Medium Access Control with Scheduled Group Contention for Massive M2M Devices

David C. Wyld et al. (Eds) : NETCOM, NCS, WiMoNe, CSEIT, SPM - 2015
pp. 39–49, 2015. © CS & IT-CSCP 2015 DOI : 10.5121/csit.2015.51604
MEDIUM ACCESS CONTROL WITH
SCHEDULED GROUP CONTENTION FOR
MASSIVE M2M DEVICES
Joo Rak Kang1
and Tae-Jin Lee2
1
Samsung Electro-Mechanics Co., Ltd., Suwon, Korea
College of Information and Communication Engineering,
Sungkyunkwan University, Suwon, Korea
jrkang@skku.edu
2
College of Information and Communication Engineering,
Sungkyunkwan University, Suwon, Korea
tjlee@skku.edu
ABSTRACT
In a dense Machine-to-Machine (M2M) network, a large number of stations contend to achieve
transmission opportunity and it creates a critical congestion problem. To solve this issue, a
group-based contention Medium Access Control (MAC) protocol is introduced. Stations are
divided into small numbers of groups and only one station in each group will contend as a
group leader to achieve the reserved time interval dedicated to a winner group. It can reduce
the contention overhead and lessen the congestion problem. In this paper, we propose
Scheduled Group Contention MAC (SGCMAC) protocol to enhance the group-based contention
MAC. The proposed SGCMAC protocol divides groups based on the traffic categories of
stations and schedules the contention groups to reduce the contention overhead. We also
propose an efficient resource management mechanism in the group grant time to prevent the
waste of time caused by idle stations. Simulations with IEEE 802.11ah parameters demonstrate
that our proposed SGCMAC has performance gains over other group-based contention MAC
protocols.
KEYWORDS
Medium Access Control (MAC), Machine-to-Machine, Grouping, Wireless LAN, Random
Access Protocols
1. INTRODUCTION
Internet of Things (IoT) is a new paradigm for the future of the Internet. IoT will integrate a
variety of sensors, actuators, smart devices and things through the Internet, and distributed smart
services will be created to change our daily life. Machine-to-Machine (M2M) communications is
considered as a basic communication technology for the realization of IoT. M2M
communications involves information exchange among machines without any human interaction.
M2M networks are expected to be widely utilized in many fields of various IoT applications such

40 Computer Science & Information Technology (CS & IT)
as public services, industrial automations, health care, intelligent transport systems, smart grids,
and agricultural networks [1]-[5].
M2M networks have unique characteristics which are quite different from conventional networks.
In M2M networks, tremendous number of devices may be involved in the service coverage and
concurrent network access attempts may occur from the devices. Typically, the amount of data
generated from each device may be small and data may be generated infrequently. If, however,
many devices participate in and generate data at the same time, it may cause a critical congestion
problem. Also, in M2M networks, high level of system automation, in which devices and systems
exchange and share data automatically, is required. Therefore, congestion control is a
fundamental issue in M2M networks. Especially, from the Medium Access Control (MAC) layer
perspective, efficient access control and management of network resources is one of the most
challenging issues in M2M networks [2]. In addition to congestion control and scalability, an
M2M network has to support various traffic categories and quality of service (QoS) requirements.
In this context, it is important how to distribute traffic and network access attempts subject to
QoS requirements, which is the motivation to our work.
To address the issue, group-based access control schemes have been introduced in the literature
[6]-[9]. The authors in [6] introduce a grouping-based radio resource management for 3GPP
M2M communications. In [7], IEEE 802.11ah task group proposes Restricted Access Window
(RAW) to allow contending channel only for a small number of stations. The group-based
contention MAC protocols, a hybrid scheme to combine the advantages of contention-based and
contention-free protocols, are presented, Group-based Medium Access Control (GMAC) [8] and
Group Leader DCF-TDMA (DCFT) [9]. In GMAC and DCFT, all stations are divided into small
numbers of groups and only one station in each group, called a group leader, involves in channel
contention. If a group leader wins the competition, a certain time duration is reserved for its
associated group and resources are allocated to the group members according to a contention-free
protocol.
In this paper, we propose an enhanced group-based contention MAC protocol, Scheduled Group
Contention MAC (SGCMAC) protocol, which is more suitable for M2M networks with
heterogeneous M2M traffics. Our SGCMAC divides the machine type communications (MTC)
stations into small numbers of groups according to their traffic categories, e.g., applications,
trigger event types and periods of traffic. The proposed group-based resource allocation scheme
is very efficient since the temporal correlation among the group members of a group is typically
very high. The Access Point (AP) can schedule the contention time duration of each group based
on its traffic characteristics. It can spread the contention groups over a wide range of time
interval. Therefore, our proposed SGCMAC is not only efficient in reducing the contention
overhead in massive M2M networks but also in being applicable to M2M networks with
heterogeneous traffics. We develop an efficient mechanism to detect active stations and idle
stations in the reserved time duration for a winner group, which removes the waste time caused
by idle stations during the reserved time. The rest of this paper is organized as follows. In Section
II and Section III, we describe the system model and the proposed SGCMAC protocol. Section
IV presents simulation results to evaluate performance of the proposed protocol. Finally, Section
V concludes the paper.
2. SYSTEM MODEL
We consider a single-hop wireless network consisting of one AP and ݊ MTC stations. The
network works based on the IEEE 802.11ah MAC and Physical (PHY) layer . Each device

