Professional Documents
Culture Documents
Soltanmohammadi, Ghavami, Naraghi-Pour - 2016 - A Survey of Traffic Issues in Machine-To-Machine Communications Over LTE
Soltanmohammadi, Ghavami, Naraghi-Pour - 2016 - A Survey of Traffic Issues in Machine-To-Machine Communications Over LTE
fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
Abstract—Machine-to-Machine (M2M) communication, also re- automation, patient monitoring, auto-identification, automotive
ferred to as Internet of Things (IoT), is a global network of sensor systems, home energy management, transportation,
devices such as sensors, actuators and smart appliances which water distribution and smart electric transmission systems. It
collect information, and can be controlled and managed in real-
time over the Internet. Due to their universal coverage, cellular is predicted that IoT will connect up to 26 billion devices by
networks and the Internet together offer the most promising 2020 [1].
foundation for the implementation of M2M communication. IoT is also prosaically referred to as Machine-to-Machine
With the worldwide deployment of the fourth-generation (4G) (M2M) communication. Key components of an M2M include
of cellular networks, the long-term evolution (LTE) and LTE- advanced software components, intelligent and autonomous
advanced standards have defined several quality-of-service classes
in order to accommodate the M2M traffic. However, cellular sensors and actuators, and a communication network infras-
networks are mainly optimized for Human-to-Human (H2H) tructure to allow the sensors to disseminate their information
communication. The characteristics of M2M traffic are different to, and acquire data and intelligence form the outside world.
from the human-generated traffic and consequently create sever For communication, a plethora of short-range, local area and
problems in both radio access and the core networks. This survey wide area technologies are available including RFID, blue-
on M2M communication in LTE/LTE-A explores the issues,
solutions and the remaining challenges to enable and improve tooth, ZigBee Wi-Fi, wireless LAN (WLAN), generic DSL
M2M communication over cellular networks. We first present an (xDSL), fiber to the x (FTTx) and cellular networks. Due to
overview of the LTE networks and discuss the issues related to its pervasive nature, wide area coverage, and mobility support,
M2M applications on LTE. We investigate the traffic issues of cellular network has received the most attention for wide area
M2M communications and the challenges they impose on both M2M connections. With its rapid growth, M2M communi-
access channel and traffic channel of a radio access network
and the congestion problems they create in the core network. cation is expected to create significant revenue growth for
We present a comprehensive review of the solutions for these cellular network providers. The number of cellular network
problems which have been proposed in the literature in recent connections worldwide used for M2M communication was
years and discuss the advantages and disadvantages of each 47.7 million in 2008 [2].
method. The remaining challenges are also discussed in detail. Although data transmission has been on the rise in the
Index Terms—Internet of thing, Machine-to-Machine (M2M) cellular networks for human-involved applications in the last
communication, Long Term Evolution (LTE), Access channel, decade, cellular networks are mainly optimized for Human-
Traffic channel. to-Human (H2H) communication. However, characteristics of
M2M traffic are different from the human-generated traffic in
I. I NTRODUCTION the cellular networks. The main differences between H2H and
To date, the vast majority of internet traffic has been a result M2M traffic can be expressed as follows [3].
of human-to-human (H2H) communication via computers or • Different uplink and downlink traffic loads: In contrast to
mobile devices. However, in recent years, a major paradigm H2H, in M2M devices the uplink traffic is higher than
shift has been on the way. The Internet of Things (IoT) is downlink.
the new marvel of information technology; an evolution from • Temporal distribution of Traffic: While H2H traffic is
isolated closed systems to a gigantic Internet-enabled network mostly concentrated during daylight and evening, M2M
of connected devices with embedded technology that sense traffic is more uniformly generated by the M2M devices
the environment, find the state of phenomena of interest, throughout the day. For further traffic distribution char-
take proper action, send information to other devices and acteristics details, we refer to the paper by Tyagi et. al.
gain intelligence by receiving information from the network. [4].
This evolution creates enormous opportunities for factory • Periodic traffic: In some applications (e.g., involving
metering devices) M2M traffic is periodic.
Copyright (c) 2012 IEEE. Personal use of this material is permitted. • Bursty traffic: In some monitoring applications, the vol-
However, permission to use this material for any other purposes must be
obtained from the IEEE by sending a request to pubs-permissions@ieee.org. ume of traffic increases sharply after the detection of
Erfan Soltanmohammadi was with the Division of Electrical and Com- events.
puter Engineering, School of Electrical Engineering and Computer Science, • Mobility: For many classes of M2M devices, M2M de-
Louisiana State University, Baton Rouge, LA. Since May 2014 he has been
with Marvell Semiconductor, Inc., 5488 Marvell Lane, Santa Clara, CA 95054 vices have a much lower mobility than H2H devices.
{email: erfan@marvell.com} However, for the healthcare devices which are expected to
Kamran Ghavami and Mort Naraghi-Pour are with the Division of Electrical be worn by the users, and for accessories such as Google
and Computer Engineering, School of Electrical Engineering and Com-
puter Science, Louisiana State University, Baton Rouge, LA 70803 {e-mail: glass and Apple watch, the mobility are the same as the
kghava1, naraghi@lsu.edu}. H2H devices.
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
Fig. 4. The framing structure for downlink physical channels and physical
signals for FDD and normal CP. The extended CP and TDD have slightly
different structure.
divided into 6 or 7 OFDM symbols depending on whether UEs to transmit control information including channel
an extended cyclic prefix is employed or not. Each sub-carrier quality indication (CQI), ACK and NACK responses for
has a bandwidth of 15 KHz and each symbol by one 15 kHz HARQ, and uplink scheduling requests indicator (SRI)3 .
sub-carrier form a resource element, as shown in Fig. 3. A set • Physical Uplink Shared Channel (PUSCH): is allocated
of 12 consecutive sub-carriers, each of duration one slot form by the uplink scheduler in eNB. The resources for
a Resource Block (RB) which is the smallest unit of resources PUSCH are multiples of sub-frames.
4
that can be allocated to a user. Therefore, an RB occupies a • Physical Random Access Channel (PRACH ): is used
bandwidth of 180 kHz. Note that RB is the smallest resource by UEs to request access from eNB.
in LTE that can be allocated between M2M and H2H devices. • Uplink Reference Signal (RS): is used to facilitate the
In the following we discuss how these RBs are aggregated to coherent demodulation in the uplink, which uses SC-
form frames for use in uplink and downlink. FDMA.
1) Downlink physical channels: In LTE different parts of
downlink frames are assigned for different purposes known B. Random Access Request
as physical downlink channels and downlink physical signals
When the UE is switched on for the very first time, it starts
(see Fig.4) including2 :
searching for the network. In order to find the network of
• Physical Downlink Shared Channel (PDSCH): is used the operator to which it subscribes, the UE goes through an
to transport user data. The RBs in this channel are shared initial cell synchronization procedure, where it acquires the
among UEs. physical cell identification (PCI), and time slot and frame
• Physical Broadcast Channel (PBCH): is used to send
3 Scheduling request could be considered as the biggest issue of M2M
2 The
channels with bold font styles play an important role in M2M communication over LTE.
communications. 4 PRACH and RACH abbreviations are used interchangeably in this paper.
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
Fig. 5. Top: The framing structure for uplink physical channels and physical
signals. Bottom: PRACH is multiplexed with PUSCH and PUCCH
Step 3.
3) Terminal Identification: In this step, the UE sends its ID,
synchronization. This will enable the UE to read system and the needed additional messages.
information blocks from a particular network. 4) Contention Resolution: In this step, the eNB announces the
After synchronization is achieved, the UE goes through a IDs of all accesses granted to UEs.
random access procedure to inform the network of its desire to Since it is possible that more than one UE uses the same
connect. Generally, in LTE networks, the UE needs the random preamble and temporary identity in step 1, the eNB may or
access procedure for several reasons: 1) to join the network may not receive this preamble successfully. Therefore, the two
for the first time; 2) after a link break down; 3) to handover to last steps are required for verifying which of the UEs, if any,
a new cell; 4) for location updates 5) for scheduling requests. must get the access permission.
The random access procedure involves the transmission of a
preamble from a predefined set. There are 64 preambles in III. M2M T RAFFIC M ODELS
each cell. When the UE attempts to connect to the network for
the first time, the network cannot inform the UE about which Access requests, originated by M2M devices, are clas-
one of the 64 preambles should be used. Therefore in this case sified by 3GPP into two groups of uncoordinated/non-
a preamble is selected at random by the UE and this results synchronized and coordinated/synchronized traffic [6], [7].
in the possibility of collision if another UE is using the same Coordinated/synchronized traffic is a type of traffic generated
preamble. However, in the case of cell handover from one eNB by many similar M2M devices in reaction to an event5 ,
to another, the eNB informs the UE which preamble (selected whereas uncoordinated/non-synchronized traffic is a result of
from a small number of preambles set aside for contention-free independent reporting of data. As shown in Table I, depending
access), to use. on whether a limited or unlimited time span is considered,
UEs can send their access requests during allocated random different statistical models are proposed for each traffic group
access resources (previously described as PRACH). Based in terms of the access probability.
on the access load, waiting time, and bandwidth, PRACH is The access request for coordinated/synchronized traffic in
formed into several configurations for each preamble format. a limited time span of T ms can be modeled by the Beta
The detailed descriptions of each configuration can be found distribution given by, [6]
in Table 5.7.1-2 and Table 5.7.1-3 of [5]. As shown in Fig. 5,
tα−1 (T − t)β−1
PRACH is multiplexed with PUCCH and PUSCH. p(t) = (1)
As shown in Fig. 6, each random access procedure consists T α+β−2 beta(α, β)
of the following steps: in which, p(t) is the probability density function (pdf) of
1) Random Access Preamble Transmission: In this step, each access requests, beta(., .) is the Beta function, and α > 0,
UE sends its access request by transmitting one of 64 orthog- β > 0. Based on experimental results, in [6] it is suggested to
onal and predefined preambles, as well as a temporary identity set α = 3 and β = 4. This distribution can model the access
RA-RNTI (random access radio network temporary identity) requests in which all events take place between t = 0 and
which is actually based on the time slot in which the preamble t = T . This pdf is suitable for the coordinated/synchronized
is transmitted. behavior in which a collision may still happen even when the
2) Random Access Response: In this step, the eNB transmits access attempt rate is below the RACH capacity. Moreover,
the access response which contains the index of the detected M2M devices with high path loss still have a chance to get
random access preamble, the scheduling for Step 3, the timing access even under heavy load.
offset (so the UE can change its timing to compensate for the
round trip delay), and the uplink resources to UE for use in 5 For example, power meters delivering their measurements after an outage.
