Content uploaded by Prasan Kumar Sahoo
Author content
All content in this area was uploaded by Prasan Kumar Sahoo on Jul 26, 2017
Content may be subject to copyright.
sensors
Article
A Novel IEEE 802.15.4e DSME MAC for Wireless
Sensor Networks
Prasan Kumar Sahoo 1,3, Sudhir Ranjan Pattanaik 2and Shih-Lin Wu 1,3,4,*
1Department of Computer Science and Information Engineering, Chang Gung University,
Taoyuan 33302, Taiwan; pksahoo@mail.cgu.edu.tw
2Department of Electrical Engineering, Chang Gung University, Taoyuan 33302, Taiwan;
d0121009@stmail.cgu.edu.tw
3Department of Cardiology, Chang Gung Memorial Hospital, Taoyuan 33305, Taiwan
4Department of Electrical Engineering, Ming Chi University of Technology, New Taipei City 24301, Taiwan
*Correspondence: slwu@mail.cgu.edu.tw; Tel.: +886-3-211-8800 (ext. 5320)
Academic Editor: Mohamed F. Younis
Received: 22 November 2016; Accepted: 9 January 2017; Published: 16 January 2017
Abstract:
IEEE 802.15.4e standard proposes Deterministic and Synchronous Multichannel Extension
(DSME) mode for wireless sensor networks (WSNs) to support industrial, commercial and health
care applications. In this paper, a new channel access scheme and beacon scheduling schemes are
designed for the IEEE 802.15.4e enabled WSNs in star topology to reduce the network discovery
time and energy consumption. In addition, a new dynamic guaranteed retransmission slot allocation
scheme is designed for devices with the failure Guaranteed Time Slot (GTS) transmission to reduce
the retransmission delay. To evaluate our schemes, analytical models are designed to analyze the
performance of WSNs in terms of reliability, delay, throughput and energy consumption. Our schemes
are validated with simulation and analytical results and are observed that simulation results well
match with the analytical one. The evaluated results of our designed schemes can improve the
reliability, throughput, delay, and energy consumptions significantly.
Keywords: wireless sensor networks; IEEE 802.15.4e; performance analysis; DSME; star topology
1. Introduction
In Wireless Sensor Network (WSNs), devices are spatially distributed to monitor the physical
or environmental conditions such as temperature, sound, pressure etc. Nowadays, WSNs are
used in many industrial and consumer applications to monitor and control the industrial process,
machine health, air pollution and many others. ZigBee devices are suitable to play major roles in
different applications of Internet of Things (IoT). Large numbers of devices are deployed for different
applications in IoT. Hence, many devices compete with each other to acquire the channel for data
communications. IEEE 802.15.4 [
1
] is a popular standard for the medium access control (MAC) scheme
of low-powered, low data rate wireless sensors. However, in order to meet the critical requirements
such as low latency in industrial and commercial applications, IEEE 802.15.4e Working Group has
redesigned the existing 802.15.4 MAC scheme.
IEEE 802.15.4e [
2
] standard defines the slotted medium access control scheme for the short-range
communication devices with low data rates. There are many MAC schemes for different applications
defined by IEEE 802.15.4e such as Time Slotted Channel Hopping (TSCH), Low Latency Deterministic
Networks (LLDN), Deterministic and Synchronous Multi channel Extension (DSME), Radio Frequency
Identification blink (RFID), and Asynchronous Multi-Channel Adaptation (AMCA). Out of these,
DSME mode proposed in IEEE 802.15.4e standard [
2
] is designed for the habitat monitoring, process
automation, factory automation, smart metering, home automation, smart building, entertainment,
Sensors 2017,17, 168; doi:10.3390/s17010168 www.mdpi.com/journal/sensors
Sensors 2017,17, 168 2 of 25
patient monitoring and telemedicine applications. The MAC scheme of DSME superframe structure
extends the number of GTS slots and the single channel operation of IEEE 802.15.4 standard to
multichannel operation for data transmission. DSME superframe structure supports the star topology
based WSNs, where collisions occur either due to hidden terminal problem or simultaneous channel
assessment for data transmission.
Hidden terminal problem can be avoided by allocating a common Contention Access Period
(CAP) in the superframe to the devices present within the same communication range. CSMA/CA
scheme avoids the collision during the transmission of other devices, which are not hidden. In WSN,
large numbers of devices are deployed for different applications in which many devices attempt
to send frames simultaneously resulting significant collisions in the system. The probability of
collisions is related directly to the number of devices in the network and this issue is significant, when
large number of devices are associated with one coordinator. To fulfill the present need of different
applications, the coordinator needs to support more devices, which may result higher rate of collisions.
The performance of CAP is deteriorated when large numbers of devices contend with each other.
In IEEE 802.15.4e MAC, nodes perform slotted carrier sense multiple access with collision
avoidance (CSMA/CA) mechanism to access the channel and uses a random backoff algorithm
to reduce the collision probability. To save power consumption, the devices use power saving mode
during the backoff duration and perform carrier sensing after the backoff duration is over. The channel
access scheme suggested by IEEE 802.15.4e standard simply adopts a random backoff with clear channel
assessment (CCA). The channel busy occurs either due to data or acknowledgement transmission.
However, the current scheme proposed in IEEE 802.15.4e increases delay as well as energy consumption
and reduces the throughput of the network due to inefficient backoff and carrier sensing management.
Hence, this problem needs to be addressed to reduce unnecessary backoff and carrier sensing.
The DSME MAC supports a group acknowledgement option to provide a retransmission
opportunity within the same superframe for any data frame, which is failed in its
DSME-GTS
transmission. This
group acknowledgement option improves the network efficiency as
acknowledgement (ACK) for multiple data frames are aggregated to a single ACK frame. However,
the devices allocate an additional DSME GTS for Retransmission (GTSR) per each allocated DSME
GTS to transmit data to the coordinator. These slots are used for retransmissions of their failed
GTS transmissions or for transmitting additional data frames. If a device is succeeded in the first
attempt of the GTS and has no new data to transmit in the allocated DSME-GTSR, the allocated slot
becomes useless. This problem needs to be addressed so that those DSME-GTSR slots can be reused
for other devices.