Computer Science & Information Technology (CS & IT) 41
follows the Distributed Coordinate Function (DCF) to access the medium and operates in the sub-
1GHz frequency band as described in [7], [10]. The MTC stations in a network have two types of
operation status, i.e., active status and idle status. When a device is in the active status, it has
upload data. We assume a single data buffer in an M2M device. If a active station successfully
transmits data, this station goes to idle status. The events are assumed to arrive at each station by
a Poisson process with the average arrival rate ߣ. Fig. 1 shows an example of DCF operation that
active stations content the channel access according to DCF rules.
Fig. 1 An example of IEEE 802.11 Distributed Coordinate Function (DCF) operation.
3. PROPOSED SGCMAC PROTOCOL
In this section we describe our proposed SGCMAC protocol. The related grouping rule and the
group leader selection algorithm are also introduced. To address the contention problem in a
dense M2M network, we divide all MTC devices into several groups according to their traffic
categories. The AP assigns the group leader role to a member in each group and schedules time
intervals for some of the group leaders to coordinate the channel access. We call this specific
time interval Group Access Window (GAW). The scheduled group leaders contend in a GAW by
a modified DCF scheme. When a group leader wins the contention, the AP assigns time slots to
the group members to allow data transmission by the group period granting procedure. The
procedure for the SGCMAC protocol is shown in Fig. 2.
3.1. Group Access Window
The AP allocates a particular time period called GAW in a beacon interval for the group-based
channel access. The information of a GAW such as the start time and the duration of the GAW
and the scheduled groups for the GAW is announced through the Scheduled Group Indication
Map (SGIM) Information Element in a beacon. The group leaders in the SGIM start contention to
acquire a Group Granted Period (GGP) at the GAW start time.
The contention procedure among the group leaders is similar to the Request to Send (RTS)/Clear
to Send (CTS) procedure based on the DCF scheme. After a random backoff, a group leader
sends a Group Access Request (GAR) frame to the AP. If there is no collision, the AP responds
by the Group Access Grant (GAG) frame. Since all the active devices in a network as well as the
contending group leaders can hear the GAG frame, the winner group is allowed the GGP, defined
in the duration field of the GAG frame. The AP checks the active group members in the winner
group and assigns time slots to the active stations by Time Division Multiple Access (TDMA)

scheme. If the total duration of the assigned time slots is less than GGP in the duration field of
the GAG frame, the AP broadcasts the Group Access End (GAE) frame after the data exchange
of the last active group member to terminate GGP. After the GGP, the group leaders resume
contention again to access the channel until the end of GAW.
Fig. 2. Group Access Window (GAW) and Scheduled Group Contention MAC (SGCMAC) protocol.
The GAW has advantages. Because an AP can schedule a contention group at a particular GAW,
M2M traffic and contention will be spread across beacon periods and QoS requirements of each
group are able to be supported. Moreover, M2M devices can coexist with legacy devices since
the mechanism of channel reservation for a group allows the protection mechanism.
(a) Transmission of stations in a GGP