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
TABLE I
S UGGESTED DISTRIBUTIONS FOR M2M ACCESS TRAFFIC
Synchronized/Coordinated Non-synchronized/Uncoordinated
Limited time Unlimited time Limited time Unlimited time
span span span span
Beta - Uniform Poisson
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
ACB approach. The eNB estimates the access load by tracking information [36]. As in the case of separate back-off, the
the number of successfully decoded RACH requests, and then devices in the cell are classified into several classes based on
enables ACB. An approach similar to dynamic ACB called their delay sensitivity. The eNB broadcasts the system load
dynamic back-off is suggested by Lin et. al. in [31] in which information and some other parameters to devices in the cell.
the eNB dynamically calculates the back-off window size In the first method, which is a pure back-off scheme, the eNB
(which is equivalent to barring time), according to the RACH broadcasts a number in the interval [0, 1] which is proportional
traffic load. to the cell load. Then after a collision occurs, the device
3GPP shows in [32] that applying ACB for dense M2M generates a random back-off time as a function of its class,
networks cannot effectively compensate the channel access cell load, and the total number of M2M devices in the cell. In
loads. In addition, while the problem of access collisions can the mixed back-off and ACB scheme, the eNB broadcasts a
be mitigated by increasing the ACB threshold, the access delay number, named barring check parameter, in addition to the cell
will be increased. To alleviate this difficulty an evolutionary load. Then each device calculates its access barring threshold
game-based ACB scheme is proposed in [33]. In this scheme as well as its back-off time. By numerical simulations in [36],
the M2M devices which fall in the overlapped area of more it is shown that by using the proposed schemes, the collision
than one eNB are guided to select the available eNB to perform probability decreases. Note that, in these two approaches, each
the random access. An evolutionary game is formulated where M2M device must know the total number of M2M devices in
the game strategies are the the devices’ choice of the eNB the cell, which is hard to obtain. Moreover, for heavy load
and the payoff is the utility of the devices defined in terms of cells, the back-off time and barring threshold can be combined
the channel capacity and the proportion of devices choosing into a single random time, in which case either the back-off
each eNB. Simulation results show that the proposed method scheme or the ACB scheme can be used instead of the mixed
improves the access delay of M2M devices. scheme.
2) Adaptive S-ALOHA: S-ALOHA achieves its maximum
throughput for the access request rate of one per slot [37]. Li
B. Back-off Scheme
et. al. use this fact to control the access request rate. The eNB
To mitigate random access overload, several authors have sets an access request probability (similar to ACB method),
proposed to employ random back-off after collisions based on using the information on the number of free spaces in the
slotted-ALOHA and its modified versions. These include random access channel so that the access request rate is one
• Separate Back-off per slot.
• Adaptive S-ALOHA 3) Adaptive MACA: Wang et. al. in [38] model the M2M
• Adaptive MACA access traffic with a Poisson process with high arrival rates,
1) Separate Back-off: 3GPP in [34] suggested a separate and proposed a modified version of Multiple Access with
back-off scheme in which delay-sensitive M2M devices to- Collision Avoidance (MACA) in [39] for access control. In
gether with H2H devices form class 1, and delay-insensitive the slotted MACA, if the traffic load exceeds the optimal load,
M2M devices form class 2. Then, different back-off schemes the throughput decreases excessively. Therefore, they suggest
are applied to these two classes. In this approach, after expe- that the eNB broadcasts an access probability for the request-
riencing a collision in the random access channel, devices in to-send (RTS) messages in order to control the access load.
class 2 back-off for a longer random time. Through simulation, This method is similar to ACB algorithm.
it is shown in [34] that this scheme prevents excessive delays
for devices in class 1. Another good point of using this scheme C. Slotted Access
is to prevent changes in the current back-off scheme for H2H In [40], 3GPP proposes a method to alleviate access over-
devices. load for delay-insensitive M2M applications. The framing
Bello et. al. have suggested to design a virtual frame of structure of the LTE cellular communication is used, in which,
RA slots in which the users send access requests [35]. Each each device can use system frames (each of duration 10 ms)
M2M device scores each RA slot in the virtual frame using as access slots. In the proposed approach, each device should
the Q-learning method. The Q-learning method keeps track use its paging channel for both ordinary downlink paging tasks
of the success probability of access for each RA slot in the and up-link transmission of access requests. Thus, the system
virtual frame which will be used in the future to find the should provide the ability of pushing mechanism in the paging
best slot to send an access request. The main idea in this channel in addition to pulling. An access cycle is defined
paper is that for the initial access request both H2H and to be a set of L consecutive access slots, which is repeated
M2M devices can share the same frames, but after a collision periodically. The M2M devices in the cell are divided into C
happens, H2H devices and M2M devices use different back-off classes and a distinct number is associated with each class.
frames. Therefore the H2H devices are restricted from sending Since each system frame in LTE is identified by a System
requests in an M2M frame and also M2M devices are not Frame Number (SFN) with a period of 1024 frames, [41],
allowed to transmit in the H2H back-off frames. This method M2M devices can use SFNs to find their access slots. Each
improves the performance because the randomness of back-off M2M device is only permitted to send an access request in a
is decreased in comparison to pure Aloha. specific time slot in each access cycle. Specifically, an M2M
Jian et. al. have suggested a pure back-off scheme as well device in class c can send an access request in the system
as a mixed back-off and ACB scheme based on the cell load frame with SFN = s, if c mod L = s. In this scheduling
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
method, M2M devices should be synchronized with the system nodes with better channel quality have the opportunity to send
framing. Fig. 10 shows the timing diagram of the slotted access their access requests. In detail, an M2M device measures the
method for two devices from two different classes using their downlink channel quality, and if its channel quality is above
assigned RBs. The first device can send access request when an assigned threshold by the eNB, it has the opportunity to
SF N = s and the second device can send an access request send the channel request. The eNB can control the access load
when SF N = r. by changing this threshold. It is shown in [40] that by using
Since there are several LTE RBs or preambles in each slot, this method, the spectrum usage efficiency will be improved.
it’s possible to serve more than one access request in each
one. Therefore, the devices should still contend for access
E. Separate Resource Allocation
to resources in each slot. According to this fact, randomly
assigning the slots to H2H and M2M devices may decrease Allocating different access resources for H2H devises is an
the performance of this approach, while pre-assigning the assured way to minimize the effect of M2M devices on H2H
resources can increase the efficiency up to three times [19]. communication. This can be achieved by different resource
allocation methods such as
Comparison of slotted access, ACB, and back-off schemes
• Separate RACH Allocation
As explained in [19], the slotted access approach has several
• Dynamic RACH Allocation
advantages in comparison to the ACB method as listed below.
• Adaptive RACH allocation
1) The simulation results show that the slotted approach has
• Asynchronous Simultaneous Small Data Transmission
a better access success rate, even in the case of randomly
• Self Optimizing Overload Control
assigned access slots.
2) By slotted access, it will be possible to more uniformly 1) Separate RACH Allocation: Since the cornerstone of
distribute the access requests in slots and RBs, and each cellular network is based on H2H services, minimizing
achieve a contention-free mechanism. the effects of M2M applications on H2H services has a
3) In terms of complexity, while ACB imposes signaling high priority for network administrators. To this point, simply
overloads, slotted access uses the already provided pag- separating the allocated resources to H2H and M2M could
ing mechanism and does not add any further complexity. be the easiest approach. As discussed in [7], this separation
The only requirement of the slotted access method is approach in LTE can be done in two ways. The first way is
providing the capability for M2M devices to access their simply splitting the RBs or preambles into two groups of H2H
paging channels in uplink direction. and M2M resources. This is also proposed by Lee in [43].
4) Slotted access can also be used for more precise net- The separate access resources is also proposed by 3GPP
work overload control. Since network access requests in [44], in which the resources are divided into two groups
may have has a bursty nature, ACB’s access blocking providing different access opportunities for different devices.
resolution is not small enough to effectively control this In [45], Lee et. al. propose two methods based on access
type of traffic. However, the precise regulation in slotted resource separation. Similar to the approach above, in their
access allows the network to control the overload more first method, they propose a simple separation of access
precisely with the resolution order of a system frame. preambles between H2H and M2M devices. In the second
Table II shows a brief comparison of the three approaches, method, the preambles are split into two groups, such that
ACB, backoff and slotted access, proposed by 3GPP for one group is used by H2H devices only, and the second group
overcoming the LTE’s RACH congestion issues, [42]. The can be used by both H2H and M2M devices. Moreover, they
numerical results in this table are provided for a Beta traffic assume that access of the M2M devices is controlled by the
arrival process over 10 seconds. For the slotted access (shown ACB method. In this algorithm, the authors set the minimum
in light gray), three slot sizes of 256, 512, and 1024 are resource preambles for the second group such that the resulting
considered. The ACB simulations (shown in medium gray) throughput for M2M devices is the same as that of the first
are considered for three barring factors of 0.9, 0.7, and 0.5, approach. To guarantee the throughput of H2H devices, the
and 4, 8, and 16 seconds of barring time. Finally, three back- barring factor is used to set a limit on the average requests
off times of 20ms, 240ms, and 960ms are used in the back- from M2M devices. As stated in [45], to control the M2M
off simulations (shown in dark gray). The three methods access requests, the eNB decreases the barring factor to lower
are compared in terms of M2M access success probability, values, so that the expected number of requests remains fixed.
