network underlaid with respect to the
primary OFDMA network.
If on one side the cognitive sensing of the unused
resources is solved by exploiting the in-band signaling,
the aim of a cognitive MAC protocol is to allow the
communications among the devices in a reliable way.
In the literature several different options for setting-
up a M2M communication have been proposed; among
other we rely on a M2M reference architecture that
allows a direct communication among the devices, as
the one described in [19]. The aim of a cognitive M2M
communication system is to allow the communications
among the M2M nodes by exploiting the unused spectrum
resources of a primary network.
To this aim, by exploiting the in-band signaling
broadcast by the primary network the M2M nodes can be
aware of the primary system occupancy and thus use the
unused resources. Hence, our proposal is toward a proper
MAC protocol, that, taking into account the information
captured from the primary network, allows to set-up an
independent CM2M network.
In particular, we focus here on a polling technique.
The polling approach is particularly suitable for those
applications where the presence of a known number of
M2M devices is considered, even if this number could
be different in different scenarios. The polling technique
allows to manage in a simple way even a high number of
devices, as it will be shown in the Section 5, by adapting
to the variable traffic generated by each M2M device. The
target application of the proposed approach is a distributed
system where a set of independent devices jointly perform
information elaboration and data sensing that appear to its
users as a simple coherent system.
The scenario is defined by the presence of Ndevices
able to detect the OFDMA signal, thus being able to
know which resources within the OFDMA frame can be
exploited, and that all the devices are connected directly
among them.
The secondary devices are able to know the primary
system occupancy, while they need to coordinate among
them for starting the communication. This problem is
solved by considering a properly designed protocol that
needs to take into account two main challenges: the
network setup and the Medium Access Control (MAC).
While in the network setup we have to consider the access
of new devices to the secondary network, also known as
4Trans. Emerging Tel. Tech. 2014; 00:1–14 c
2014 John Wiley & Sons, Ltd.
DOI: 10.1002/ett
Prepared using ettauth.cls
D. Tarchi, R. Fantacci and D. Marabissi A M2M Cognitive MAC protocol for Overlaid OFDMA Environments
4G eNodeB
4G UE
4G UE
4G UE
M2M device
M2M device
M2M device
4G UE
Figure 1. The schematic representation of the Cognitive M2M scenario.
Admission Control, the MAC needs to solve the medium
access of the devices.
The network setup has been already considered in [23],
while we focus herein on the MAC protocol. As for the
device list we can suppose that the M2M network has
been managed previously in an offline phase, so that each
device is capable to be aware of the other M2M devices
and the polling order; such scenario can be effectively used
in applications where a fixed and predetermined number
of M2M devices is set up (e.g., smart meters, in-home
applications).
The polling approach is based on the token passing
technique; due to the framing structure of the primary
network, the amount of time to be assigned to each device
is referred to the primary network frame time. However,
the amount of resources that can be used by each device
can vary based on the occupancy of the primary network
in each frame.
The proposed M2M MAC protocol based on the polling
technique is illustrated in Algorithm 1, that refers to the
behavior of each M2M device. A certain M2M device
remains in a waiting state until it receives the token.
Whenever the token is received, the M2M device holds
it for a certain amount of time. This time can be fixed or
made dependent on the data in the transmitting queue of
the M2M device.
Algorithm 1 The M2M MAC protocol
Wait for the token
while Token is received do
Wait for the Primary Network in-band signaling
Decode the in-band signaling
if The Primary Network frame has free resources
then
Send data
end if
end while
Release the token
The behavior of the proposed DACT is represented in
Fig. 2where a simplified scenario with three M2M devices
and one 4G eNodeB is considered. In order to simplify
the representation, the propagation delays have not been
considered. It is possible to note that the 4G eNode B
Figure 2. The behaviour of the DACT in a simplified scenario
sends regularly the inband signaling in broadcast to all the
M2M devices that, in turn, exchanges the token. The M2M
device holding the token is also allowed to send data to
the other devices. The data transmission depends on the
availability of the resources of the primary as signaled by
the 4G eNodeB, and its duration depends on the technique
we use.
Indeed, to this aim, different techniques that take into
account these different approaches have been considered
in the following. During the time a M2M device holds the
token, it waits for the primary network in-band signaling
and decodes it when received. If the primary network
frame has free resources, the M2M device sends data in
the unused portions; the data are sent in a broadcast way
so that all the M2M devices can receive them. The amount
of data differs, and depends on the selected approach, as
discussed in the following. At the end, the M2M device
Trans. Emerging Tel. Tech. 2014; 00:1–14 c
2014 John Wiley & Sons, Ltd. 5
DOI: 10.1002/ett
Prepared using ettauth.cls
A M2M Cognitive MAC protocol for Overlaid OFDMA Environments D. Tarchi, R. Fantacci and D. Marabissi
releases the token that will be received by the following
M2M device. Even if the access to the network is exclusive
to one device at a time, depending on the token passing, the
data are broadcast so that each device can receive them.