(b) Frame exchange sequence during GGP
Fig. 3. An example of Group Granted Period (GGP).
3.2. Transmission of Stations in a GGP
When a group leader wins the contention, the winner group gets the dedicated channel access
time called GGP. The transmission time for each station in a winner group is assigned by an AP
using a TDMA scheme. However, if a station is not activated in GGP, the time slot for this
station will be wasted. In order to prevent this waste, the AP has to check whether each station is
in the active status or not before assigning time slots. This status check procedure may be an
overhead. To minimize this overhead, we use a signal multiplexing scheme in the frequency
domain as in [13]. When the GGP starts, the AP transmits the Status Request (SReq) frame to the
group members. The SReq frame includes the sub-carrier assignment map for each group
member. Active group members can hear the SReq frame and knows the sub-carrier assigned to
itself. After Short Inter-Frame Space (SIFS), the active group members respond by sending Status
Response (SRes) by using single tone signals assigned by the AP. Based on this SRes, the AP
will be able to assign time slots to the active group members without wasting time. Since we
assume that the group members have fixed data size, the AP can schedule the time slots for each
of the active members by announcing the order of the active members. Then the AP transmits a
Resource Allocation (RA) frame including the schedule information.
In our protocol, a time slot is assigned for a particular station. So, very small time interval larger
than the propagation delay is enough to separate data exchange sequences of each active member.
In this paper, this inter-time slot of a member is designed and utilized as in Reduced Inter-Frame
Space (RIFS) defined in the IEEE 802.11 standard. Fig. 3 shows an example of GGP. The total
channel occupation time for one group access consists of the time duration to check the active
stations and the scheduled time slots for data transmission. Let ܶ஼ and ܶௌ denote the time
durations to confirm the active stations and the scheduled time slots, respectively.
ܶ஼ = ܶௌோாொ + ܶௌோாௌ + ܶோ஺ + ܵ‫ܵܨܫ‬ × 4 (1)
where ܶௌோாொ, ܶௌோாௌ, and ܶோ஺ denote the duration for the SReq frame, the SRes, and the RA
frame, respectively.

ܶௌ = (ܶ஽ + ܵ‫ܵܨܫ‬ + ܶ஺ + ܴ‫)ܵܨܫ‬ × ܰ஺ீெ
+(ܵ‫ܵܨܫ‬ − ܴ‫ܵܨܫ‬ + ܶா) (2)
where ܶ஽, ܶ஺ and ܶா denote the duration for data, acknowledgment (ACK), and GAE frame,
respectively. ܰ஺ீெ denotes the number of active stations in a specific GGP. The total channel
occupation time ܶீ௥௔௡௧ for one group access is then calculated as follows.
ܶீ௥௔௡௧ = ܶ஼ + ܶௌ. (3)
3.3. Grouping and Group Leader Selection Algorithm
We divide MTC stations into groups according to the traffic patterns of stations while GMAC
and DCFT make groups based on the region or the coverage of nodes. After grouping, there is
only a small number of stations involved in contention at a certain time duration and the hidden
terminal issue may be reduced by the RTS/CTS mechanism as defined in [10]. If the temporal
correlation between group member stations is low, it will adversely affect the channel efficiency
of the group-based contention MAC.
A traffic pattern of each MTC station depends on its application or service type. The traffic of an
M2M application or service is classified into Fixed-Scheduling (FS) or Event-Driven (ED) [11].
Furthermore, it can be categorized into periodic, trigger event, or random based on its application
[6], [12]. When an MTC station associates with the AP, a station informs its traffic category and
application type to the AP. The AP assigns this station to a proper group using the Group ID
(GID) and the Group Member ID (GMID). In the proposed protocol, we limit the maximum
number of group members to a number less than the number of sub-carriers for efficient group
management. Accordingly, it can create multiple groups with the same traffic category.
One of the group members is selected as a group leader by the AP. The group leader consumes
more power than the other group members due to the contention for its GGP. To prevent too
early burning out of a group leader station, the AP may change the group leader every GGP. The
AP chooses the next group leader among the active group members in a round-robin manner and
sends its information in the RA frame.
4. SIMULATION RESULTS
In this section, we present the simulation results of the proposed SGCMAC protocol, GMAC [8]
and DCFT [9]. We develop our own simulator which have been performed with the PHY and
MAC layer characteristics based on the IEEE 802.11ah standard [7] as in Table 1. We assume
that the number of stations in each group is fixed to 50. We measure the temporary reservation
time for a winner group succeeded in the group contention and the length of group management
frames such as the Polling frame in GMAC and the SReq frame and the RA frame in SGCMAC.
The data rate of stations in the simulations is 650Kbps defined as the basic data rate in IEEE
802.11ah.