Mobility of devices are imminent in many IoT applications such as habitat monitoring, home
automation, smart building, entertainment and patient monitoring etc. Due to mobility, devices
frequently leave or enter into the network, by which their synchronization is lost. In order to join a
beacon enabled network, devices first discover the coordinator by scanning channels and then join
the network by using association messages. The channel scanning durations are defined based on the
Beacon Order (BO) of the coordinator. The longer is the BO, the higher is the latency of the coordinator
discovery. Again, neither the channel on which the coordinator operates nor its frequency of beaconing
is known to a new device that enters into the network. Therefore, minimization of association time for
a device with the coordinator is a challenging task, which needs to be addressed.
Motivation
In a WSN, normally nodes are deployed densely around the coordinator and several nodes may
compete with each other to transmit data simultaneously. Though, IEEE 802.15.4e has its own collision
avoidance mechanism, there is no provision to reduce the probability of collisions by restricting the
number of devices to be associated with the coordinator at the same time. Hence, it is essential to
restrict large numbers of devices to be associated with the coordinator or none of devices can get
benefits to transmit data. In IEEE 802.15.4e medium access mechanism, a node has to go for clear
Sensors 2017,17, 168 3 of 25
channel assessment (CCA) after its random backoff period is over. It senses the channel busy if any
other node within its transmission range is transmitting either data or ACK and therefore it increases
the delay and consume more power due to another longer backoff. It is necessary to devise some
mechanism by which unnecessary backoffs can be avoided. Hence, we design a new carrier sensing
mechanism by which a node can avoid sensing the channel busy situation when the exchange of ACK
is going on by other node.
The group ACK option given in DSME MAC of IEEE 802.15.4e provides a retransmission
opportunity to the nodes within the same superframe. Accordingly, an additional DSME GTS is
allocated for Retransmission (GTSR) to the coordinator. However, these allocated slots become useless,
if a device is succeeded to transmit data in its first attempt and has no new data to transmit. Therefore,
we design here one dynamic GTS allocation scheme by which the retransmission delay can be reduced
and the network throughput can be improved. In the beacon enabled IEEE 802.15.4e network, devices
first discover the coordinator by scanning the channels. The easiest way for coordinator discovery is to
scan all channels sequentially with maximum beacon interval, which is approximately 250 s. Hence,
discovery of a beacon by scanning 16 channels would take an average of 33 min (250 s
×
16/2). In order
to mitigate this delay, we design a novel MAC mechanism for the IEEE 802.15.4e enabled WSNs and
analyze its performance. The main contributions of our work can be summarized as follows.
•
A new contention channel access scheme is designed for wireless sensors to enhance the
throughput, reliability and to reduce the power consumption.
•
A new dynamic GTS allocation scheme is designed to reduce the delay for the devices who have
failed to transmit data.
•A new beaconing scheme is designed to reduce the association delay for orphan mobile devices.
•
A Markov chain model is designed for the performance analysis of the system in terms of delay,
throughput, reliability and energy consumption due to data transmission.
Rest of the paper is organized as follows. Related works of the existing IEEE 802.15.4e standard
are given in Section 2. Network model and proposed methods are presented in Section 3. Analytical
models are given in Section 4. Performance analysis of different network parameters based on our
models are given in Section 5. Simulation results are presented in Section 6and concluding remarks
are made in Section 7.
2. Related Works
Many channel access schemes have been proposed to enhance IEEE 802.15.4 performance
for different performance parameters like throughput, delay, reliability, and energy consumption.
A generalized approach to analyze the performance of the slotted CSMA/CA scheme in the IEEE
802.15.4 standard is designed in [
3
]. Authors in [
4
] present an analytical model for the MAC scheme
of IEEE 802.15.4 to predict the energy consumption as well as the throughput of the saturated and
unsaturated traffic rates. A Markov chain based analytical model is introduced in [
5
] to evaluate the
impact of throughput and energy consumption on the probability of delivering a packet. An integrated
cross-layer model of the MAC and physical layers for unslotted IEEE 802.15.4 networks is presented
in [
6
] by considering explicit effects of multi-path shadow fading channels in the presence of interferers.
In an event driven application, all the associated devices start transmitting data simultaneously in non
beacon enabled wireless sensor network. Reference [
7
] provides an accurate performance analysis for
event-driven CSMA/CA WSNs. Authors in [
8
,
9
] have designed the time critical services and priority
based adaptive MAC schemes for the WSNs. However, these schemes do not talk how to increase
reliability of the network.
An analytical model based on a Markov chain for the GTS allocation scheme in the contention-free
period (CFP) is presented in [
10
]. Authors in [
11
,
12
] have designed the GTS allocation scheme based
on the data size and mobility of the devices for an optimal utilization of the channel bandwidth. In a
wireless network due to some hardware failure, devices associated with particular application start
Sensors 2017,17, 168 4 of 25
sending faulty data. Again limited link capacity forces the devices to send the quantized data. Under
such scenario to enhance the reliability authors in [
13
] propose a multiple superframe structure and a
non-preemptive GTS scheduling algorithm. In a wireless network, devices synchronously alternate
active and sleep modes and completely turnoff the radio during sleep mode. Authors in [
14
] provide
comprehensive review synchronous MAC schemes with respect to latency. However, authors in these
papers have not considered the retransmission opportunity for the transmission failure during GTS.
A significant amount of packet loss occurs due to interference, which degrades the expected
performance in wireless network. To avoid Wi-Fi interference among the devices in a network, devices
should use orthogonal channels. However, limited number of orthogonal channels are available.
Hence, authors in [
15
] propose one frequency agility based interference avoidance algorithm that
detects the interference and devices switch to the safe channel adaptively. Authors in [
16
] perform
theoretical analysis of throughput for WSNs under external interference with deployment of forward
error correction. Also, authors in [
17
] have studied different types of interferences, direction of signal
arrival and attenuation losses associated with smart building environment. However, none of the
works has considered the reliability of the data communications.
An analytic model is designed in [
18
] to study the MAC performance and reliability. Energy
is a major constraint for device and the major energy consumption of device is due to transmission.
As channels are considered with temporal and fixed path loss values in body area networks, authors
in [
19
] have suggested through experimental results that transmitting in
−
12 dBm to
−
15 dBm is
suitable for energy efficient body area networks. A comprehensive analysis of energy consumption
of body area networks including the effect of packet inter-arrival time is given in [
20
]. Using the
neighbors information regarding packet size, data interval, energy and consumed bandwidth, a new
MAC scheme is proposed in [
21
] to minimize the energy consumption in wireless sensor network.