The MAC protocol we propose is based on the
polling scheme, where a token is passed among all the
devices. Differently from the classical polling approach we
consider the presence of the primary network that does
not allow to use the communication spectrum during all
the time. In that sense the MAC protocol needs to take
into account that, after the reception of the token, a certain
device could not be able to send data due to the presence
of the primary network.
In the following we consider the overall token delay
¯
Das the time elapsing between two consecutive arrivals
of the token at the same device, while ¯
THis the amount
of time that each station holds the token. It has to be
emphasized that due to the occupancy of the primary
network a device could not send data during the whole ¯
TH
interval.
Within this scenario five different approaches can be
considered, where the holding time assumes different
values:
a) An exhaustive approach, where each device that
receives the token sends data until its queue is
empty;
b) A gated approach, where each device sends all the
data that is in the queue at the moment the token is
received;
c) A limited approach, where each device sends a
limited amount of data;
d) A fixed approach, where each device holds the
token only for a limited amount of time;
e) An adaptive approach where each device sends a
certain amount of data that depends on a defined
cost function.
In the following we will consider that there are N
devices and that the primary network has a percentage p
of free resources within each frame. Indeed, it has been
shown that, even if depending on the type of data carried
on primary OFDMA networks, there is a certain amount of
data that remains free within the framing structure [6,7].
In Tab. I, the parameters used for the protocol analysis are
summarized.
Exhaustive approach In this case we consider that each
device sends data until it has an empty transmitting queue.
It has to be noted that this approach could lead to a
starvation in case the data generated by each device is
higher with respect to the free resources in the primary
network. This could result in an unfair behavior in the case
the amount of free resources of the primary network are
less than the data generated by each device.
Table I. List of the used parameters.
Parameter Description
NNumber of M2M devices
TfFrame length of the primary network [s]
RResources of the primary network in
each frame [b]
pPercentage of free resources of the
primary network at each frame
SThroughput [b/s]
¯
DOverall token delay [s]
¯
THToken holding time [s]
βiTransmission probability of i-th M2M
device
LpSecondary network packet length
(including both payload and header) [b]
P L Payload of each packet of the secondary
network [b]
jToken length [b]
kiPackets sent by i-th M2M device during
the token holding time
λPacket generation rate of each M2M
device
By resorting for convenience to the same notation as
in [32] it is possible to derive the throughput SEas
SE=P L Pki
N(j/pR)Tf+ (LpPki/pR)Tf
(4)
where kiis the amount of packets sent by the i-th device,
Ris the amount of resources within an OFDMA frame, Tf
is the OFDMA frame time, P L is the payload in bits in
one packet of the secondary network, jis the size in bits
of the token exchanged by the secondary network MAC
protocol, Lpis the M2M packet length (including both
header and payload) and pis the primary network free
resource probability. The packets sent by each device are:
ki=λ·¯
DE,
where λis the packet generation rate at each M2M device
and the overall token delay is:
¯
DE=Nj
pR Tf+LpPki
pR Tf.(5)
The mean holding time is:
¯
THE=j
pR Tf+LpPki
NpR Tf.(6)
The probability βB,i that the i-th CS can transmit data at
each token reception is:
βE,i =kiLp/pR
N(j/pR) + (LpPki/pR),(7)
where Lpis the secondary network packet length including
both the payload and the header.
6Trans. Emerging Tel. Tech. 2014; 00:1–14 c
2014 John Wiley & Sons, Ltd.
DOI: 10.1002/ett
Prepared using ettauth.cls
D. Tarchi, R. Fantacci and D. Marabissi A M2M Cognitive MAC protocol for Overlaid OFDMA Environments
Gated Approach In this scheme each device sends all
the data that is in the queue at the time instant the token is
received. It differs from the previous exhaustive approach
by the fact that the data incoming in the output queue when
the token is hold is not sent immediately, while needs to
wait the following token. The behavior is similar to the
exhaustive approach, and is affected also by the starvation
problem.
The throughput SG, overall token delay ¯
DG, the mean
holding time ¯
THG, and the transmission probability βG,i
have the same behavior of the exhaustive approach. It
should be noted, however, that the exhaustive and the gated
approaches differ on how the amount kiis evaluated: in the
exhaustive case it takes into account all the packets in the
queue when receiving the token plus those arriving when
the token is still at that device, while for the gated case the
value kiis limited to the number of packets in the queue
when the token arrives to the i-th device.
Limited Approach In this scheme the aim is to reserve a
certain amount of capacity to each device. Differently from
the previous approaches the system can go in a starvation
state if the available resources in the primary network are
few, preventing the transmission of the data.
The throughput SL, overall token delay ¯
DL, the mean
holding time ¯
THL, and the transmission probability βL,i
have the same behavior of the exhaustive and limited
approaches. However, in the limited approach the amount
kiis held constant to a value K, limiting the amount of
data that each device can send.