RACH preamble collision probability, and H2H VoIP success In this case, the variance of the number of requests will be
probability for different device populations. It can be seen that increased due to the fact that in the binomial distribution, if
in all scenarios, the slotted access method outperforms the the mean is kept constant, then the variance increases with
other mechanisms. Also, these results show that the slotted the number of independent trials. It is noteworthy that, while
access has the minimum effect on the H2H access traffic. this paper applies the Poisson distribution, using the Beta
distribution which is proposed by 3GPP, may lead to more
accurate results.
D. Opportunistic Access The complete numerical results for these two methods by
Another method proposed in [40] uses the fact that M2M Lee et. al. can be found in [46] where the authors evaluate
devices are experiencing different time varying shadowing. It the random access throughput for three scenarios. First, they
proposes an algorithm based on the channel quality, in that the assume the H2H devices have low arrival rates. In this case,
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
10
Fig. 10. Timing diagram of the slotted access method where two different devices from two different classes are using the assigned RBs.
TABLE II
P ERFORMANCE OF THE SLOTTED ACCESS ( LIGHT GRAY ), ACB ( MEDIUM GRAY ), AND THE BACK - OFF ( DARK GRAY ) APPROACH UNDER (10 S , B ETA )
SCENARIO .
they show that there is a value for the arrival rate of the M2M any similar type activity, it must start the traditional RACH
devices, below which the second method is slightly better than procedure and contend with other H2H and M2M devices. The
the first method, but above that the first method is superior. In eNB removes the allocated RBs to a certain type, if it does
the second scenario, a moderate arrival rate of H2H devices not receive any access requests from the devices of that type.
is considered. A similar behavior to the first scenario was The simulation results in [47] shows that in comparison to the
observed for larger M2M arrival rates. Finally, in the third ordinary static RACH allocation, using the dynamic allocation
scenario, they show that for extremely high H2H arrival rates, method will provide large improvement in the access success
the second method results in a better throughput. This result probability as well as in access delay.
is intuitively expected, since allocating fewer access resources In [48] the authors have proposed a dynamic resource allo-
for M2M devices will provide more access opportunities for cation procedure for LTE-A using a game theoretic framework.
H2H devices. In this approach the random access resources are divided into
2) Dynamic RACH Allocation: Since M2M traffic is bursty, three groups for H2H, for M2M and for hybrid utilization.
and numerous nodes try to access the network in event occa- A game is formulated and it is shown that a mixed strategy
sions, using fixed static channel separation may be wasteful of Nash equilibrium exists for which the M2M devices will have
resources. In contrast, a dynamic algorithm can more smoothly a positive probability of transmitting preambles. A maximum
assign the access resources to H2H and M2M devices and likelihood estimation is also proposed to estimate the number
improve resource utilization. of M2M devices and the hybrid preamble pool. It is shown
through simulations that the performance in the case of esti-
In dynamic RACH allocation, as proposed in [47], M2M
mated parameters is almost the same as the case where the
devices are categorized by types. In this approach, when eNB
parameters are perfectly known.
accepts the access request from an M2M device, in addition
to granting access, it also allocates some RBs to devices from 3) Adaptive RACH resource allocation: To adaptively allo-
the same type, in which M2M devices from the same type can cate the RACH resources in LTE, [49] suggests a mechanism
content for access resources. As it is described in Algorithm 2, for cellular networks assuming a Poisson process model for
in the proposed method, each M2M device must listen to the the access traffic. They define an optimization problem to
channel, to determine if there is any activity from devices with minimize the RACH resources while keeping the average
the same type. Then it can start contenting with them in the delay below a certain threshold. A moving average filter over
assigned contention area. As shown in Fig. 11, this contention the arrival rates and the periodicity profile of the users is
area is located inside the RACH. If the device cannot find utilized by the eNB to predict the upcoming access rate.
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
11
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
12
and for each access group, a cluster head is assigned. Based and the main eNB at the macro cell. This paper proposes
on this grouping, they propose to change the LTE’s RACH some interesting solutions concerning the inter-cluster access
procedure to five steps: 1) the eNB pages a desired group, 2) interference issues. Similar to traditional LTE networks, cluster
the cluster head at this group sends the access preambles to members use preambles and the similar 4-step Random Access
the eNB, 3) the eNB sends the random access response to that Procedure (see section II-B) to send access requests to their
group, 4) the terminal identification is sent to the eNB, 5) the cluster heads. Since the neighboring clusters may use similar
eNB assigns resources by sending contention resolution. They preambles, a threshold on the magnitude of received preambles
also proposed a group-based procedure for uplink data report. is used in each cluster head to detect the corresponding
In this mechanism, the devices in each access group send preambles. This ensures that the cluster head in the machines
their data to the cluster head. A resource reuse mechanism cluster can successfully detect the information and feed back
is proposed such that the resources can be used in different to the machines. In the feedback information the cluster heads
access groups at the same time, which provides better resource must transmit its ID so that the machines in the neighboring
utilization. The simulation results show that increasing the size clusters do not decode the information. Only the machines in
of the access group reduces the average access delay. the desired cluster will decode the information.
A group-based extension to LTE RACH procedure is sug- In [61], Niyato et. al. consider the effect of cluster size on
gested by Lee et. al. in [56]. In the proposed mechanism, it is the total cost of the system considered to be the superposition
assumed that the devices are grouped, and in each group an of installation cost, packet delay cost, and packet loss cost.
M2M device is supposed to periodically transmit the access The authors investigate the cost dependency on the packet
preambles to the eNB. And, in the second stage, the eNB generation rate for different cluster sizes and a fixed number
grants the uplink access resources. In the third stage and in of nodes. They show that for low packet generation rates, the
each group, resources are randomly allocated to M2M devices cost is lower for larger cluster sizes. On the other hand, for
and they send their required resources. And, the contention higher packet generation rates, the cost is higher for larger
resolution is done at the forth stage. The simulation results cluster sizes.
in [56] shows a decrease in random access delay by applying A cluster based network structure to solve the RACH issue
this mechanism. A similar grouping mechanism is suggested of M2M over LTE is suggested by Lee et. al. in [62]. In their
in [57], in which group based timing control is used for RAN proposed network, M2M devices can access the eNB either
overload control. directly or via cluster heads. The cluster heads are responsible
Lien et. al. propose a decision making procedure upon for transmitting the aggregated data of M2M nodes to the eNB.
receiving the data about the characteristic (access rate) and Hence, to implement these two access methods, they suggest
requirement (delay) after the third step of RACH procedure that direct connection to the eNB is over LTE channels, and
in LTE [58]. In this approach, the M2M devices are grouped any connection between M2M devices and the cluster heads
based on access request rate and maximum tolerable delay. are by using cognitive radio links in TV white space. They
Corresponding to these requirement and characteristics the show that using this mechanism the number of nodes using
eNB allocates different access slots. After receiving the device direct access can be decreased. In addition, they show that the
ID, the requirement, and characteristic at the third step, the optimal cluster size and the optimal number of cluster heads
eNB check to see if there is an assigned group for this device. increase with increasing the cognitive radio resources. In their
If such group exists, then the eNB only checks to see if there model, they do not consider the aggregation algorithm and
is enough resource to support one more device or not. The accessing mechanism of cluster heads to the eNBs, which can
device can be served upon the existence of enough resources. influence the number of RACH requests. As they note, the
If there is no such group, and the requirement does not violate proposed approach does not fulfill the requirements of delay-
other groups, a new group will be created for that specific sensitive applications.
requirement and the device will be served. They show that Since both intuitively and practically, clustering can solve
this method improves the access delay performance. the RACH issue of M2M networks, we discuss in more detail
In [59] the authors have provided a dynamic version of the proposed clustering algorithm by Azquez-Gallego et. al.
the decision making procedure. The procedure is similar to in [63] in order to provide a more clear vision for the reader.
the above method but the scheduler is modified to monitor In their model, the cluster heads are capable of accessing into
the occupancy of each sub-frame. When it identifies the sub- both a local radio network and the cellular network, while
frames are not being used, they are reassigned to new M2M typical M2M devices can only connect to the local radio
devices. network. For M2M devices to access the M2M server, they
Wang et. al. propose a cluster-based approach for mitigating need an intermediate connection to a cluster head. The authors
the inefficiencies of the ACB algorithm [60]. In their proposed also assume that each M2M node has prior information about
model, all M2M devices are grouped into different clusters and its neighbors and its cluster head. If any M2M device wants
access the eNB via a cluster head while the H2H devices ac- to transmit data to the M2M server, it must use the following
cess the eNB directly. In this method, cellular communication algorithm:
is also used in links between cluster heads and their members. Each beacon is followed by a specific number of slots that
There are several aspects that needs to be precisely considered. can be used by other neighboring M2M devices that have
For example power control (adjustment) in each cluster is re- data to send. The authors have not explained how an M2M
quired in order to avoid interference between cluster members device can find the proper cluster head, and how a cluster head
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
13
begin Tu et. al. in [67] and Ho et. al. in [68], also suggest a
Initialization: clustering mechanism (not for any specific cellular standard)
An M2M device, A, has some data to send to M2M to decrease the power consumption. In their model, the effects
server: of collision and retransmissions are not considered. They
A: send a request to the associated cluster head, G; assume that all the channel coefficients are known. Then,
G: start transmitting periodic beacons; an optimization problem is defined to minimize the overall
while The network is active, or G’s activity duration power consumption when the nodes can be grouped up to a
has not ended do maximum number of clusters. The solution in terms of clusters
G: transmit a beacon; and cluster heads are found using the K-means algorithm.