Authors in [
22
] propose a priority-based adaptive time slot allocation scheme for the IEEE 802.15.4
MAC scheme. On the other hand, these schemes are based on ideal channel condition only and are
not realistic.
An analytical model is designed in [
23
] to predict the throughput and energy consumption of the
IEEE 802.15.4 networks under non-saturated wireless environment. Authors in [
24
] have designed
one additional carrier sensing (ACS) algorithm to get the information of busy channel due to data or
an acknowledge packet transmission during the second CCA. However, it is wastage of energy if the
second CCA is found busy due to data. In order to increase the chance of transmission by ignoring
the channel busy condition due to end of ACK packet transmission during first CCA, authors in [
25
]
propose one segmentized CCA. However, it has no improvements for the data transmission, if it is a
data packet instead of ACK one. In order to avoid collision in a dense network, a new variable Clear
Channel Assessment MAC scheme for WSNs along with performance analysis is designed in [
26
].
However, these schemes consumes significant amount of energy for CCAs. A mathematical model
for deriving the energy consumptions in clustered IEEE 802.15.4 WSNs is designed in [
27
], which is
suitable only for non beacon-enabled WSNs.
Performance study of sending and receiving capabilities of commonly used devices in wireless
network are made in [
28
]. Though, they investigate the communication performances of various
devices, data reliability, energy consumption and other related performance metrics have not been
studied in their work. Authors in [
29
] have experimentally analyzed the aggregate throughput
with different traffic loads and under interference of IEEE 802.11b. Performance of IEEE 802.15.4e
Time Slotted Channel Hopping (TSCH) and DSME have been studied in [
30
] to analyze the energy
consumption of the system. However, none of these work proposes new MAC mechanism for the IEEE
802.15.4e DSME. Besides, none of the work also analyze the performance in terms of reliability and
latency. A discrete-time Markov chain model is used in [
31
] to analyze the throughput of IEEE 802.15.4
MAC scheme in presence of the hidden devices. Most of the existing analytical models of IEEE 802.15.4
MAC scheme assume ideal channel conditions. However, wireless channels exhibit burst errors in real
world scenarios. A three-dimensional discrete-time Markov Chain model is proposed in [
32
], which is
Sensors 2017,17, 168 5 of 25
applied to analyze the performance of IEEE 802.15.4 MAC scheme under burst channel errors. In [
33
],
authors evaluate the impact of losing synchronization in beacon-enabled IEEE 802.15.4 network with
star topology. However, they did not considered how to reduce synchronization time for orphan or
mobile devices. To study the aggregate network throughput and packet delivery delay, authors in [
34
]
propose an analytical framework. However, they do not consider the transmission reliability. From the
survey of the current literature, it is observed that most of the performance analysis are based on the
IEEE 802.15.4 standard.
In typical WSN applications such as smart home, smart industrial automation and smart health
care, all devices are connected to a coordinator, which is a single hop scenario. It can be extended to
multihop networks by considering the communications among coordinators. In this paper, we focus
on how to optimize the performance of the single hop network. This paper not only proposes the
DSME MAC schemes for single hop WSN, but also designs the analytical models to evaluate the
performance in terms of maximum throughput, delay, energy consumption and reliability based on
the IEEE 802.15.4e instead of IEEE 802.15.4 standard.
3. The Proposed DSME MAC
In order to avoid collision due to simultaneous transmissions, a new channel access scheme is
designed for the IEEE 802.15.4e star topology WSNs. The network model and the proposed schemes
are illustrated as follows.
3.1. Network Model
Let us consider a beacon enabled star topology of IEEE 802.15.4e based WSNs with
N
numbers
of devices as
S1
,
S2
, ....,
SN
, which are connected to a PAN coordinator. It is assumed that each device
follows the DSME superframe structure of IEEE 802.15.4e as shown in Figure 1.
CAP CFP CAP CFP CAP CFP CAP CFP
Beacon Rx Beacon Rx Beacon Rx
Multi-superframe Multi-superframe
Beacon Interval
Beacon Tx
Figure 1. DSME superframe structure with BO = 6, MO = 5, SO = 4.
The CSMA/CA scheme is the key component in IEEE 802.15.4e MAC. It is adopted to arrange
the devices in the network with an appropriate order when they access the channel. All devices use
CSMA/CA channel access scheme in the CAP period to transmit data frames to the coordinator. Every
device will go for the power saving mode after packet transmission. When a device access the channel
in the CAP period, it may defer transmission, if the remaining slots are not enough to transmit the data.
3.2. A New Association Scheme
In this section, a new scheme of beaconing by the PAN coordinator is designed for the fast
association of the devices with it. As per the IEEE 802.15.4e standard, the numbers of superframes,
multisuperframes present in a DSME beacon interval are 2
BO−SO
and 2
BO−MO
, where
BO
,
SO
,
MO
are
macBeaconOrder
,
mac Supe r f rameOr der
and
mac Mul tisu per f ra meOrder
respectively. As shown
in Figure 1, if the coordinator chooses
BO =
6,
MO =
5 and
SO =
4, the number of superframes
Sensors 2017,17, 168 6 of 25
2
BO−SO
is 4. Each superframe has one exclusive beacon slot and therefore the total numbers of
available beacon slots in the DSME superframe are 2
BO−SO
. According to the current standard,
coordinator transmits a beacon only at the start of DSME superframe in one selected channel. In order
to reduce the discovery/association delay, we propose that the coordinator should broadcast the
beacon in different channels during beacon slots. Let
CH ={C H1
,
CH2
, ....
CHk}
be the set of channels
and
SUP ={SF1
,
SF2
, ....
SF2BO−SO }
be the set of superframes present in the DSME beacon interval.
According to our proposed scheme, the coordinator broadcasts the beacon in the ith channel, i.e., in
CHi
and jth superframe, i.e., in SFj.
(j+BN ∗2BO−SO)≡i MO D k f or 1≤i≤k and 1≤j≤2BO−SO (1)
where the variable BN represents the beacon sequence number.