Fixed Approach In this case we consider that each
device can hold the token for a fixed amount of time
without any consideration to the loading factor of the
primary network. Regarding to this, in this approach it
is not guaranteed that a device can send data within its
transmission period, while preserving a regular rotation of
the token among the different devices.
It is possible to derive the throughput SFas
SF=NpRH ·P L
N(j/pR)Tf+N H Tf
,(8)
where Hstands for the number of frame that the token is
held by each M2M device. Concerning the overall token
delay, it can be derived as:
¯
DF=Nj
pR Tf+NHTf,(9)
and the mean holding time is:
¯
THF=j
pR Tf+HTf.(10)
It should be noted that in this case the performance
figures do not depend on the amount of data. Finally, the
probability βFthat a device can transmit data at each token
reception is:
βF=H
N(j/pR) + N H .(11)
Adaptive Approach In this scheme the holding time
is adaptive, by exploiting a variable parameter αthat
multiplies the Hvalue, in order to have a variable interval
in which each station can send packets, aiming to avoid
the starvation that could occur in the previous cases. It is
possible to derive the throughput SAas:
SA=min(NpRαH ·P L, P L Pki)
N(j/pR)Tf+ min (NαH , LpPki/pR)Tf
,
(12)
while the overall token delay is:
¯
DA=Nj
pR Tf+ min N αH Tf,LpPki
pR ,(13)
and the mean holding time is:
¯
THA=j
pR Tf+ min αH Tf,LpPki
NpR .(14)
The probability βCthat a device can transmit data at each
token reception is:
βA=min(αH, kiLp/pR)
N(j/pR) + min (N αH, (LpPki/pR)) .(15)
The optimal value for the parameter αcan be obtained
through the minimization of a multi-objective cost function
aiming at maximize the throughput SAand minimize the
overall token delay ¯
DA, corresponding to
C=
maxαmin(NpRαH ·P L,P L Pki)
N(j/pR)Tf+min(N αH,LpPki/pR)Tf
minαnNj
pR Tf+ min N αH Tf,LpPki
pR o
(16)
whose optimization is a non-trivial problem. To this
aim we propose an heuristic solution that exploiting
the knowledge of the queue length at each device and
the overall token delay aims at minimize the previously
defined cost function. Indeed, our proposal is to vary the
value of αat each transmission interval for each station
based on a cost function defined as:
Cheu =qi−¯q
¯q−Di−¯
DA
¯
DA
,(17)
where qiis the queue length of the i-th device, ¯qis the
average length of the queue of all the users, Diis the
time interval between two consecutive token receptions
last by the i-th device. It should be noted that ¯qand
¯
DAcan be easily obtained during the token passing as a
supplementary field within the token. The value of αis
increased by 1 if the cost is higher than zero and decreased
by 1 if it is lower than zero.
4.1. Energy consumption
The energy efficiency of the proposed MAC protocols is
not one of the main focus of the paper, however, it is an
important issue when M2M are considered [33], hence we
Trans. Emerging Tel. Tech. 2014; 00:1–14 c
2014 John Wiley & Sons, Ltd. 7
DOI: 10.1002/ett
Prepared using ettauth.cls
A M2M Cognitive MAC protocol for Overlaid OFDMA Environments D. Tarchi, R. Fantacci and D. Marabissi
introduce here a rough evaluation of energy consumption
with the aim of comparing our proposed approaches.
By resorting to the model presented in [34], it is possible
to derive the energy consumption of a certain device as:
E=Eid +Es+Etx +Erx (18)
where Eid is the energy consumed in sleeping mode,
Esis the energy consumed for sensing∗,Etx is the
energy consumed for sending data, and Erx is the energy
consumed for receiving data.
In order to derive the energy efficiency of the proposed
approaches it is possible to resort to an energy cost measure
defined as the ratio between the consumed energy and
the total amount of bits, L, sent and received in a certain
amount of time T, i.e.,
Ce=Etx +Erx +Es+Eid
L(19)
In order to simplify our analysis we assume that the
energy consumed in sleeping mode is negligible with
respect to the other quantities; moreover, regarding the
energy needed for transmitting data, usually composed by
one term that takes into account the energy consumption of
the transmitter electronics and digital processing and one
term for the radiated power necessary to transmit the bits
over a certain distance, we can neglect the latter term due
to the short distances we take into account in this paper.
This means that, by considering with Etx/rx equal to
Etx +Erx, (19) can be rewritten as:
Ce∼
=Etx/rx +Es
L
=Ee·L+Es
L
(20)
where Eestands for the energy consumption per bit of
the transmitter electronics and digital processing, that can
be considered equal for the transmission and reception
phases [34].