A: wait for your slot;
A: wake up; H. Strong Coding/Modulation Scheme
A: negotiate with G for a dedicated slot; A new RACH procedure is suggested by Zhou et. al. in [69].
A: transmit data to G; To reduce the connection establishment latency, it is suggested
A: enter sleep mode; to remove the first two steps of LTE RACH procedure. In
end
the proposed procedure, the eNB broadcasts the allocated
end
resources, and then the M2M devices send their data and ID
Algorithm 3: Clustering mechanism proposed in [63] together. It is assumed that the transmission duration of data
and ID is equal to the first step of LTE RACH procedure.
To increase the probability of correct reception of IDs, it is
dedicates different slots to the cluster members. assumed that strong modulation and channel coding schemes
Kwon and Choi divide the M2M devices into different are used for ID transmission. Then with a high probability,
groups based on their SIR (Signal to Interference Ratio) the transmitted ID can still be decoded even in the case of
requirements [64]. They investigate the problem for noiseless collisions. Therefore the eNB can ask for data retransmission.
communication link and include channel fading characteristics. The suggested duration for the data packets is assumed to be
In their model they set the position of M2M nodes and eNBs less than that of LTE preamble, which may not be practical
according to the Poison Point Process (PPP). They solve the in all applications. Moreover, using strong modulation and
optimization problem for minimizing the number of resources channel coding degrades the total throughput. Also, it is
(time or frequency needed for random access channel) with noteworthy that preambles in the contention part of the LTE
respect to the required SIR for all groups. They do not consider RACH procedure are small (see section II-B), in order to
other factors, such as latency, in defining the QoS. Finally the reduce the collision probability, resource wastage , etc, [41].
positions of the eNBs are assumed to be random.
In [65], Pratas and Popovski propose to use the standard
I. predictive uplink resource allocation
cellular terminals such as the hand sets as relays for supporting
the M2M traffic. In this way the M2M traffic does not have a A predictive scheduling for uplink resource allocation is
direct link to the eNB but rather through nearby relays. They suggested by Brown and Khan in [70]. In LTE after a
propose two network-assisted schemes that enable cooperation device has established connection by RACH procedure, it can
between M2M devices and the standard cellular devices, so schedule the next uplink data transfer at the end of the current
as to meet the outage requirements of M2M devices while data transfer. Brown and Khan suggest that if an M2M device
maximizing the rate of the broadband services for the other asks for future uplink resources, then the eNB predicts the
devices. future requests from the neighbors of that M2M device. Then,
the appropriate uplink resources will be allocated to those
neighbors and reported to them. The proposed algorithm is
G. Energy-saving Access Mechanisms applicable to event driven M2M devices, which are in RRC-
A cluster based energy-saving mechanism for random ac- Connected state (see section II-B). This approach is suitable
cess overload control is suggested by Sun et. al. in [66]. In the for M2M devices which do not have power limitations, such
proposed model, they assume two different links, one is based as in Smart Grid networks.
on WLAN and connects M2M devices to each other using a
cluster head, and the other one is based on the cellular protocol J. Learning-based eNB Selection
and connects the cluster heads to the eNB. It is also assumed
Hasan et. al. in [71] suggest the access delay of M2M
that the M2M devices have periodic and static traffic, each
devices as a QoS measure. Since in LTE networks, it is
M2M device is capable of being a cluster head, each M2M
possible for the M2M devices to be covered by more than one
device has only one packet to send with known transmission
eNB, the authors propose the Q-learning approach for eNB
power, and a specific number of sub-carriers from the eNB
selection based on QoS.
are allocated to the cluster heads. Their solution is based
on a two-stage optimization. In the first one, based on the Pull-based Mechanisms
K-means algorithm, the optimal clusters, cluster heads, and The pull-based approaches are used by the M2M server in
cluster members are selected. In the second stage, the optimal collaboration with the cellular network in order to collect data
resource (e.g., sub-carrier) allocation is developed. from M2M devices, [7], [20]. This mechanism is useful for
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
14
reading the smart meters’ data [72] say in smart gird network.
While this mechanism can simply mitigate the overload issues
at RACH, it may create overload in the paging channel. The
paging channel overload is discussed in [21], [22].
Wei et. al. in [73] develop an analytical model for per-
formance evaluation of group paging in LTE. In the group
paging approach, after a number of M2M devices are paged,
they will try to access the network in a highly synchronized
manner. Moreover, after each M2M device is served, the
access probability of the other M2M devices will be changed.
Therefore the previous analysis on the throughput and access
delay cannot be used here. Wei et. al. suggest an analytical
Fig. 12. Three different models of carrier aggregation. From top to bottom:
model by including LTE specifications in: random back-off Intra-band contiguous allocation, Intra-band non-contiguous allocation, Inter-
procedure, capacity of access channel, the maximum number band non-contiguous allocation.
of retransmissions, and the power ramping exponential law.
They show that all their analytical results are matched with
LTE system simulations.
2) Advanced MIMO techniques: Advanced MIMO tech-
V. DATA C HANNEL OVERLOAD niques, [77] are suggested for LTE-A to increase the capacity.
In LTE-A higher order multi-antenna MIMO is used for down-
Compared to H2H devices, the size of the data for each link and uplink channels, where in downlink both transmitter
M2M devices is small. However, due to the massive number and receiver are equipped with up to 8 antennas, and in the
of M2M devices, the aggregated traffic is noticeably large. In uplink by up to 4 antennas. For high SNR, this number of
this section, we describe the suggested mechanisms to handle antennas are used for different space-time coding (MIMO
this new overload in cellular networks. We first review the techniques), [78], and for low SNR they can be used for
enhancements included in LTE-A to increase the cell capacity. diversity combining schemes, [79]. Both MIMO and diversity
Next we review other possibilities for improving the traffic combining increase the channel capacity.
data for M2M networks.
3) Wireless relays: Relay Nodes are low power eNBs with
reduced functionality, [80], [81]. They are planed in the cell
A. 3GPP Solutions to increase Capacity in LTE-Advanced to enhance the coverage and capacity at cell edges. The cell-
LTE-Advanced (LTE-A) has envisioned new techniques to edge devices can be connected to the eNB via an intermediate
achieve higher data rates in both uplink and downlink, [74] relay node.
including:
4) Enhanced Inter-Cell Interference Coordination (eICIC):
• Carrier aggregation
Inter-Cell Interference Coordination (ICIC), [82] is used in
• Advanced MIMO techniques
order to attain high spectral efficiency and to improve the
• Wireless relays
channel condition for the users under interference. The eNB
• Enhanced Inter-Cell Interference Coordination (eICIC)
allocates a wide range of frequencies to the cell area and to
• Coordinated Multipoint (CoMP) transmission/reception
avoid interference with the adjacent cells, the eNB allocates
1) Carrier aggregation: Carrier aggregation increases bit different frequencies to cell edges. To assure low interference
rates for all users in a cell, and facilitates bursty applications. and good QoS at the edges, the eNB uses low transmission
It can be used for both FDD and TDD modes of operation, power for areas near the center of the cell and higher trans-
[75], [76]. It increases the transmission bandwidth up to 100 mission power for the edge areas, as shown in Fig. 13. This
MHz by aggregating up to five LTE carriers. This enhancement fact is known as Fractional Frequency Reuse (FFR). In LTE
allows to use non-contiguous carriers and to manage con- release 10, this approach has been enhanced by considering
trol channels in heterogeneous networks6 . Carrier aggregation the heterogeneous network structure, known as eICIC.
works in three different modes:
1) Intra-band contiguous allocation: uses up to five adjacent 5) Coordinated Multipoint (CoMP) transmission/reception:
carriers at the same band as shown at the top of Fig. 12. Due to the use of fiber optics in the back-haul network, the
2) Intra-band non-contiguous allocation: uses up to five latency and capacity of the links between eNBs has been
non-adjacent carriers at the same band, as shown in the improved. This improvement enables features such as clock
middle of Fig. 12. synchronization and synchronous data exchanges between
3) Inter-band non-contiguous allocation: uses up to five eNBs. On the other hand, this allows a device to be served
non-adjacent carriers at different bands, as shown at the by more than one eNB, which is known as Coordinated
bottom of Fig. 12. Multi-Point (CoMP) transmission and reception, [83]. By
combining signals from multiple sources, a higher bit rate can
6 Here, heterogeneous networks refers to networks of macro-cells, micro- be achieved, and cell-edge devices can also experience a better
cells, etc. QoS.