When a new device enters into the network and wants to communicate, it should be associated
with the PAN coordinator. To perform association, the device must perform the channel scanning to
receive a beacon broadcast by the coordinator. The device keeps its radio ON during the scanning
until it receives a beacon. There are two types of scanning procedures (active and passive) suggested
in the IEEE 802.15.4e standard. In active scanning, devices have to send beacon request command in
the chosen channel, which is an energy consuming process. In passive scanning, devices have to listen
to the chosen channel until they locate the coordinator. However, passive scanning to discovery the
coordinator increases the delay. Since, power consumption is a major constraint in WSN, we focus on
passive scanning scheme to save the power. In our proposed method, the new device that wants to
join the network should choose a random channel and perform the passive scan, i.e., it waits for the
beacon in that channel. The device definitely gets the beacon as the coordinator broadcasts the beacon
in every channel. Upon receiving the beacon frame, the devices send the association request frame
using CSMA/CA and the coordinator sends back the association response frame, if it has received the
association request successfully. Otherwise, the devices repeat the association procedure as suggested
in the IEEE 802.15.4e standard.
In our proposed scheme, un-associated devices can chose one random channel and perform
the passive scan. Hence, our scheme can reduce the busy channel conditions and collisions when
un-associated devices are increase. However, according to the IEEE 802.15.4e, all un-associated devices
can locate the coordinator simultaneously through the passive scan and all of them can have access to
the channel simultaneously. This procedure can increase the delay for the discovery of the coordinator
and association with it. On the other hand, our procedure can significantly reduce the delay for
discovering the coordinator and association with it. The detailed procedure of our proposed schemes
are given in Algorithm 1and Algorithm 2for the coordinator and devices, respectively.
Algorithm 1 A new beaconing scheme for coordinator.
Require:
Set of channels
CH ={C Hi|
1
≤i≤k}
, beacon order (
BO
), superframe order (
SO
) and
beacon sequence number (BN).
Ensure: Beacon broadcast in every channel.
1: l=2BO−SO;
2: Set of superframes SUP ={SFj|1≤j≤l};
3: for (1 ≤i≤k)do
4: for (1 ≤j≤l)do
5: if ((j+BN ×l)MOD k =i)then
6: Broadcast beacon for superframe SFjin channel CHi;
7: end if
8: end for
9: end for
Sensors 2017,17, 168 7 of 25
Algorithm 2 A new association scheme for device.
Require: Set of channels CH ={CHi|1≤i≤k}.
Ensure: Successful association with a suitable coordinator.
1: Generate a random number r=random (1, k);
2: Passive scan in channel CHr.
3: if Beacon received then
4: repeat
5: Compete with other devices to associate with the coordinator;
6: until Successful association.
7: else
8: Wait for beacon;
9: end if
3.3. A New Channel Access Scheme
CSMA/CA scheme is the key component in IEEE 802.15.4e MAC. It is adopted to arrange the
devices in the network with an appropriate order when they access the channel. When too many
devices have data to transmit to the coordinator at the same time, they try to access the channel
simultaneously, which increases the collision and makes the channel busy. As per IEEE 802.15.4e
standard, the numbers of superframes present in the DSME superframe are 2
BO−SO
. Hence, the number
CAP period presented in the DSME superframe will be 2
BO−SO
. To restrict the number of users to
access the channel simultaneously, we propose here a new channel access scheme in which each device
is allowed to contend the channel only in one CAP period.
Upon receiving the beacon successfully, each device knows the value of BO and SO from the
beacon and gets the numbers of CAP period presented in the DSME superframe as 2
BO−SO
. Based on
the following equation, each device calculates the index of its assigned CAP period.
CAPindex = (SID +No f f s et)MOD(2BO−SO )(2)
where
SID
is ID of each device and
No f f s et
represents the offset value in the mapping function.
No f f s et
is determined based on a time stamp divided by a beacon interval field in a beacon frame from the
coordinator. It is to be noted that
No f f s et
improves the fairness among the devices as it is changed
with respect to the beacon interval. Hence, the mappings of devices to the CAPs are different among
beacon intervals. All devices upon receiving the beacon are synchronized and go for the power
saving mode. The devices wake-up at the start of the assigned CAP period to access the channel.
The devices those are assigned to a CAP period contend with each other in that CAP period only.
Our proposed mapping function can allocate the devices in to different CAPs so that the collision rate
can be reduced significantly.
After the allocation, the variables NB, CW, BE and RT are initialized for each allocated device.
NB is the number of times the CSMA/CA scheme required to delay while attempting the current
transmission, which is initialized to 0 before every transmission. CW is the contention window length
and BE is the backoff exponent. RT is the maximum number of retransmission attempts. Each device
generates a random backoff counter
(Rb)
in the range of
[
0; 2
BE −
1
]
units. When the backoff period is
count down to zero, the tagged device performs the CCA if the remaining time
RCA P
in the CAP period
is enough to perform the CCA and to transmit the required data. Let the duration of one backoff slot,
time duration to perform one CCA, time for data transmission be
σ
,
TCCA
and
TD
, respectively. It is
to be noted that the CAP and CFP period in the superframe is fixed. According to the standard, GTS
slots are allotted to different devices based on their requests received by the coordinator during CAP.
However GTS slots may be under utilized, if coordinator receives less number of requests as compared
to the available slots in the CFP, which will affect the network throughput. Hence, we modify the
beacon format by adding one bit field called Cross CAP Boundary (CCB) as shown in Figure 2to
utilize the unused slots, which is allowed by the standard.
Sensors 2017,17, 168 8 of 25
Octets: 1/2 0/1 variable 0/1/ 5/6/10/14 variable variable 1 2
Frame control
Sequence
number
Adrressing
fields
Auxiliary Security
Header
Information elements Beacon
payloads CCB
GACK
FCS
Header IES Payload IES
MHR MAC payload MFR
Proposed beacon frame
Figure 2. A new beacon frame for IEEE802.15.4e.
As per the IEEE 802.15.4e standard, it is to be noted that a device doubles the contention window
size up to
MAXN B
and performs the CCA again by choosing a random delay within the contention
window and discards the packet after maximum number of retransmission
MAXRT
attempts if the
device finds the channel busy during any of the CCA. After finding the channel idle in the first CCA,
the second CCA is performed and transmission is started if the channel is found idle. This procedure is
repeated if the channel is found busy during the second CCA, which may be incurred by two reasons.