If we introduce the overall network throughput S
defined as S=L/T , it is possible to write the cost Ce
as:
Ce∼
=Ee·S·T+Es
S·T
=Ee+Es
S·T
(21)
Hence, the consumed energy per bit is composed by two
terms: the first is the same for all the proposed approaches,
while the second decreases for higher throughput values,
given that the energy consumed for sensing the in-
band signaling can be considered the same for all the
approaches. This means that those approaches having a
higher transmission efficiency (i.e., throughput) and a
∗For the proposed DACT, it corresponds to the energy needed for receiving the
in-band signaling.
lower resource wastage, can reach also a lower energy
cost. The obtained formula gives a method for comparing
the previously defined approaches in terms of energy
consumption.
4.2. Theoretical Analysis
The effectiveness of the proposed approach has been
validated by resorting to a theoretical model that allows
to describe the polling technique. To this aim the analysis
considers the results obtained in [35], where the average
delay in the exhaustive approach has been derived as:
ˆ
¯
DE= ¯x+Λ¯x2
2(1 −ρ)+¯
W2
2¯
W+(N−1) ¯
W
2(1 −ρ),(22)
where ¯xis the average service time for sending a single
message, ρis the loading factor, ¯
Wis the average token
passing time, Nis the number of devices, and Λis the
overall generated traffic by considering all the devices. It
is possible to derive the delay also for the gated approach
as:
ˆ
¯
DG=ˆ
¯
DE+ρ¯
W
1−ρ(23)
and for the limited approach as:
ˆ
¯
DL=1−ρ
1−ρ−Λ¯
Wˆ
¯
DE(24)
Differently from the classical approach introduced in [35],
we are now facing with a system where, even if a device
has the token for sending data, it may not access to the
channel due to the occupancy of the primary system. To
this aim we introduce the probability pfr that a certain
frame has a sufficient amount of resources for allowing
to the M2M device to send one message; this probability
can be derived by the primary network free resource
probability pas:
pfr =Prob{Lp> pR}.
In the following we consider a simplified approach
where the message is not divisible between different
frames, so that the probability of free resources
corresponds to the probability that the available resources
are higher than the message length.
It is possible to derive the probability of sending a
message. Due to the fact that the message can be sent only
if the free resources are higher than the message length,
a message will be sent with a probability pfr . Moreover,
due to the fact that the message transmission needs to be
completed within the OFDMA frame, but not necessarily
at the beginning, the probability of sending a message at
the k-th trial can be derived as a geometrical distribution
with a constant offset:
Px=kTf−Tf
2= (1 −pf r )k−1pfr (25)
where we have assumed that the probability of sending a
message within a frame is uniformly distributed and hence,
8Trans. Emerging Tel. Tech. 2014; 00:1–14 c
2014 John Wiley & Sons, Ltd.
DOI: 10.1002/ett
Prepared using ettauth.cls
D. Tarchi, R. Fantacci and D. Marabissi A M2M Cognitive MAC protocol for Overlaid OFDMA Environments
in average, the message is sent in the middle of the frame.
It is possible to derive the average service time as:
¯x=E{x}=Tf
pfr
−Tf
2(26)
Before deriving the mean squared value of the service time,
we define
x=x1−x2
where x1=kTfand x2=Tf
2, so that:
¯x2=E{x2}=E{x2
1}+E{x2
2} − 2E{x1x2}
=T2
f(2 −pfr )
p2
fr
+T2
f
4−2Tf
pfr
Tf
2
=p2
fr −8pf r −8
4p2
fr
T2
f
(27)
By substituting (26) and (27) in (22), (23), and (24), it is
possible to derive the average delay in the three selected
scenarios.
5. NUMERICAL RESULTS
This section deals with the numerical results obtained
through the analytical approach outlined in Section 4.2 and
computer simulations aiming to evaluate the system perfor-
mance and validate the obtained analytical predictions.
Concerning the primary system we consider to have a
LTE-A based network. To this aim we have supposed to
use a 10 MHz bandwidth, characterized by the presence
of 50 Physical Resource Blocks (PRB), each characterized
by the presence of 12 subcarriers and 6 OFDM symbols; by
considering to work with a QPSK modulation and a coding
rate Rcequal to 1/2, we have that R, the overall amount of
resources in each frame, is equal to 36000 bits [22]. As for
the frame duration Tf, we suppose that it is equal to 10 ms.
Concerning the M2M system we have considered
to resort to the quantities related to the ZigBee
specifications [36] concerning the packet size and
generation. To this aim we have considered that each
packet is composed by Lpequals to 1048 bits, 832 of
which are for the payload and the remaining for the
overhead. The token length jis supposed to be equal
to 192 bits. Finally we suppose that the travel time
between successive M2M devices is equal to one Tfthat
corresponds to say that each device has to wait one Tf
for sending data after the token is released by the previous
station.
The parameters used for deriving the numerical results
have been summarized in Tab. II for the readers’
convenience.