Table 1. Simulation Parameters
Parameter Value Description
aSlotTime 52 ߤ‫ݏ‬ Backoff slot time
aRIFSTime 20 ߤ‫ݏ‬ Reduced Inter-Frame Space
aSIFSTime 160 ߤ‫ݏ‬ Short Inter-Frame Space
aDIFSTime 264 ߤ‫ݏ‬ DCF Inter-Frame Space
‫ܹܥ‬௠௜௡ 16 Minimum contention window size
‫ܹܥ‬௠௔௫ 512 Maximum contention window size
ܰீெ 50 Number of group members
Data Rate 650 Kbps S1G MCS0 for 2MHz channel
Payload 512 Bytes Data payload size
MAC Overhead 22 Bytes S1G short MAC header+FCS
ܶ௉ு௒ 280 ߤ‫ݏ‬ PHY header
ܶோ்ௌ 560 ߤ‫ݏ‬ RTS, DCFT poll, GAR frames
ܶ஼்ௌ 480 ߤ‫ݏ‬ CTS, CFEND, GAG, GAE frames
ܶோ஺ 1120 ߤ‫ݏ‬ SReq, RA, GMAC polling frames
ܶௌோாௌ 40 ߤ‫ݏ‬ SRes frame
ܶ஺஼௄ 240 ߤ‫ݏ‬ ACK frame
4.1. Performance in Homogeneous Traffic
Fig. 4 and Fig. 5 show the performance under the homogeneous traffic condition. The number of
stations associated with the AP is 4000 and they are divided into 80 groups. All stations belong to
one traffic category with the same event arrival rate. Our proposed SGCMAC shows gains in
throughput and delay compared with the other group-based contention MAC schemes.
Fig. 4. Throughput for varying arrival rates under the homogeneous traffic condition.

Fig. 5. Average delay for varying arrival rates under the homogeneous traffic condition.
Fig. 4 shows the throughput for varying event arrival rates from 0.1 to 2.0. As the arrival rate
increases above 1.0, the network traffic reaches the saturation condition. Then almost all stations
are in the active status whenever their groups achieve the reserved time slots for the group
members. In this case, the contention overhead and the static overhead of each MAC protocol
mainly affects the throughput. If the arrival rate decreases, stations stay in the idle status for a
long time, and a large portion of the reserved time interval for a winner group will be wasted. In
order to reduce this waste, it is important to efficiently detect idle stations in the reserved time
interval for a winner group. Our proposed SGCMAC exhibits better performance than the others
in terms of channel efficiency and throughput.
The average delay of stations is presented in Fig. 5. Our proposed SGCMAC has a gain of 850ms
compared with the others at the arrival rate of 0.1. The proposed SGCMAC protocol has lower
contention overhead and less waste of time in the granted group duration than the others. When
the network traffic reaches its saturation condition, the average contention overhead and the static
MAC and PHY overhead only affects the delay. In this case, our proposed SGCMAC also shows
lower delay than the others.
4.2. Performance in Heterogeneous Traffic
In Fig. 6 and Fig. 7, the throughput and the delay are presented for varying numbers of stations
under the heterogeneous traffic condition. There are 20 traffic categories which are classified
according to their arrival rates from 0.01 to 0.2. All stations are grouped into 20 traffic categories
such that equal number of stations is assigned to each of the categories.

Fig. 6. Throughput for varying numbers of stations under the heterogeneous traffic condition.
Fig. 7. Average delay for varying number of stations under heterogeneous traffic condition.
The throughput for varying number of stations is shown in Fig. 6. The throughput of our
proposed SGCMAC is higher than those of GMAC and DCFT. The average gain of the
throughput is 20 Kbps and 113 Kbps compared with those of GMAC and DCFT, respectively. If
the number of groups increases due to varying number of stations, the contention overhead in
DCF also increases, which then decreases throughput. In the simulation, the arrival rate of each
group is lower than 0.2. So most stations stay in the idle status for a long time. However, if the
channel access interval of each group gets longer due to increasing number of contention groups,
the member stations in the active status will increase in the reserved time duration for a winner