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
15
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
16
uplink channel [92]. They suggest to assign the resource blocks its imposed computational complexity on each node are not
to H2H devices first, and then the remaining RBs are assigned investigated.
to M2M devices by using two different methods. In the first Abdalla and Venkatesan in [100] propose two separate
method, for each RB, the M2M devices are sorted based on the resource unit (RU) allocation in LTE uplink channel, such
SNRs. If the device with best SNR is delay sensitive, then that that specific percentage of resources dedicated for H2H users
RB will be allocated to it. In the second approach, the M2M and the remainder for M2M devices. They propose to allocate
devices are sorted according to the tolerable delay and the resources base on the QoS requirements.
device with lowest delay gets the RB with the best SNR. The Shin and Triwicaksono suggest an adaptive RB allocation
main issue for these method is that the eNB should estimate for H2H and M2M devices in LTE network [101]. They also
the SNR for each individual device over all the RBs. compared their algorithm with two previously suggested algo-
Delgado-Luque et. al. in [74] apply and evaluate three rithm for resource allocation. In proportional fair algorithm,
different scheduling schemes previously proposed by [93]– the eNBs assign the resources based on the channel quality
[96] in LTE up-link traffic Channel. In Opportunistic Hard feedback from nodes. In round robin (RR) based algorithms,
Priority (OHP), which is proposed in [93], [94], as long as the eNBs do not consider any quality feedback from nodes,
the packet delays are under a threshold, a uniform priority is and schedule the resources in a fixed pattern. While these
allocated to all packets. Otherwise, the more delay sensitive two allocation methods are fair, their resource allocations
packets will be served sooner. The other algorithm, Channel is not optimum. In the proposed algorithm in [101], the
Dependent Earliest Deadline Due (CD-EDD), suggested in number of unoccupied RB by H2H devices, are adaptively
[95], combines a channel dependent term with delay to pro- dedicated to M2M nodes. The simulation results show that this
duce a priority label for each packet. Finally, as a modification algorithm improves the peak throughput, while achievement in
to the previous method, in CD-EDD with postponed EDD the average throughput performance is small.
term, as suggested in [96], the dominance of the channel 5) Novel waveform design: As mentioned in Section II,
dependence term has been increased. In [97], it is shown that OFDMA is used in the downlink of LTE to allocate a subset
all scheduling schemes outperform the non-scheduled system. of subcarriers to each user. To avoid intercarrier interference
While the delay performance improves in low SNRs, it is (ICI) in OFDMA, the signals from different users must be
shown that the latter algorithm has a better performance in synchronized at the receiver input. This is easily achieved in
all SNRs. the downlink since all the subcarriers are transmitted from
Zhenqi et. al. in [98] define two sets: set 1 includes H2H the same point (eNB) and they undergo the same Doppler fre-
devices and delay-sensitive M2M devices, and set 2 includes quency shift before reaching each user. However, synchroniza-
delay-insensitive M2M devices. Through timing scheduling, tion is not trivial in the uplink where signals are transmitted
the eNB first serves set 1 in a long duration and then it serves from a number of geographically distributed mobile users7 . In
set 2 in a shorter duration. Although, if there is no more traffic the absence of perfect synchronization, ICI occurs and results
from set 1, then the eNB serves set 2 in the first interval. In in significant system performance degradation. For this reason
detail, they define a metric to measure the QoS that each RB OFDMA is not used in LTE uplink.
can provide for each device. They first assign the RBs to set ICI in OFDMA is particularly detrimental due to the
1 according to each device’s metric and after serving them , high sidelobe levels in subcarrier spectrum which extend
they similarly serve set 2. In the numerical result, they show to a wide frequency band. To overcome this effect, filter-
that the proposed method decreases the delay while keeping bank based multicarrier systems have been proposed where
the throughput fixed. prototype filters are designed to ensure much lower sidelobe
A customized packet scheduling (PS) algorithm of LTE levels. The universal filtered multicarrier (UFMC) and filter
for M2M devices is proposed by Gotsis et. al. in [92] with bank multicarrier (FBMC) introduced in [102], [103] and
including three considerations: specific traffic model, queue [104], respectively, reduce the out-of-band sidelobe levels
awareness, and clustering. In their proposed mechanism, they and minimize the effects of ICI between adjacent users.
modeled M2M traffic by Poisson process, let the eNBs be UFMC and FBMC provide a number of advantages over LTE
aware of the queue length at each M2M devices, and classify waveform including good spectrum containment, increased
the M2M devices into limited number of groups. It is note- spectral efficiency, relaxed synchronization requirements and
worthy that this approach is a good fit for event-driven M2M compatibility with LTE pilot pattern and MIMO modes.
applications and not for polling. 6) Novel multiple access schemes: The requirements of an
extremely high number of connected devices and low end-to-
Off-peak transfer approach is suggested by Kitahara et. al.
end latency are not easily achieved through LTE. Therefore,
in [99]. They try to shift the delay-insensitive traffic of M2M
in addition to the physical layer enhancements, the multiple
devices to the off-peak hours, which helps to distribute the
access scheme also needs to be improved. Multi-user MIMO
traffic more uniformly during a day. They suggest an adaptive
(MU-MIMO) has already extended the multiplexing space
approach which is implemented on each M2M devices rather
into the spatial domain. Two other candidates proposed for
than eNBs. In their algorithm, each day is divided to specific
multiplexing are the power domain and the code domain. One
number of time slots, and then each node evaluates the
promising approach is to relax the orthogonality of OFDMA
throughput of its working time slot. If the throughput is not
good enough, it will move to a new time slot for the next 7 The signals may undergo different Doppler frequency shifts and may also
data transmission. The algorithm’s convergence, speed, and have different carrier frequency offsets.
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
17
from the multiple access perspective. Non-orthogonal multiple specific rate µi [116]. Then they formulate the problem of
access (NOMA) multiplexes users in the power domain by allocation the incoming rate, λi (≤ µi ), for eNB i to maximize
exploiting the channel gain difference among different users the total incoming rate in the networks of eNBs subject to
in the cellular network. This approach can enhance the tech- a constraint defined as incoming rate cost. Their analysis is
nology of macro cells [105] and can be combined with MU- based on modeling each eNB as M/M/1 queue. They verified
MIMO to increase spectral efficiency and also the cell-edge that increasing the incoming rate (average waiting time in
throughput. Sparse Code Multiple Access (SCMA) [106] is queue) increases delay and cost.
another promising approach to increase the throughput by
providing code multiplexing on top of power multiplexing. C. Session Management Back-off
7) Context aware networking: Using a small side informa- 3GPP classifies the M2M devices based on their service
tion known as data context can enhance the network through- priorities [117]. The priority level of each device is determined
put. The context can be information on a device location and by the manufacturer or the M2M operator. When an M2M
prediction of its future location, QoS requirements, and/or data device registers its location in the Mobility Management
size. This side information can help the network to adapt the Entity (MME), its priority identifier will be sent to Serving
resource allocation, solve the mobility related issues, reduce Gateway (SGW) and Packet Data Network Gateway (PDN-
call drops and improve combined M2M and H2H traffic issues GW). Later, this information is used for packet scheduling and
[107], and also alleviate backhaul management issues [108]. also for congestion control, using a back-off scheme. When
a large number of UEs attempt to communicate via eNB,
VI. C ORE N ETWORK OVERLOAD the mobile core network aggregating the eNBs’ traffic may
Other than RAN issues raised by massive number of M2M experience congestion due to the high load. To resolve this
devices, Core Network (CN) is also prone to be affected. An congestion, a back-off mechanism is developed by 3GPP in
analyze of the congestion issue of M2M in the LTE’s CN is which when a UE sends a Service Request message to the
investigated by Yang et. al. in [109], by considering M/G/k/k core network after establishing the RRC connection, MME
queuing system model for Poisson arrival M2M process. An sends a Service reject to the UE with a back-off timer value.
approach similar to ACB is suggested by Ksentini et. al. The back-off time is set according to the priority level of
for controlling the CN congestion [110]. In this method, the the M2M devices. The M2M device is not allowed to initiate
barring factor is adjusted using a PID controller, in which data any mobility management procedures until the timer expires,
queue in core notwork is always compared with a reference except for emergency services. This control mechanism is used
value. To prevent and control congestion in Core Network to alleviate congestion at the SGW and PDN-GW and is known
(CN) the previous RACH control mechanisms like ACB, as session management back-off (SM Back-off).
EAB, slotted access, etc can be used though there are a few
approaches proposed for congestion in the CN. VII. C ONCLUSION
In this survey we investigated the issues imposed by M2M
A. Extended Wait Timer communication on cellular networks, specially LTE and LTE-
A networks. We briefly reviewed the LTE networks in terms
When eNB notified the traffic overload in CN, it can use
of frame structure, downlink and uplink physical channels,
an extended wait time scheme to inform user equipment with
random access request mechanism, and the LTE MAC sched-
low priority or delay insensitive traffic not to request within
uler. Next, we investigated M2M traffic models and the effect
an specific period of time, [111]–[113]. The extended wait
of synchronized/coordinated traffic on access channels. Then
time is an enhancement to defer channel access by UEs,
the simulation models for M2M traffic access requests are
[114]. [113] suggests to use passive or active mechanisms for
presented.
applying extended wait time to the relevant user equipment. In
The first major problem imposed by M2M communication
passive mechanism a connection request is rejected in RACH
is the overload in the access channel. We investigated the
procedure, or a message to release the dedicated resources
proposed solutions, their advantages and disadvantages and the
is send to the relevant UEs, and a waiting time is assigned in
remaining challenges for this problem, including access class
which the UEs should not send a request. In active mechanism,
Barring scheme, back-off scheme, slotted access, opportunistic
once CN overload happens, the eNB broadcast a waiting time
access, separate resource allocation, group-based approaches,
information over the network. Then the UEs with low priority
energy-saving access mechanisms, strong coding/modulation
and delay insensitive traffic waits until after the waiting time.
scheme, predictive uplink resource allocation, and learning-
A randomization maybe applied to the wait time to avoid
based eNB selection. We described each of these schemes and
collision for the upcoming access attempts. This randomization
any possible modifications in details. The comparisons among
can be achieved using the timers for ACB which is already
these methods were also presented.
considered in LTE, [115]. A delay tolerant indicator can be
The second major problem imposed by M2M communi-
used for determining the delay sensitive devices.
cation is the overload in the data channel. We investigated
the 3GPP proposed technologies to increase the capacity in
B. Traffic Rate Allocation for Different eNBs LTE and LTE-Advanced including carrier aggregation, ad-
Yao et. al. look at the problem of rate allocation in a network vanced MIMO techniques, wireless relays, enhanced inter-
of eNBs. In this network, each eNB, i, can serve up to a cell interference, coordination (eICIC), coordinated multipoint
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
18
(CoMP) transmission/reception. enhancing cell-edge M2M de- [22] 3GPP:R2-102962, “RACH overload control for MTC devices,” 3GPP
vices, overhead reduction, cooperative broadcast, and separate TSG RAN WG2 Meeting 70, Montreal, Canada, 2010.
[23] 3GPP:TD-S2-100, “MTC signal congestion control,” 3GPP TSG SA
resource allocation. WG2 Meeting 77, Shenzhen, China, 2010.