First of all we have focused our attention on the delay
analysis by comparing the theoretically obtained values,
following the analysis in the Section 4.2, with computer
simulation results. In Fig. 3, the performance in terms of
average delay for sending one packet is considered by
Parameter Value
Bandwidth 10 MHz
PRB per frame 50
PRB size 12 subcarriers ×6
OFDM symbols
Modulation order QPSK
Coding rate (Rc)1/2
Resources per frame
(R)
36000 bits
Frame duration (Tf)10 ms
Packet length (Lp)1048 bits
Payload 832 bits
Overhead 216 bits
Token length (j)192 bits
Table II. Parameters used for deriving the numerical results
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Secondary network traffic load [pkt/s]
Delay [s]
Exaustive
Gated
Limited
Exaustive (Theory)
Gated (Theory)
Limited (Theory)
Figure 3. Performance in terms of average delay for deliver one
packet
varying the loading factor of each device. The performance
has been evaluated by supposing that the idle probability
in the primary system pfr is equal to 10%, while the
secondary network traffic load has been considered to vary,
following a Poisson distribution; the effectiveness of the
selected value is also validated by taking into consideration
previous studies [6,7], where the selected idle resource
probability is considered as realistic. It is possible to
note that the performance obtained through computer
simulations almost agrees with the results obtained through
the theoretical analysis. This confirms the correctness of
the theoretical analysis.
After the theoretical evaluation of the proposed model,
the performance of the considered approaches has been
derived. In particular we have resorted to three different
scenarios, where we have evaluated the performance
by varying the free resource probability in the primary
network, the packet generation rate at the M2M network,
and the number of M2M devices, while the remainder have
been fixed.
We have focused our attention on four performance
indexes, namely ¯
D, the average delay that each secondary
Trans. Emerging Tel. Tech. 2014; 00:1–14 c
2014 John Wiley & Sons, Ltd. 9
DOI: 10.1002/ett
Prepared using ettauth.cls
A M2M Cognitive MAC protocol for Overlaid OFDMA Environments D. Tarchi, R. Fantacci and D. Marabissi
station has to wait for accessing to the network, the
throughput of the M2M devices, the resource wastage
expressed as the amount of resources of the primary
network that are not used by the M2M network, and
the average queue length at each M2M device evaluated
in the time instant in which the token is received by
each device. In the following we suppose that the data
packets generation rate at each device follows the Poisson
distribution, and that each simulated scenario is 1000
seconds.
Firstly, we have derived the performance by varying the
probability of free resources in the primary system. We
consider here that the overall average Poisson distribution
of generated packets by all the M2M devices ¯
λis equal to
10 pkt/s, and that the number of M2M devices, N, is equal
to 10; the selected values are consistent with the reference
scenario, where it is supposed to have a data traffic almost
equal to 10 kbit/s while 10 nodes can be considered a a
realistic value for a ZigBee cluster.
In Fig. 4we report the four above defined performance
indexes when varying the idle probability of the primary
network. By analyzing the throughput performance it
is possible to highlight that the exhaustive and gated
approaches allow to achieve the best results, while the fixed
and limited approaches, on the contrary, suffer in terms
of throughput; in particular the limited approach has an
intrinsic upper limit in terms of throughput. A specular
analysis can be done by considering the overall token delay
¯
D, where it is possible to highlight that, especially in case
of low resource probability the exhaustive and the gated
approaches suffer of a very high round trip time due to
their characteristic of going in a starvation state. At the
same time it is possible to see that the fixed and limited
approaches allow to maintain lower ¯
Dvalues. Similar
conclusions can be highlighted in the resource wastage and
queue length results, where the exhaustive and the gated
approaches allows to have a lower queue length at the
expense of a higher delay.
It is worth to noticing that the adaptive approach allows
to tradeoff between overall throughput performance and
¯
Dissues, by achieving throughput values similar to the
exhaustive and gated approaches while maintaining low ¯
D,
resulting in optimized performance values.
Moreover, it has to be noted in the exhaustive and gated
approaches that, despite the throughput is maximized, the
delay is very high; this is due to the fact that these two
approaches are unfair resulting in the hold of the token by
only one device and, hence, an increased average delay.
It is worth to be noticed that, due to the relationship
between the energy consumption per bit, as defined in (21),
and the throughput and the resource wastage, the adaptive
approach results to have the best performance in terms of
energy cost.
In Fig. 5the performance results are obtained by varying
the average value of packets generated by each M2M
device, by considering again to have a Poisson distribution
of packet generation. We have now fixed the amount of free
resources probability ¯pin the primary network to 0.1, i.e.,
10%, and that the number of M2M device Nis still equal
to 10; the free resource probability has been set to 10% also
according to the analytical study previously discussed.
The performance results show that the exhaustive
and gated techniques allow to increase the performance
in terms of throughput at the cost of a higher ¯
D,
that corresponds to a higher starvation probability. By
considering the adaptive approach, it is possible to
adapt the αparameter to the variable conditions of the
channel and traffic load for increasing the performance.