group. Then the network throughput increases as the number of stations increases as shown in
Fig. 6.
In this traffic condition, the delay is a more important performance metric. The average delay of
stations is presented in Fig. 7. The delay of SGCMAC is better than those of the other group-
based contention MAC protocols. The delay of SGCMAC is 600 ms to 2.2 seconds lower than
those of the others. In our proposed SGCMAC, the contention overhead does not change
regardless of the number of stations since the AP is able to schedule a certain number of
contention groups in GAW. In the simulation, the number of contention groups in every GAW is
controlled to 10. Since the proposed SGCMAC keeps the low contention overhead and minimizes
the waste time in the reserved time duration for a winner group, it shows lower delay over the
whole range of the number of stations than the others.
5. CONCLUSION
In this paper, we have proposed a group-based contention MAC protocol for a dense M2M
network. Our proposed SGCMAC divides groups based on the traffic categories of stations and
schedules group contention time slots. It can reduce the contention overhead for channel access
and solve a congestion problem in a dense M2M network. In addition, we propose an efficient
mechanism to determine active stations in the reserved time duration for a winner group and
resources are allocated to reside within the actual transmission time. We can reduce the waste
time caused by idle stations in the group access time. Through simulations with the IEEE
802.11ah MAC and PHY parameters, we evaluate the throughput and delay performance of our
proposed SGCMAC compared to GMAC and DCFT. The simulation results demonstrate that our
proposed SGCMAC has performance gains compared with the other group-based contention
MAC protocols. As a future work, we would like to evaluate our SGCMAC protocol
performance based on real deployed M2M network traffic model.
ACKNOWLEDGMENT
This work was supported by the National Research Foundation of Korea (NRF) grant funded by
the Korean government (MSIP) (2014R1A5A1011478).
REFERENCES
[1] L. Atzori, A. Iera, and G. Morabito, “The Internet of Things: A Survey,” Computer Networks, vol.
54, no. 15, pp. 2787-2805, Oct. 2010.
[2] A. Rajandekar and B. Sikdar, “A Survey of MAC Layer Issues and Protocols for Machine-to-
Machine Communications,” IEEE Internet of Things Journal, vol. 2, no. 2, pp. 175-186, Apr. 2015.
[3] J. Kim, J. Lee, J. Kim, and J. Yun, “M2M Service Platforms: Survey, Issues, and Enabling
Technologies,” IEEE Communications Surveys & Tutorials, vol. 16, no. 1, Feb. 2014.
[4] L. Karim, A. Anpalagan, N. Nasser, and J. Almhana, "Sensor-based M2M Agriculture Monitoring
Systems for Developing Countries: State and Challenges," Network Protocols and Algorithms, vol. 5,
no. 3, pp. 68-86, 2013.

[5] A. Reche, S. Sendra, J. R. Diza, and J. Lloret, "A Smart M2M Deployment to Control the Agriculture
Irrigation," Ad-hoc Networks and Wireless of the series Lecture Notes in Computer Science, vol.
8629, pp. 139-151, Feb. 2015.
[6] S.-Y. Lien, K.-C. Chen, and Y. Lin, “Toward Ubiquitous Massive Accesses in 3GPP Machine-to-
Machine Communications,” IEEE Communication Magazine, vol. 49, no. 4, pp. 66-74, Apr. 2011.
[7] IEEE P802.11ah Draft Ver.5.0, IEEE Std. 802.11 TGah, Mar. 2015.
[8] Z. Abichar and J. M. Chang, “Group-Based Medium Access Control for IEEE 802.11n Wireless
LANs,” IEEE Trans. Mobile Computing, vol. 12, no. 2, pp. 304-317, Feb. 2013.
[9] Y. Yang and S. Roy, “Grouping-Based MAC Protocols for EV Charging Data Transmission in Smart
Metering Network,” IEEE J. Selected Areas in Communications, vol. 32, no. 7, pp. 1328-1343, Jul.
2014.
[10] IEEE Std 802.11-2012, Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer
(PHY) Specifications, IEEE, Mar. 2012.
[11] O. Al-Khatib, W. Hardjawana, and B. Vucetic, “Traffic Modeling for Machine-to-Machine (M2M)
Last Mile Wireless Access Networks,” in Proc. of IEEE Global Communications Conference
(GLOBECOM), pp. 1199-1204, Dec. 2014.
[12] R. Liu, W. Wu, H. Zhu, and D. Yang, “M2M-Oriented QoS Categorization in Cellular Network,” in
Proc. of 7th International Conference on Wireless Communications, Networking and Mobile
Computing (WiCOM), Sep. 2011.
[13] S. Sen, R. Roy Choudhury, and S. Nelakuditi, “No Time to Countdown: Migrating Backoff to the
Frequency Domain,” in Proc. of 17th Annual International Conference on Mobile Computing and
Networking (MobiCom), pp. 241-252, Sep. 2011.

Medium Access Control with Scheduled Group Contention for Massive M2M Devices

More Related Content

What's hot (16)

Similar to Medium Access Control with Scheduled Group Contention for Massive M2M Devices (20)

Recently uploaded (20)

Medium Access Control with Scheduled Group Contention for Massive M2M Devices