Finally, the core network overload was investigated and the [24] H. Wu, C. Zhu, R. La, X. Liu, and Y. Zhang, “Fast adaptive s-
solutions including extended wait timer and different traffic aloha scheme for event-driven machine-to-machine communications,”
in Vehicular Technology Conference (VTC Fall), 2012 IEEE, 2012, pp.
rate allocation for different eNBs were explained. 1–5.
[25] 3GPP:R2-100182, “Access control of MTC devices,” 3GPP TSG RAN
WG2 Meeting 68bis, Valencia, Spain, 2010.
R EFERENCES [26] 3GPP:R2-103143, “Discussion on separating RACH resources for
MTC,” Alcatel-lucent Shanghai Bell, Alcatel-lucent, 2010.
[1] Gartner, “Gartner says the internet of things installed base will grow
to 26 billion units by 2020,” 2013. [27] A. Larmo and R. Susitaival, “RAN overload control for machine type
[2] Berg Insight, “The global wireless M2M market,” 2009. communications in LTE,” in Globecom Workshops (GC Wkshps), 2012
IEEE, 2012, pp. 1626–1631.
[3] M. Shafiq, L. Ji, A. Liu, J. Pang, and J. Wang, “Large-scale mea-
surement and characterization of cellular machine-to-machine traffic,” [28] R. Cheng, J. Chen, D. Chen, and C. Wei, “Modeling and analysis of
Networking, IEEE/ACM Transactions on, vol. 21, no. 6, pp. 1960–1973, an extended access barring scheme for machine-type communications
2013. in lte networks,” Wireless Communications, IEEE Transactions on,
[4] R. Tyagi, F. Aurzada, K.-D. Lee, and M. Reisslein, “Connection estab- vol. PP, no. 99, pp. 1–1, 2015.
lishment in lte-a networks: Justification of poisson process modeling,” [29] S.-Y. Lien, T.-H. Liau, C.-Y. Kao, and K.-C. Chen, “Cooperative ac-
Systems Journal, IEEE, vol. PP, no. 99, pp. 1–12, 2015. cess class barring for machine-to-machine communications,” Wireless
[5] G. T. 36.211, “Physical channels and modulation (release 12),” 3rd Communications, IEEE Transactions on, vol. 11, no. 1, pp. 27–32,
Generation Partnership Project; Technical Specification Group Radio 2012.
Access Network; Evolved Universal Terrestrial Radio Access (E- [30] J.-P. Cheng, C. han Lee, and T.-M. Lin, “Prioritized random access with
UTRA);, 2013. dynamic access barring for RAN overload in 3GPP LTE-a networks,”
[6] G. T. GERAN47, “Common assumptions for MTC simulations on ccch in GLOBECOM Workshops (GC Wkshps), 2011 IEEE, 2011, pp. 368–
and pdch congestion,” 3GPP TSG GERAN 47,Kunming, China, 2010. 372.
[7] 3GPP-TR-37.868, “Study on RAN improvements for machine-type [31] G.-Y. Lin, S.-R. Chang, and H.-Y. Wei, “Estimation and adaptation for
communications (release 11),” Technical Specification Group Radio bursty lte random access,” Vehicular Technology, IEEE Transactions
Access Network, 2011. on, vol. PP, no. 99, pp. 1–1, 2015.
[8] X. Jian, X. Zeng, Y. Jia, L. Zhang, and Y. He, “Beta/m/1 model for [32] 3GPP:TSG-RAN-WG2-71, “MTC simulation results with specific so-
machine type communication,” Communications Letters, IEEE, vol. 17, lutions,” 3GPP TSG RAN WG2 ,Madrid, Spain, 2010.
no. 3, pp. 584–587, March 2013. [33] T. Jiang, X. Tan, X. Luan, X. Zhang, and J. Wu, “Evolutionary game
[9] R. Paiva, R. Vieira, and M. Saily, “Random access capacity evaluation based access class barring for machine-to-machine communications,”
with synchronized MTC users over wireless networks,” in Vehicular in Advanced Communication Technology (ICACT), 2014 16th Interna-
Technology Conference (VTC Spring), 2011 IEEE 73rd, 2011, pp. 1–5. tional Conference on, Feb 2014, pp. 832–835.
[10] R. Paiva, H. Wilhelm, M. Saily, D. Navratil, and M. Taponen, [34] 3GPP:R2-103776, “Separate backoff scheme for MTC,” 3GPP TSG-
“Overload control method for synchronized MTC traffic in geran,” in RAN2 70bis meeting,Stockholm, Sweden, 2010.
Vehicular Tec hnology Conference (VTC Fall), 2011 IEEE, 2011, pp. [35] L. Bello, P. Mitchell, and D. Grace, “Frame based back-off for q-
1–5. learning rach access in lte networks,” in Telecommunication Networks
[11] M. Laner, P. Svoboda, N. Nikaein, and M. Rupp, “Traffic models for and Applications Conference (ATNAC), 2014 Australasian, Nov 2014,
machine type communications,” in Wireless Communication Systems pp. 176–181.
(ISWCS 2013), Proceedings of the Tenth International Symposium on, [36] X. Jian, Y. Jia, X. Zeng, and J. Yang, “A novel class-dependent back-off
Aug 2013, pp. 1–5. scheme for machine type communication in LTE systems,” in Wireless
[12] N. Nikaein, M. Laner, K. Zhou, P. Svoboda, D. Drajic, M. Popovic, and Optical Communication Conference (WOCC), 2013 22nd, 2013,
and S. Krco, “Simple traffic modeling framework for machine type pp. 135–140.
communication,” in Wireless Communication Systems (ISWCS 2013), [37] J. Li, H. Tian, L. Xu, and Y. Huang, “An optimized random access
Proceedings of the Tenth International Symposium on, Aug 2013, pp. algorithm for MTC users over wireless networks,” in Vehicular Tech-
1–5. nology Conference (VTC Spring), 2013 IEEE 77th, 2013, pp. 1–5.
[13] K. Smiljkovic, V. Atanasovski, and L. Gavrilovska, “Machine-to-
[38] G. Wang, X. Zhong, S. Mei, and J. Wang, “An adaptive medium access
machine traffic characterization: Models and case study on integration
control mechanism for cellular based machine to machine (M2M)
in lte,” in Wireless Communications, Vehicular Technology, Information
communication,” in Wireless Information Technology and Systems
Theory and Aerospace Electronic Systems (VITAE), 2014 4th Interna-
(ICWITS), 2010 IEEE International Conference on, 2010, pp. 1–4.
tional Conference on, May 2014, pp. 1–5.
[39] J. Garcia-Luna-Aceves and C. Fullmer, “Performance of floor acquisi-
[14] A. Ahmadian, O. Galinina, S. Andreev, and Y. Koucheryavy, “Modeling
tion multiple access in ad-hoc networks,” in Computers and Commu-
contention-based m2m transmissions over 3gpp lte cellular networks,”
nications, 1998. ISCC ’98. Proceedings. Third IEEE Symposium on,
in Communications Workshops (ICC), 2014 IEEE International Con-
1998, pp. 63–68.
ference on, June 2014, pp. 441–447.
[15] J.-B. Seo and V. Leung, “Design and analysis of backoff algorithms for [40] 3GPP:R2-103759, “Load distribution for MTC devices,” 3GPP TSG-
random access channels in umts-lte and ieee 802.16 systems,” Vehicular RAN WG2 Meeting , Alcatel-Lucent, Alcatel-Lucent Shanghai Bell,
Technology, IEEE Transactions on, vol. 60, no. 8, pp. 3975–3989, Oct 2010.
2011. [41] E. Dahlman, S. Parkvall, and J. Skold, 4G: LTE/LTE-Advanced for Mo-
[16] A. Pourmoghadas and P. Poonacha, “Performance analysis of a bile Broadband: LTE/LTE-Advanced for Mobile Broadband. Elsevier
machine-to-machine friendly mac algorithm in lte-advanced,” in Ad- Science, 2011.
vances in Computing, Communications and Informatics (ICACCI, 2014 [42] 3GPP:R2-105623, “Comparison on RAN loading control schemes for
International Conference on, Sept 2014, pp. 99–105. MTC,” Alcatel-Lucent, Alcatel-Lucent Shanghai Bell, 2010.
[17] 3GPP:TSG-RAN-WG2-70, “Discussion on RACH overload for MTC,” [43] K. Lee, “Preamble set separation for random access control in
3GPP TSG RAN WG2 Meeting 70,Montreal, Canada, 2010. large scale cellular networks,” Apr. 5 2012, wO Patent App.
[18] 3GPP:TSG-RAN-WG2-69, “Discussion on RACH congestion for PCT/KR2011/007,095.
MTC,” 3GPP TSG RAN WG2 69bis,Beijing, China, 2010. [44] 3GPP-TS-36.321, “Evolved universal terrestrial radio access (e-utra);
[19] 3GPP:R2-112247, “Merits of the slotted access methods for MTC,” medium access control (MAC) protocol specification,” 3GPP Technical
Alcatel-lucent Shanghai Bell, Alcatel-lucent, 2011. Specification Group Radio Access Network (Release 12), 2013.
[20] 3GPP:R2-104007, “Pull vs push approach for MTC,” 3GPP TSG RAN [45] K.-D. Lee, M. Reisslein, K. Ryu, and S. Kim, “Handling randomness of
WG2 70bis, Stockholm, Sweden, 2010. multi-class random access loads in LTE-Advanced network supporting
[21] 3GPP:R2-102781, “Paging and downlink transmission for MTC,” small data applications,” in Globecom Workshops (GC Wkshps), 2012
3GPP TSG RAN WG2 Meeting 70, Montreal, Canada, 2010. IEEE, 2012, pp. 436–440.
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
19
[46] K.-D. Lee, S. Kim, and B. Yi, “Throughput comparison of random cations and Networking Conference (WCNC), 2013 IEEE, 2013, pp.
access methods for M2M service over LTE networks,” in GLOBECOM 4706–4711.