In particular by varying the secondary network traffic
load it is possible to see that the proposed adaptive
approach allows to have good throughput performance
while maintaining low delay, confirming what stated for
the variable idle resource probability cases.
Finally in Fig. 6the performance has been evaluated
by varying the number of devices while maintaining fixed
the idle resource probability to 10% and the M2M devices
average loading factor ¯
λto 10 packets per second. As
expected the throughput decreases when increasing the
number of M2M devices due to the increased number
of devices; however, it should be emphasized that the
proposed adaptive approach allows to achieve a low
delay, while having a higher throughput with respect to
the limited and fixed cases, corresponding to permit the
communication to a higher number of devices with a lower
delay.
6. CONCLUSION
Cognitive radio networks are a promising technique for
exploiting unused radio resources. We have focused our
attention on a M2M scenario where multiple M2M
devices exploit the free resources of an overlaid OFDMA
based primary network. We have proposed the DACT
protocol based on the polling principle for the secondary
network aiming to exploit the unused resources. In
particular five different approaches have been considered,
by highlighting their performance in an OFDMA cognitive
scenario; among others, the proposed adaptive approach
allows to tradeoff by optimizing the CM2M system
performance, increasing the throughput while minimizing
delay, resource wastage and average queue length with
respect to the other techniques. Both theoretical and
numerical results obtained through computer simulations
reveal the effectiveness of the proposed method for
allowing to set up a Cognitive M2M network within an
OFDMA environment.
REFERENCES
1. Akyildiz IF, Lee WY, Vuran MC, Mohanty S. Next
generation/dynamic spectrum access/cognitive radio
10 Trans. Emerging Tel. Tech. 2014; 00:1–14 c
2014 John Wiley & Sons, Ltd.
DOI: 10.1002/ett
Prepared using ettauth.cls
D. Tarchi, R. Fantacci and D. Marabissi A M2M Cognitive MAC protocol for Overlaid OFDMA Environments
0 0.02 0.04 0.06 0.08 0.1
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
Primary network free resource probability
Throughput [b/s]
Fixed
Limited
Gated
Exaustive
α adaptive
(a) Throughput
0 0.02 0.04 0.06 0.08 0.1
0
1
2
3
4
5
6
7
8
9
Primary network free resource probability
Resource wastage [%]
Fixed
Limited
Gated
Exaustive
α adaptive
(b) Resource Wastage
0 0.02 0.04 0.06 0.08 0.1
0
20
40
60
80
100
120
Primary network free resource probability
Overall token delay [s]
Fixed
Limited
Gated
Exaustive
α adaptive
(c) Overall token delay
0 0.02 0.04 0.06 0.08 0.1
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5x 106
Primary network free resource probability
Average queue length [b]
Fixed
Limited
Gated
Exaustive
α adaptive
(d) Queue Length
Figure 4. Performance results by varying the probability of free resources in the primary network.
wireless networks: A survey. Computer Networks
2006; 50(13):2127–2159.
2. Wu G, Talwar S, Johnsson K, Himayat N, Johnson
K. M2M: From mobile to embedded internet. IEEE
Communications Magazine Apr 2011; 49(4):36–43.
3. Li Y, Chai KK, Chen Y, Loo J. Duty cycle con-
trol with joint optimisation of delay and energy
efficiency for capillary machine-to-machine net-
works in 5G communication system. Transac-
tions on Emerging Telecommunications Technolo-
gies 2015; 26(1):56–69, doi:10.1002/ett.2891. URL
http://dx.doi.org/10.1002/ett.2891.
4. Karim L, Anpalagan A, Nasser N, Almhana J,
Woungang I. Fault tolerant, energy efficient and
secure clustering scheme for mobile machine-
to-machine communications. Transactions on
Emerging Telecommunications Technologies 2014;
25(10):1028–1044, doi:10.1002/ett.2801. URL
http://dx.doi.org/10.1002/ett.2801.
5. Yang B, Zhu G, Wu W, Gao Y. M2M access
performance in LTE-A system. Transactions
on Emerging Telecommunications Technologies
2014; 25(1):3–10, doi:10.1002/ett.2746. URL
http://dx.doi.org/10.1002/ett.2746.
6. Grønsund P, Pham HN, Engelstad P. Towards
dynamic spectrum access in primary OFDMA
systems. Proc. of IEEE PIMRC 2009, Tokyo, Japan,
2009; 848–852.
7. Pham HN, Grønsund P, Engelstad P, Grøndalen O.
A dynamic spectrum access scheme for unlicensed
systems coexisting with primary OFDMA systems.
Proc. of IEEE CCNC 2010, Las Vegas, NV, USA,
2010.