When a tagged device performs the first CCA if another device performs its second CCA, the channel
remains busy. Also the channel remains busy, if the tagged device performs the first CCA during the
acknowledgment waiting time
TACK
of another device. One solution to reduce the channel busy is to
increase the duration of the backoff period of a sensor. However, by doing so, the network performance
like transmission delay will be increased and fairness may not be maintained. Besides, it could be
possible that a node may still assess the channel busy during the transmission of ACK by other nodes.
For example, as shown in Figure 3, node
B
assesses busy channel, when node
A
transmits the ACK
packet to the coordinator. If the backoff period is increased, it will be applicable for all devices and
accordingly, node
B
can have same problem, when the ACK transmission is going on in the medium.
However, our proposed mechanism can avoid this problem as a node has to adopt two idle slots after
its first CCA and by that time ACK transmission if any can be completed.
Device A
Device B
Device A
Device B
(a) IEEE 802.15.4e CCA
(b) Our proposed CCA
Channel busy
t0
t1
t0
t1
CCA Idle Data Ack
backoff
t
t
t
t
t
t
1
Figure 3. A new CCA scheme for IEEE802.15.4e to avoid channel busy due to acklodgement.
Sensors 2017,17, 168 9 of 25
The tagged device can transmit the data packet and waits for the acknowledgment, if the channel
is found idle during the second CCA. After the
TACK
time is out, if the device does not receive the
acknowledgment, it assumes that collision might have occurred in the channel and therefore repeats
the procedure. The detailed procedure of our proposed scheme is given in Algorithm 3.
Algorithm 3 New channel access scheme.
Require: NB =
0,
BE =macMinBE
,
RT =
0,
σ=
20
symbols
,
TCCA =
8
symbols
,
CCB =
0
or
1 and
packet duration time TD.
Ensure: Transmission success/failure.
1: ∆=2×TCCA +2×σ+TD
2: if Beacon received then
3: Calculate the CAPindex and wakeup at start of the C APindex ;
4: else
5: Wait for beacon;
6: end if
7: Generate a random backoff time Rb=random(0, 2BE−1)×σ;
8: if ((RC AP ≤∆)or CCB =0)then
9: Wait for next superframe and go to step 2;
10: else
11: Perform CCA1 after Rbduration;
12: if Channel found busy then
13: Switch to sleep state for the random(0, TD)and go to step 30;
14: else
15: Wait for 2σand perform CCA2;
16: if Channel found busy then
17: Switch to sleep state for TDand go to step 30;
18: else
19: Start transmission;
20: if Transmission failure then
21: RT =RT +1;
22: if RT ≤MAXRT then
23: Go to step 7;
24: else
25: Transmission failure;
26: end if
27: end if
28: end if
29: end if
30: NB =NB +1, BE =min(BE +1, macMaxBE);
31: if (NB >MAXN B )then
32: Transmission failure;
33: else
34: Go to step 7;
35: end if
36: end if
According to IEEE 802.15.4e enabled star topology, all devices need to be associated with a single
coordinator. Accordingly, all devices can have right to access the channel simultaneously and, as a
result, collisions may occur. However, in our proposed scheme, devices are distributed to different
superframes during their channel access mechanism so that less number of devices can access to
Sensors 2017,17, 168 10 of 25
the channel as compared to the IEEE 802.15.4e standard and therefore can minimize the collisions.
Thus, our proposed mechanism can be suitable even in the large scale network.
3.4. Proposed Dynamic GTS Allocation Scheme
Group acknowledgement (GACK) plays an important role to provide the transmission
opportunity within the same DSME superframe. The failed devices who get the information in
GACK can have a chance to transmit data again in the allotted slots within the DSME superframe.
In the current IEEE 802.15.4e standard, the coordinator allocates two separate slots within the CFP for
this propose. Again the device has to reserve two slots, i.e., GTS and GTSR for each transmission slot
in order to avail the retransmission opportunity within the DSME superframe. The next GTSR allotted
slot will be used for new data packet of that device, if the transmission is successful in the first GTS
allotted slot. Most of the time the devices may not have new data to send in the GTSR allotted slots.
Hence, the overall throughput may be reduced significantly under this scenario.
To overcome the drawbacks of the GTS scheme of IEEE 802.15.4e standard, we design here a
new low complexity and reliable GTS allocation scheme. In order to increase the network throughput
and to reduce the delay for retransmission, we modify the beacon format by adding a new seven bit
field to the beacon frame called GACK as shown in Figure 2. The CFP of one superframe contains
seven GTS slots and these seven bits will be one or zero based on the corresponding GTS transmission
is success or failed, respectively. Upon receiving the beacon, each device counts the number of bits
having value zero among these seven bit field and starts the channel access during CAP after those
number of slots. The failed devices find the suitable slot during the CAP based on the received seven
bit field and transmit again without following the CSMA/CA procedure. Hence, the delay can be
minimized and unnecessary energy consumption due to contention in the CAP can also be reduced.
As shown in Figure 4, let devices
S1
,
S2
...,
S7
be allotted to GTS slots 1, 2, ..., and 7, respectively.
It is assumed that devices
S2
,
S3
,
S7
fail to transmit the data during their allotted 2nd, 3rd and 7th
GTS slots, respectively due to interference or fading or channel error. The coordinator will inform
the transmission failure through the GACK field present in the beacon frame as shown in Figure 2.
The value in the 2nd, 3rd and 7th bit of the GACK field is 0. The superframe presents in a DSME
superframe contains 16 slots out of which one slot is a beacon slot, next eight (1–8) slots are CAP and
the rest seven (9–15) slots are CFP. devices those who receive the beacon get the information about
these failed transmission by counting the number of zero bits in the GACK field and start the channel
access from the 4th slot of the current CAP. The failed devices
S2
,
S3
,
S7
will start retransmission at slot
number 1st, 2nd and 3rd, respectively in CAP without contention.
BEACON
DSME Superframe
BEACON
S1
S2
S3
S4 S5
S6 S7 S2
S3
S7
BEACON
BEACON
Figure 4. Dynamic GTSR allocation for the failed GTS transmission.
According to the existing mechanism of IEEE 802.15.4e, failed devices has to wait for the next
retransmission slots to be allocated in the GTS. However, as shown in Figure 2, we add a GACK slot in
our proposed MAC so that the failed devices neither have to contend for the channel nor have to wait
for the next allotted retransmission slots. Besides, the proposed scheme can have the same superframe
structure defined by the IEEE 802.15.4e standard.