Workshops (GC Wkshps), 2011 IEEE, 2011, pp. 373–377. [67] C.-Y. Tu, C.-Y. Ho, and C.-Y. Huang, “Energy-efficient algorithms and
[47] 3GPP:R2-113328, “Dynamic separate RACH resources for MTC,” evaluations for massive access management in cellular based machine
3GPP TSG RAN WG2 74, Institute for Information Industry (III), to machine communications,” in Vehicular Technology Conference
Coiler Corporation, 2011. (VTC Fall), 2011 IEEE, 2011, pp. 1–5.
[48] Y.-C. Pang, S.-L. Chao, G.-Y. Lin, and H.-Y. Wei, “Network access for [68] C.-Y. Ho and C.-Y. Huang, “Energy-saving massive access control and
m2m/h2h hybrid systems: a game theoretic approach,” Communications resource allocation schemes for M2M communications in OFDMA
Letters, IEEE, vol. 18, no. 5, pp. 845–848, 2014. cellular networks,” Wireless Communications Letters, IEEE, vol. 1,
[49] S. Choi, W. Lee, D. Kim, K.-J. Park, S. Choi, and K.-Y. Han, no. 3, pp. 209–212, 2012.
“Automatic configuration of random access channel parameters in LTE [69] K. Zhou, N. Nikaein, R. Knopp, and C. Bonnet, “Contention based
systems,” in Wireless Days (WD), 2011 IFIP, 2011, pp. 1–6. access for machine-type communications over LTE,” in Vehicular
[50] S.-T. Sheu, C.-H. Chiu, Y.-C. Cheng, and K.-H. Kuo, “Self-adaptive Technology Conference (VTC Spring), 2012 IEEE 75th, 2012, pp. 1–5.
persistent contention scheme for scheduling based machine type com- [70] J. Brown and J. Khan, “Predictive resource allocation in the LTE uplink
munications in LTE system,” in Mobile and Wireless Networking for event based M2M applications,” in Communications Workshops
(iCOST), 2012 International Conference on Selected Topics in, 2012, (ICC), 2013 IEEE International Conference on, 2013, pp. 95–100.
pp. 77–82. [71] M. Hasan, E. Hossain, and D. Niyato, “Random access for machine-
[51] M. Giluka, N. Rajoria, A. Kulkarni, V. Sathya, and B. Tamma, to-machine communication in LTE-Advanced networks: issues and
“Class based dynamic priority scheduling for uplink to support m2m approaches,” Communications Magazine, IEEE, vol. 51, no. 6, pp. 86–
communications in lte,” in Internet of Things (WF-IoT), 2014 IEEE 93, 2013.
World Forum on, March 2014, pp. 313–317. [72] G. R. 104873, “Comparing push and pull based approaches for MTC,”
[52] R. R. Tyagi, K.-D. Lee, F. Aurzada, S. Kim, and M. Reisslein, 3rd Generation Partnership Project;, 2010.
“Efficient delivery of frequent small data for u-healthcare applications [73] C.-H. Wei, R.-G. Cheng, and S.-L. Tsao, “Performance analysis of
over lte-advanced networks,” in Proceedings of the 2Nd ACM group paging for machine-type communications in LTE networks,”
International Workshop on Pervasive Wireless Healthcare, ser. Vehicular Technology, IEEE Transactions on, vol. 62, no. 7, pp. 3371–
MobileHealth ’12. New York, NY, USA: ACM, 2012, pp. 27–32. 3382, 2013.
[Online]. Available: http://doi.acm.org/10.1145/2248341.2248354 [74] I. M. Delgado-Luque, F. Blanquez-Casado, F. J. Martin-Vega, M. G.
[53] C. Bontu, J. Ghimire, S. Periyalwar, and M. Pecen, “Asynchronous Fuertes, G. Gomez, M. C. Aguayo-Torres, J. Entrambasaguas, and
simultaneous small packet transmission in cellular wireless system,” J. Banos, “Performance evaluation of cooperation-based techniques
in Wireless Personal Multimedia Communications (WPMC), 2013 16th for M2M traffic over LTE,” in Personal Indoor and Mobile Radio
International Symposium on, 2013, pp. 1–5. Communications (PIMRC), 2013 IEEE 24th International Symposium
[54] A. Lo, Y. W. Law, M. Jacobsson, and M. Kucharzak, “Enhanced lte- on, 2013, pp. 144–148.
advanced random-access mechanism for massive machine-to-machine [75] G. T. 36.808, “Evolved universal terrestrial radio access (e-utra); carrier
(m2m) communications,” in 27th World Wireless Research Forum aggregation; base station (bs) radio transmission and reception,” 3rd
(WWRF) Meeting, 2011, pp. 1–5. Generation Partnership Project; Technical Specification Group Radio
Access Network, 2013.
[55] G. Farhadi and A. Ito, “Group-based signaling and access control for
cellular machine-to-machine communication,” in Vehicular Technology [76] J. Wannstrom, “LTE-Advanced.”
Conference (VTC Fall), 2013 IEEE 78th, 2013, pp. 1–6. [77] 3GPP:TR36.817, “Uplink multiple antenna transmission; base station
(BS) radio transmission and reception (release 10),” 3rd Generation
[56] K. Lee, J. Shin, Y. Cho, K. S. Ko, D. K. Sung, and H. Shin, “A
Partnership Project; Technical Specification Group Radio Access Net-
group-based communication scheme based on the location information
work; Evolved Universal Terrestrial Radio Access (E-UTRA);, 2011.
of MTC devices in cellular networks,” in Communications (ICC), 2012
[78] H. Jafarkhani, Space-time coding: theory and practice. Cambridge
IEEE International Conference on, 2012, pp. 4899–4903.
Univ Pr, 2005.
[57] A.-H. Tsai, L.-C. Wang, J.-H. Huang, and T.-M. Lin, “Overload con-
[79] G. L. Stüber, Principles of mobile communication. Springer Science
trol for machine type communications with femtocells,” in Vehicular
& Business Media, 2011.
Technology Conference (VTC Fall), 2012 IEEE, 2012, pp. 1–5.
[80] G. T. 36.826, “Relay radio transmission and reception (release 11),”
[58] S.-Y. Lien, K.-C. Chen, and Y. Lin, “Toward ubiquitous massive 3rd Generation Partnership Project; Technical Specification Group
accesses in 3GPP machine-to-machine communications,” Communica- Radio Access Network; Evolved Universal Terrestrial Radio Access
tions Magazine, IEEE, vol. 49, no. 4, pp. 66–74, 2011. (E-UTRA);, 2013.
[59] K. Edemacu and T. Bulega, “Resource sharing between m2m and h2h [81] K. Zheng, F. Hu, W. Wang, W. Xiang, and M. Dohler, “Radio resource
traffic under time-controlled scheduling scheme in lte networks,” in allocation in lte-advanced cellular networks with m2m communica-
Telecommunication Systems Services and Applications (TSSA), 2014 tions,” Communications Magazine, IEEE, vol. 50, no. 7, pp. 184–192,
8th International Conference on, Oct 2014, pp. 1–6. July 2012.
[60] S.-H. Wang, H.-J. Su, H.-Y. Hsieh, S. ping Yeh, and M. Ho, “Random [82] D. Kimura and H. Seki, “Inter-cell interference coordination (icic)
access design for clustered wireless machine to machine networks,” in technology,” FUJITSU Sci. Tech. J, vol. 48, no. 1, pp. 89–94, 2012.
Communications and Networking (BlackSeaCom), 2013 First Interna- [83] D. Lee, H. Seo, B. Clerckx, E. Hardouin, D. Mazzarese, S. Nagata,
tional Black Sea Conference on, 2013, pp. 107–111. and K. Sayana, “Coordinated multipoint transmission and reception
[61] D. Niyato, L. Xiao, and P. Wang, “Machine-to-machine communica- in LTE-Advanced: deployment scenarios and operational challenges,”
tions for home energy management system in smart grid,” Communi- Communications Magazine, IEEE, vol. 50, no. 2, pp. 148–155, Febru-
cations Magazine, IEEE, vol. 49, no. 4, pp. 53–59, 2011. ary 2012.
[62] H. kwan Lee, D. M. Kim, Y. Hwang, S. M. Yu, and S.-L. Kim, [84] A. Hamza, S. Khalifa, H. Hamza, and K. Elsayed, “A survey on inter-
“Feasibility of cognitive machine-to-machine communication using cell interference coordination techniques in OFDMA-based cellular
cellular bands,” Wireless Communications, IEEE, vol. 20, no. 2, pp. networks,” Communications Surveys Tutorials, IEEE, vol. 15, no. 4,
97–103, 2013. pp. 1642–1670, Fourth 2013.
[63] F. Azquez-Gallego, J. Alonso-Zarate, I. Balboteo, and L. Alonso, [85] S. Andreev, O. Galinina, and Y. Koucheryavy, “Energy-efficient client
“DPCF-M: A medium access control protocol for dense machine-to- relay scheme for machine-to-machine communication,” in Global
machine area networks with dynamic gateways,” in Signal Processing Telecommunications Conference (GLOBECOM 2011), 2011 IEEE,
Advances in Wireless Communications (SPAWC), 2013 IEEE 14th 2011, pp. 1–5.
Workshop on, 2013, pp. 490–494. [86] S.-T. Sheu, K.-H. Kuo, C.-C. Yang, and Y.-M. Sheu, “A go-back-n harq
[64] T. Kwon and J.-W. Choi, “Multi-group random access resource alloca- time bundling for machine type communication devices in LTE tdd,” in
tion for M2M devices in multicell systems,” Communications Letters, Wireless Communications and Networking Conference (WCNC), 2013
IEEE, vol. 16, no. 6, pp. 834–837, 2012. IEEE, 2013, pp. 280–285.