8. Zhang Y, Yu R, Nekovee M, Liu Y, Xie S, Gjessing
S. Cognitive machine-to-machine communications:
visions and potentials for the smart grid. IEEE
Network May/Jun 2012; 26(3):6–13.
9. Bader F, Demestricas P, DaSilva L, Harada H.
Cognitive radio in emerging communications sys-
temssmall cells, machine-to-machine communica-
tions, TV white spaces and green radios. Transac-
tions on Emerging Telecommunications Technologies
Trans. Emerging Tel. Tech. 2014; 00:1–14 c
2014 John Wiley & Sons, Ltd. 11
DOI: 10.1002/ett
Prepared using ettauth.cls
A M2M Cognitive MAC protocol for Overlaid OFDMA Environments D. Tarchi, R. Fantacci and D. Marabissi
0 5 10 15 20
0
0.5
1
1.5
2
2.5 x 104
Secondary network traffic load [pkt/s]
Throughput [b/s]
Fixed
Limited
Gated
Exaustive
α adaptive
(a) Throughput
0 5 10 15 20
0
0.5
1
1.5
2
2.5
Secondary network traffic load [pkt/s]
Resource wastage [%]
Fixed
Limited
Gated
Exaustive
α adaptive
(b) Resource Wastage
0 5 10 15 20
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Secondary network traffic load [pkt/s]
Overall token delay [s]
Fixed
Limited
Gated
Exaustive
α adaptive
(c) Overall Token Delay
0 5 10 15 20
0
1
2
3
4
5
6
7
8
9x 106
Secondary network traffic load [pkt/s]
Average queue length [b]
Fixed
Limited
Gated
Exaustive
α adaptive
(d) Queue Length
Figure 5. Performance results by varying the secondary network traffic load.
2013; 24(7-8):633–635, doi:10.1002/ett.2732. URL
http://dx.doi.org/10.1002/ett.2732.
10. Stevenson C, Chouinard G, Lei Z, Hu W, Shellham-
mer S, Caldwell W. IEEE 802.22: The first cognitive
radio wireless regional area network standard. IEEE
Communications Magazine Jan 2009; 47(1):130–
138.
11. Wang R, Lau VKN, Lv L, Chen B. Joint cross-
layer scheduling and spectrum sensing for OFDMA
cognitive radio systems. IEEE Transactions on
Wireless Communications May 2009; 8(5):2410–
2416.
12. Jeong SS, Jeong DG, Jeon WS. Nonquiet primary
user detection for OFDMA-based cognitive radio
systems. IEEE Transactions on Wireless Communi-
cations Oct 2009; 8(10):5112–5123.
13. Tu SY, Chen KC, Prasad R. Spectrum sensing of
OFDMA systems for cognitive radio networks. IEEE
Transactions on Vehicular Technology Sep 2009;
58(7):3410–3425.
14. Cao Y, Jiang T, Wang C, Zhang L. CRAC: Cognitive
radio assisted cooperation for downlink transmissions
in OFDMA-based cellular networks. IEEE Journal
on Selected Areas in Communications Oct 2012;
30(9):1614–1622.
15. Rahimzadeh P, Moradian M, Ashtiani F. Saturation
throughput analysis of a cognitive IEEE 802.11-
based WLAN overlaid on an IEEE 802.16e WiMAX.
Proc. of IEEE PIMRC 2012, Sydney, NSW, Australia,
2012; 1515–1520.
16. Choe S. OFDMA cognitive radio medium access
control using multichannel ALOHA. Proc. of ISWCS
2010, York, United Kingdom, 2010; 244–249.
17. Joshi GP, Nam SY, Kim SW. An analysis of
channel access delay in synchronized MAC
protocol for cognitive radio networks. Transactions
on Emerging Telecommunications Technologies
2014; 25(5):485–489, doi:10.1002/ett.2589. URL
http://dx.doi.org/10.1002/ett.2589.
18. Hussain F, Anpalagan A, Vannithamby R. Medium
access control techniques in M2M communication:
12 Trans. Emerging Tel. Tech. 2014; 00:1–14 c
2014 John Wiley & Sons, Ltd.
DOI: 10.1002/ett
Prepared using ettauth.cls
D. Tarchi, R. Fantacci and D. Marabissi A M2M Cognitive MAC protocol for Overlaid OFDMA Environments
101102103
0
2000
4000
6000
8000
10000
12000
# M2M devices
Throughput [b/s]
Fixed
Limited
Gated
Exaustive
α adaptive
(a) Throughput
101102103
1
1.5
2
2.5
3
3.5
# M2M devices
Resource wastage [%]
Fixed
Limited
Gated
Exaustive
α adaptive
(b) Resource Wastage
101102103
0
20
40
60
80
100
120
# M2M devices
Overall token delay [s]
Fixed
Limited
Gated
Exaustive
α adaptive
(c) Overall Token Delay
101102103
0
1
2
3
4
5
6x 106
# M2M devices
Average queue length [b]
Fixed
Limited
Gated
Exaustive
α adaptive
(d) Queue Length
Figure 6. Performance results by varying the number of M2M devices.
survey and critical review. Transactions on
Emerging Telecommunications Technologies
2014; :n/a–n/adoi:10.1002/ett.2869. URL
http://dx.doi.org/10.1002/ett.2869.