Sensors 2017,17, 168 11 of 25
4. Analytical Models for DSME MAC
In this section, we propose the Markov model to analyze the channel access scheme of the devices
in the shared slots. It is assumed that the traffic condition is unsaturated. Based on such unsaturated
traffic condition and the notation given in Table 1, the analytical model can be designed as follows.
Table 1. Notation table.
Notation Meaning
MTotal number of devices assigned to one superframe;
LDLength of a data packet;
LACK Length of an acknowledgement packet;
qPacket generation probability;
RbDuration of random backoff;
pProbability that remaining time slots are sufficient to transmit a packet;
φProbability that a device is sensing the channel first time;
αProbability that the channel is busy during first CCA;
βProbability that the channel is busy during second CCA;
Pb(ζ)Probability of channel bit error rate;
PSProbability of transmission success;
PCProbability of transmission collision;
Ptr Probability of transmission;
TCCA Duration of CCA;
TDDuration of data transmission;
We assume that Mdevices are allotted to one superframe for data communication with the
coordinator. For a given device, the stochastic processes
s(t)
and
c(t)
represent the backoff stage for
NB and backoff counter for CW, respectively as shown in Figure 5. Let
bi,x=lim
t→∞P{s(t) = i,c(t) = x},
where
x∈ {−
1, ...,
Wi−
1
}Wi=
2
min(i+macM I NBE,macM AX BE)i∈ {
0, ...,
m}
. The time
t
corresponds to
the beginning of the slot time and is directly related to the system time.
bi,−1
represents the states
corresponding to the start of the channel access, where
i∈ {
0, ...,
m}
. In our our proposed model,
since a device has to wait for two backoff slots before performing the second CCA, we represent the
states
bi,−2
and
bi,−3
corresponding to the waiting states prior to the second CCA, where
i∈ {
0, ...,
m}
.
bi,−4
represents the states corresponding to the second CCA before start for transmission, where
i∈ {
0, ...,
m}
.
b−1,l
, represents the transmission state, where
l∈ {
1, 2, ...,
L}
.
L
is the number of
transmission slots. The state
I
stands for the idle state when a device has no request packet to transmit.
q
is defined as the probability that the device has a packet to transmit. After generating the request
packet, a device goes to the state b−1,0. Upon reaching at the state b−1,0, the device chooses a random
backoff and goes to the state
bi,0
, where
i∈ {
0, 1, 2...,
W0−
1
}
. The state
bi,j
represents the backoff
states, where i∈ {0, ..., m},j∈ {−1, ..., Wi−1}.
Let,
α
and
β
be the probability of accessing channel busy during the first and second CCA,
respectively. Let
p
be the probability that the remaining time in the CAP period is sufficient to complete
the data transmission. A device goes to the transmission state, if the channel is idle and attempts to
transmit requests. Let PSbe the probability that the transmission is successful.
Sensors 2017,17, 168 12 of 25
!
"#$#! "#$%! "#$&!
'!
%!
%! %! %!
%('!
")$#! ")$%!
'!
%! %! %!
")$&!
"#$*#(%!
"m !"#$%
$#&%
'(%
$#)%
)%
$*!+%
,#$ +
$#'(%
$*!"%
$%
$%
,#$ -% ,#$ -#$% ,%#$ $%
,+ #$%
,+ #.%
,+ #/%
,+ #0%
," #$%
," #.%
," #/%
," #0%
&%
1%
$#1%
$#1%
&%
1%
$#'%
$#&%
$%
$%
Figure 5. Markov model for the proposed CSMA-CA scheme.
Based on the proposed Markov chain model given in Figure 5, the steady state probabilities can
be derived as follows.
P{bi,k|bi,k+1}=1f or 0≤i≤m and 0≤k≤Wi−1. (3)
P{bi,k|bi−1,0}=p+ (1−p)α+ (1−p)(1−α)β
Wi
f or 0≤i≤m and 0≤k≤Wi−1.
(4)
P{I|bm,0}= ( p+ (1−p)α+ (1−p)(1−α)β) + (1−p)(1−α)(1−β)PS.(5)
Equation (3) represents the decrement of backoff counter, which occurs with probability 1.
Equation (4) represents the probability of finding channel busy during CCA and thereafter a device
selects uniformly a state in the next backoff state. Equation (5) gives the probability of going back to
the idle state from the last backoff state after finding channel idle or busy during the CCA. Based on
these transition probabilities, we can derive the steady state probabilities as follows. The steady state
probability that a device enters into the next backoff state after the channel access failure in the current
backoff state is
bi,k=Wi−k
Wi
[bi−1,0{p+ (1−p)α+ (1−p)(1−α)β}];
f or 1≤i≤m and 0≤k≤Wi−1.
(6)
It is to be noted that a tagged device starts transmission upon accessing to a channel successfully.
However, if the transmission is successful or failed without exceeding the retry limits, it will again
enter into the initial backoff state. Hence, the corresponding steady state probability is
b0,k=W0−k
W0
[{
m
∑
i=0
bi,0}(1−p)(1−α)(1−β){(1−PS) + PSq}
+bm,0(p+ (1−p)α+ (1−p)(1−α)β)q]; 0 ≤k≤Wi−1.
(7)
Sensors 2017,17, 168 13 of 25
Normally, a device goes for CCA after the random backoff period and the remaining time in
the CAP period is sufficient for the data transmission. Therefore, the corresponding steady state
probability can be deduced as follows.
bi,−1= (1−p)bi,0; for 0 ≤i≤m.(8)
Normally, a device goes for the second CCA, if the channel is found idle during the first CCA.
Hence, the corresponding steady state probability can be derived as follows.
bi,−y= (1−α)bi,−1; for 0 ≤i≤mand −4≤y≤ −2. (9)
After accessing the channel successfully in the
ith
backoff state, the device starts transmitting the
data and the corresponding steady state probability is given as follows.
b−1,l= (1−β)bi,−4; for 0 ≤i≤mand for 1 ≤l≤L.(10)
A device that enters to the idle state either completes the data transmission or has no packet to
transmit. Hence, the corresponding steady state probability is
I=1
q(PSb−1,1 + (p+ (1−p)α+ (1−p)(1−α)β)bm,0 ).(11)
If a device has new packet or needs to retransmit the unsuccessful packet, it enters to the active
state. The corresponding steady state probability is
b−1,0 = (1−PS)b−1,1 +qI.(12)
According to the rule of the probability, i.e., sum of the probabilities of all states should be equal
to 1, we can get the following equation.