[65] N. K. Pratas and P. Popovski, “Low-rate machine-type communi- [87] K.-R. Jung, A. Park, and S. Lee, “Machine-Type-Communication
cation via wireless device-to-device (D2D) links,” in arXiv preprint (MTC) Device Grouping Algorithm for Congestion Avoidance of MTC
arXiv:1305.6783, 2013. Oriented LTE Network,” in Security-Enriched Urban Computing and
[66] L. Sun, H. Tian, and L. Xu, “A joint energy-saving mechanism for Smart Grid, Communications in Computer and Information Science,
M2M communications in LTE-based system,” in Wireless Communi- Volume 78. ISBN 978-3-642-16443-9. Springer Berlin Heidelberg,
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
20
2010, p. 167, T.-H. Kim, A. Stoica, and R.-S. Chang, Eds., 2010, p. [107] J. Costa and G. Miao, “Context-aware machine-to-machine com-
167. munications,” in Computer Communications Workshops (INFOCOM
[88] A. Laya, K. Wang, L. Alonso, and J. Alonso-Zarate, “Multi-radio WKSHPS), 2014 IEEE Conference on, April 2014, pp. 730–735.
cooperative retransmission scheme for reliable machine-to-machine [108] P. Makris, D. Skoutas, N. Nomikos, D. Vouyioukas, and C. Skianis,
multicast services,” in Personal Indoor and Mobile Radio Communi- “A context-aware backhaul management solution for combined h2h
cations (PIMRC), 2012 IEEE 23rd International Symposium on, 2012, and m2m traffic,” in Computer, Information and Telecommunication
pp. 1–6. Systems (CITS), 2013 International Conference on, May 2013, pp. 1–
[89] C. Ide, B. Dusza, M. Putzke, C. Muller, and C. Wietfeld, “Influence of 5.
M2M communication on the physical resource utilization of LTE,” in [109] F.-M. Yang, C.-Y. Hsu, and W.-C. Hung, “Group-based control method
Wireless Telecommunications Symposium (WTS), 2012, 2012, pp. 1–6. for machine type communications in LTE system,” in Complex,
[90] C. Ide, B. Dusza, M. Putzke, and C. Wietfeld, “Channel sensitive Intelligent, and Software Intensive Systems (CISIS), 2013 Seventh
transmission scheme for v2i-based floating car data collection via lte,” International Conference on, 2013, pp. 274–280.
in Communications (ICC), 2012 IEEE International Conference on, [110] A. Ksentini, Y. Hadjadj-Aoul, and T. Taleb, “Cellular-based machine-
June 2012, pp. 7151–7156. to-machine: overload control,” Network, IEEE, vol. 26, no. 6, pp. 54–
[91] C. Ide, L. Habel, T. Knaup, M. Schreckenberg, and C. Wietfeld, 60, 2012.
“Interaction between machine-type communication and h2h lte traffic [111] M.-Y. Cheng, G.-Y. Lin, H.-Y. Wei, and A.-C. Hsu, “Overload control
in vehicular environments,” in Vehicular Technology Conference (VTC for machine-type-communications in LTE-Advanced system,” Commu-
Spring), 2014 IEEE 79th, May 2014, pp. 1–5. nications Magazine, IEEE, vol. 50, no. 6, pp. 38–45, 2012.
[92] A. Lioumpas and A. Alexiou, “Uplink scheduling for machine-to- [112] 3GPP:R2-110465, “Rrc message to carry the extended wait time,”
machine communications in LTE-based cellular systems,” in GLOBE- 3GPP TSG-RAN WG2 72bis, Dublin, Ireland, 2011.
COM Workshops (GC Wkshps), 2011 IEEE, 2011, pp. 353–357. [113] 3GPP:R2-110403, “Further consideration on extended wait time as-
signment,” 3GPP TSG-RAN WG2 72bis, Dublin, Ireland, 2011.
[93] S. Choi, K. Jun, Y. Shin, S. Kang, and B. Choi, “MAC scheduling
[114] 3GPP:R2-110137, “Extended wait timer,” 3GPP TSG-RAN WG2 72bis,
scheme for VoIP traffic service in 3G LTE,” in Vehicular Technology
Dublin, Ireland, 2011.
Conference, 2007. VTC-2007 Fall. 2007 IEEE 66th, 2007, pp. 1441–
[115] 3GPP:R2-110158, “Extended wait timers,” 3GPP TSG-RAN WG2
1445.
72bis, Dublin, Ireland, 2011.
[94] M. Andreozzi, G. Steay, A. Bacioccola, and R. Rossi, “Flexible [116] H. Yao, T. Hui, Z. Jie, and Q. Cheng, “Rate allocation scheme for
scheduling for real-time services in high-speed packet access cellular machine-to-machine service based on 3GPP in heterogeneous wireless
networks,” in Wireless Conference, 2009. EW 2009. European, 2009, networks,” Communications, China, vol. 10, no. 9, pp. 65–71, 2013.
pp. 189–193. [117] G. T. 23.401, “LTE; general packet radio service (GPRS) enhancements
[95] A. K. F. Khattab and K. M. F. Elsayed, “Channel-quality dependent for evolved universal terrestrial radio access network (E-UTRAN) ac-
earliest deadline due fair scheduling schemes for wireless multimedia cess (version 11.5.0 release 11),” 3rd Generation Partnership Project;,
networks,” in Proceedings of the 7th ACM International Symposium on 2013.
Modeling, Analysis and Simulation of Wireless and Mobile Systems, ser.
MSWiM ’04. New York, NY, USA: ACM, 2004, pp. 31–38.
[96] G. Barriac and J. Holtzman, “Introducing delay sensitivity into the
proportional fair algorithm for cdma downlink scheduling,” in Spread
Spectrum Techniques and Applications, 2002 IEEE Seventh Interna-
tional Symposium on, vol. 3, 2002, pp. 652–656 vol.3.
[97] I. M. Delgado-Luque, F. Blanquez-Casado, M. Fuertes, G. Gomez,
M. C. Aguayo-Torres, J. Entrambasaguas, and J. Banos, “Evaluation
of latency-aware scheduling techniques for M2M traffic over LTE,” in
Signal Processing Conference (EUSIPCO), 2012 Proceedings of the
20th European, 2012, pp. 989–993.
[98] S. Zhenqi, Y. Haifeng, C. Xuefen, and L. Hongxia, “Research on uplink Erfan Soltanmohammadi (S12, M15) was born in
scheduling algorithm of massive M2M and H2H services in LTE,” in Karaj, Iran, in 1984. He received his BSc in electri-
Information and Communications Technologies (IETICT 2013), IET cal engineering from K.N. Toosi University of Tech-
International Conference on, 2013, pp. 365–369. nology (KNTU), Tehran, Iran, in 2007, MSc from
[99] T. Kitahara, Y. Hiehata, H. Koto, N. Fukumoto, H. Nakamura, and Amirkabir University of Technology (AUT), Tehran,
S. Ano, “Autonomous data transmission using off-peak hours for Iran, in 2010, and PhD degree in systems (commu-
cellular-based M2M communications,” in Computer Software and nication & signal processing) from Louisiana State
Applications Conference (COMPSAC), 2013 IEEE 37th Annual, 2013, University (LSU), Baton Rouge, Louisiana, USA, in
pp. 167–168. 2014. He is currently a DSP engineer at Marvell
Semiconductor, Inc., Santa Clara, California, USA.
[100] I. Abdalla and S. Venkatesan, “A qoe preserving M2M-aware hy-
He received George Reymond scholarship from LSU
brid scheduler for LTE uplink,” in Mobile and Wireless Networking
in 2013, and was awarded as one of the top 100 engineers at Marvell
(MoWNeT), 2013 International Conference on Selected Topics in, 2013,
Semiconductor, Inc. in 2015. His research interests include signal processing,
pp. 127–132.
machine learning, and wireless communications.
[101] S. Y. Shin and D. Triwicaksono, “Radio resource control scheme for
machine-to-machine communication in LTE infrastructure,” in ICT
Convergence (ICTC), 2012 International Conference on, 2012, pp. 1–6.
[102] V. Vakilian, T. Wild, F. Schaich, S. ten Brink, and J.-F. Frigon,
“Universal-filtered multi-carrier technique for wireless systems beyond
lte,” in Globecom Workshops (GC Wkshps), 2013 IEEE, Dec 2013, pp.
223–228.
[103] F. Schaich and T. Wild, “Waveform contenders for 5g 2014; ofdm vs.
fbmc vs. ufmc,” in Communications, Control and Signal Processing
(ISCCSP), 2014 6th International Symposium on, May 2014, pp. 457–
460.
[104] B. Farhang-Boroujeny, “Ofdm versus filter bank multicarrier,” Signal
Processing Magazine, IEEE, vol. 28, no. 3, pp. 92–112, May 2011. Kamran Ghavami received his B.Sc. and M.S. in
electrical engineering from the University of Tehran
[105] Y. Saito, Y. Kishiyama, A. Benjebbour, T. Nakamura, A. Li, and
in 1998 and 2000 respectively. Upon the completion
K. Higuchi, “Non-orthogonal multiple access (noma) for cellular future
of his M.Sc. degree, he conducted research and
radio access,” in Vehicular Technology Conference (VTC Spring), 2013
implementation of telecommunication systems as a
IEEE 77th, June 2013, pp. 1–5.
research engineer at NRI, where he participated in
[106] H. Nikopour and H. Baligh, “Sparse code multiple access,” in Personal several related projects. In January 2013, he started
Indoor and Mobile Radio Communications (PIMRC), 2013 IEEE 24th his Ph.D. program at Louisiana State University. His
International Symposium on, Sept 2013, pp. 332–336. research interests include channel coding, iterative
detection, and MIMO wireless systems.
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2533541, IEEE Internet of
Things Journal
21
2327-4662 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.