19. Zheng K, Hu F, Wang W, Xiang W, Dohler
M. Radio resource allocation in LTE-advanced
cellular networks with M2M communications. IEEE
Communications Magazine Jul 2012; 50(7):184–192.
20. Yucek T, Arslan H. A survey of spectrum sensing
algorithms for cognitive radio applications. IEEE
Communications Surveys and Tutorials First 2009;
11(1):116–130.
21. Papapanagiotou I, Toumpakaris D, Lee J, Devetsiki-
otis M. A survey on next generation mobile WiMAX
networks: objectives, features and technical chal-
lenges. IEEE Communications Surveys and Tutorials
Fourth Quarter 2009; 11(4):3–18.
22. Ghosh A, Ratasuk R, Mondal B, Mangalvedhe N,
Thomas T. LTE-advanced: next-generation wireless
broadband technology. IEEE Wireless Communica-
tions Magazine Jun 2010; 17(3):10–22.
23. Fantacci R, Tarchi D. A novel cognitive networking
scenario for IEEE 802.16 networks. Proc. of IEEE
GLOBECOM 2009, Honolulu, HI, USA, 2009.
24. Adhikary A, Ntranos V, Caire G. Cognitive fem-
tocells: Breaking the spatial reuse barrier of cellu-
lar systems. Proc. of 2011 Information Theory and
Applications Workshop, La Jolla, CA, USA, 2011.
25. Reconfigurable radio systems (RRS); cognitive pilot
channel (CPC). Technical Report TR 102 683, ETSI
Sep 2009.
26. Ji Z, Ganchev I, O’Droma M, Zhang X. A
realisation of broadcast cognitive pilot channels
piggybacked on T-DMB. Transactions on
Emerging Telecommunications Technologies 2013;
24(7-8):709–723, doi:10.1002/ett.2690. URL
http://dx.doi.org/10.1002/ett.2690.
27. Tarchi D, Fantacci R, Bonciani E. Analysis and
comparison of scheduling techniques for a BWA
OFDMA mobile system. Wireless Communications
and Mobile Computing Jul 2010; 10:888–898.
Trans. Emerging Tel. Tech. 2014; 00:1–14 c
2014 John Wiley & Sons, Ltd. 13
DOI: 10.1002/ett
Prepared using ettauth.cls
A M2M Cognitive MAC protocol for Overlaid OFDMA Environments D. Tarchi, R. Fantacci and D. Marabissi
28. Fantacci R, Marabissi D, Tarchi D. Adaptive
scheduling algorithms for multimedia traffic in
wireless OFDMA systems. Physical Communication
Sep 2009; 2(3):228 – 234.
29. Bartoli G, Tassi A, Marabissi D, Tarchi D, Fantacci
R. An optimized resource allocation scheme based
on a multidimensional multiple-choice approach with
reduced complexity. Proc. of IEEE ICC 2011, Kyoto,
Japan, 2011.
30. IEEE standard for local and metropolitan area
networks part 16: Air interface for broadband
wireless access systems. Std. IEEE 802.16-2009,
IEEE 2009.
31. Dahlman E, Parkvall S, Skold J, Beming P. 3G
Evolution: HSPA and LTE for Mobile Broadband.
3G Evolution Series, Elsevier Science: Oxford, UK,
2010.
32. Ergen M, Lee D, Sengupta R, Varaiya P. WTRP–
wireless token ring protocol. IEEE Transactions on
Vehicular Technology Nov 2004; 53(6):1863–1881.
33. Tirronen T, Larmo A, Sachs J, Lindoff B,
Wiberg N. Machine-to-machine communication
with long-term evolution with reduced device
energy consumption. Transactions on Emerging
Telecommunications Technologies 2013;
24(4):413–426, doi:10.1002/ett.2643. URL
http://dx.doi.org/10.1002/ett.2643.
34. Arienzo L, Tarchi D. Statistical modeling of
spectrum sensing energy in multi-hop cognitive radio
networks. IEEE Signal Processing Letters Mar 2015;
22(3):356–360, doi:10.1109/LSP.2014.2360234.
35. Hayes JF, Ganesh Babu TVJ. Modeling and Analysis
of Telecommunications Networks. John Wiley &
Sons, Inc.: Hoboken, NJ, USA, 2004.
36. Farahani S. ZigBee wireless networks and
transceivers. Newnes, 2011.
14 Trans. Emerging Tel. Tech. 2014; 00:1–14 c
2014 John Wiley & Sons, Ltd.
DOI: 10.1002/ett
Prepared using ettauth.cls