I+b−1,0 +
m
∑
i=0
Wi−1
∑
j=0
bi,j+
4
∑
x=1
m
∑
i=1
bi,−x+
L
∑
l=1
b−1,l=1(13)
By solving these equations, we can get the probabilities of all states. Then, the probability that
a device attempts the CCA for the first time within a slot for transmitting data can be calculated
as follows.
φ=
m
∑
i=0
bi,0(1−p).(14)
5. Performance Analysis
The probability of the bit error rate
Pb(ζ)
for the transceiver of IEEE 802.15.4 radios is calculated [
1
]
as given in the following equation.
Pb(ζ) = 8
15
1
16
16
∑
k=2
(−1)k16
ke20ζ(1
k−1)(15)
where,
ζ
is the signal-to-interference-plus-noise ratio (SINR). Let,
LD
be the length of the data packet
and
LAck
be the length of the acknowledgement packet. Let,
q
be the probability that a device has a
packet to transmit and
α1
be the probability that the first CCA is busy due to data transmission by
other devices. A device starts transmission, if the channel is found ideal. Hence,
α1=LD(1−(1−φ)M−1)(1−α)(1−β)(16)
Sensors 2017,17, 168 14 of 25
Coordinator will send the acknowledgement packet, if it receives the packet without any collision
or channel error. In other words, we can say if exactly one device transmits without any channel error,
then the coordinator will send the acknowledgement packet. Let
α2
be the probability that the first
CCA is busy due to transmission of acknowledgement by the coordinator. Hence,
α2=Mφ(1−φ)M−1(1−α)(1−β)(1−Pb(ζ))LD ata (17)
where,
LData
is the length of the data packet in bits. Let,
α
be the probability that the first CCA is
busy. During the first CCA, the device will find the channel busy due to the transmission of data or
acknowledgement. Hence, the probability of finding the channel busy during the first CCA is given
as follows.
α= [LD(1−(1−φ)M−1) + Mφ(1−φ)M−1(1−Pb(ζ))LData ](1−α)(1−β).(18)
Let
β
be the probability that the second CCA is busy. A device will perform the second CCA,
if the channel is found ideal during the first CCA. A device transmits the data, if the channel is found
idle after the second CCA. In our scheme, the channel will be busy during the second CCA due to
data transmission as the acknowledgement packet is avoided by keeping two ideal slots between two
CCAs. Hence, the tagged device will find the channel busy during its second CCA, if and only if any
other device completes his second CCA and starts transmitting data. Hence,
β=3(1−(1−φ)M−1)(1−β).(19)
The device will go for the next backoff, if the channel access fails. Hence, the probability of going
to the next backoff is
Pbac ko f f =p+ (1−p)α+ (1−p)(1−α)β.(20)
5.1. Reliability
A transmission is successful if exactly one device is transmitting without any collision and channel
error. A collision occurs if at least one of the remaining devices starts sensing the channel in the same
slot and transmits data upon finding the channel idle. Hence, the probability of collision is
PCol lision =1−(1−φ)M−1(21)
where,
M
is the number of devices. A transmission will be successful, if the device receives
the acknowledgement packet corresponding to its transmission. The coordinator sends the
acknowledgement, if exactly one device transmits data without any channel error. Therefore,
the probability of successful transmission by a device is
PS=Mφ(1−φ)M−1
(1−(1−φ)M)(1−Pb(ζ))LData .(22)
The packets are discarded mainly either due to channel access failure or due to retry
limits. Channel access failure occurs, if the tagged device does not find the channel idle during
maxM acBa cko f f number of backoffs. Thus, the probability of channel access failure is
Pc f ={p+ (1−p)α+ (1−p)(1−α)β}bm,0 .(23)
Considering the maxFrameRetries =r, the probability of transmission failure in rattempts is
Pt f = (1−PS)r+1(24)
Sensors 2017,17, 168 15 of 25
Probability of successful packet reception within the maximum number of channel access and
retransmission is called as reliability. Hence the reliability is
PR=1−Pc f −Pt f (25)
5.2. Throughput
Normalized throughput is the ratio of average time occupied by the successful transmission to
the interval between two consecutive transmissions. Let
S
be the normalized throughput of a device.
The channel is sensed busy if at least one device is transmitting data. Let
Ptr
be the probability that
there is at least one transmission. Since there are
M
devices attached to the coordinator and
φ
is the
probability that a device attempts its first CCA, then
Ptr = (1−(1−φ)M)(1−α)(1−β)(26)
Hence the normalized throughput of a device is
S=Ptr PSLD
(1−Ptr )σ+PtrPSTs+Ptr (1−PS)Tc(27)
where,
Ts
is the mean time of successful transmission, and
Tc
is the mean time that the channel is
occupied because of the collision.
LD
is the time of payload and
σ
is the duration of a unit backoff slot.
Tsand Tcare calculated as
(Ts=2TCCA +2σ+TD+δ+TACK
Tc=2TCCA +2σ+TD+δmax )(28)
where,
TCCA
,
TD
,
TACK
,
δ
,
δmax
and
σ
are the time duration for performing
CCA
, transmitting data,
receiving an acknowledgement, waiting time for an acknowledgement, maximum waiting time for an
acknowledgement and duration of one backoff slot, respectively.
5.3. Network Delay
The total time required for the information to be generated by the source device and to be received
by the destination device is defined here as the network delay. Let,
Dj
be the delay that the packet is
transmitted successfully at time j+1 and the packet has failed for jtimes. Then
Dj=Ts+jTc+
j
∑
i=0
Db(29)
where,
Db
is the delay that the device successfully found the channel idle during
maxM acBa cko f f
number of backoff slots.
Ts
is the mean time of successful transmission and
Tc
is the mean time that the
channel is occupied because of the collision. Let
Bk
be the event that the channel access is successful
for
(k+
1
)
th times and the channel access is failed for
k
th times. Let
B
be the event that the channel
access is successful within
maxM acBa cko f f
number of backoff. Hence, the expected backoff delay in
one transmission attempt is
Db=
m
∑
i=0
Pi
back o f f (1−Pbacko f f )
1−Pm+1
back o f f
i
∑
k=0
Wi−1
2(30)
If the packet is generated outside its allocated CAP, then it has to wait until the next allocated
CAP comes. Therefore, if the packet arrival follows the Poisson process with packet arrival rate
λ
, then
waiting time =ZBI
0(BI −x)λe−λxdx.(31)
Sensors 2017,17, 168 16 of 25
Propagation delay is the amount of time it takes for the head of the signal to travel from the sender
to the receiver. It can be computed as the ratio between the link length and the propagation speed over
the specific medium. Let
Ej
be the event that the packet is transmitted successfully
j+
1 times and
the packet has failed for
j
times. Let
E
be the event that the packet is transmitted successfully within
maxFrameRetries. Considering the maxFrameRetries =r,
Pr(Ej/E) = (1−PS)jPS
1−(1−PS)r+1.(32)
Therefore the network delay is
D=
r
∑
j=0
Pr(Ej/E)Dj+waiting time +propagation delay.(33)
The packet may not be transmitted in the current CAP period if
D>CAPperiod
and has to wait
for the next superframe. Therefore, the total average delay is
Del ay =bD
CAPperiod
cBI +D mod CAPperiod.(34)
5.4. Energy
In this section, the energy consumption, which is defined as the average energy consumption
by a device to transmit one slot amount of data successfully is analyzed. We consider
PRX
and
PTX
are the energy consumption to receive and to transmit a packet, respectively. Sensor devices are
assumed to be in sleep state during the backoff procedure by which they do not consume any energy.
The energy consumption due to turnaround process from the sleeping state to the sensing one is
assumed to be
PRX+PTX
2
. Taking
TL
,
TA
,
TCCA
,
Tta
and
δmax
as the time duration of transmitting a packet,
receiving an acknowledgement, each successful CCA, turnaround time, maximum time to wait for the
acknowledgement, respectively, the total energy consumption per device can be analyzed as follows.
E=αTCCA PRX +2(1−α)βTCCA PRX + (1−α)(1−β){(1−PS)Ec+PSEs}
(1−α)(1−β)PSTdata
(35)
where,
Es
is the energy consumption of successful transmission, and
Ec
is energy consumption
due to collision. Esand Eccan be calculated as
Es=2TCCA PRX +2Tta PRX+PTX
2+TLPTX +TAPRX
Ec=2TCCA PRX +2Tta PRX+PTX
2+TLPTX +δma x PRX )(36)
6. Simulation
In this section, we validate and evaluate our models by using OMNeT++ [
35
] simulator.
We perform simulations to compare our proposed schemes with IEEE 802.15.4e. In our simulation,
star topology is considered, where each device is deployed randomly around the coordinator. Our work
is compared with the IEEE 802.15.4e standard for successful transmission, throughput, reliability,
average network delay and energy consumption taking poisson arrival rate. The simulation parameters
given in Table 2are set as per the IEEE 802.15.4e standard. The simulation is performed with 50 runs
and the results are given taking average of those 50 runs.
Sensors 2017,17, 168 17 of 25
Table 2. Simulation parameters.
Parameters Value
Radio band 2.4 GHz
Channel bandwidth 250 kbps
Carrier sense sensitivity −85 dBm
Channel number 11∼26
Beacon interval 15.6 ×24ms
Unit backoff period 20 symbol
PHY overhead 6 byte
MAC overhead 3 byte
Transmission current consumption 9.1 mA
Receiving current consumption 5.9 mA
Turnaround current consumption 7.5 mA
Sleep current consumption 0.001 mA
As shown in Figure 6, different number of devices are considered to evaluate the corresponding
packet success probability. It is observed that the packet success probability decreases as the number
of devices attached to a coordinator increases irrespective of the beacon order. When the data payload
of 100 bytes and 500 number of devices are considered, the success probability is 0.96, if the beacon
order is 10. But, the probability is 0.85, 0.70 and 0.47 for the beacon order 9, 8 and 7, respectively.
The difference between the simulation and analytical results varies little due to the distribution of
devices to different superframes. The number of associated devices to one CAP decreases by increasing
the beacon order.
!"
!#
!$
!%
!&
!'
!(
!)
!*
"
" # $ % & ' ( ) *
"
!"#$%&'(""$''&)*+,!,-.-%/&
0(1,$*&+2&sensor devices
!"#$%&'()*+,-().
!"#$%/,01)(+,2'.
!"#3%&'()*+,-().
!"#3%/,01)(+,2'.
!"#4%&'()*+,-().
!"#4%/,01)(+,2'.
!"#56%&'()*+,-().
!"#56%/,01)(+,2'.
Figure 6. Validation of packet success probability for different beacon order.
As shown in Figure 7, our work is compared with the IEEE 802.15.4e standard based on the
simulation results for different values of beacon order and superframe order. It is observed that
the packet success probability of our work is higher as compared to the IEEE 802.15.4e. Figure 8,
demonstrates the throughput corresponding to the number of devices. It is observed that the
throughput decreases as the number of devices attached to a coordinator increases irrespective of the
Sensors 2017,17, 168 18 of 25
beacon order. When the data payload of 100 bytes and five hundred number of devices are considered,
the throughput is 0.27, if the beacon order is 10.
!"
!#
!$
!%
!&
!'
!(
!)
!*
"
" # $ % & ' ( ) * "
!"#$%&'(""$''&)*+,!,-.-%/&
0(1,$*&+2& !" #$%&!'()! &
+++++) #!"&!%,+-./01/21
+++++324546,1+scheme
Figure 7.
Comparison of packet success probability of our proposed scheme with the IEEE
802.15.4e standard.
! "
!#
!#"
!$
!$"
!%
# $ % & " ' ( ) *
#
!"#$%!&$'(
)$*+,"(#-(sensor devices
!"#$%&'()*+,-().
!"#$%/,01)(+,2'.
!"#3%&'()*+,-().
!"#3%/,01)(+,2'.
!"#4%&'()*+,-().
!"#4%/,01)(+,2'.
!"#56%&'()*+,-().
!"#56%/,01)(+,2'.
Figure 8. Validation of throughput for different beacon order.
We compare our work with the IEEE 802.15.4e standard for equal number of beacon and
superframe order as shown in Figure 9. It is found that the throughput of our scheme is higher
than the standard except when the number of devices are less than 15. This happens as we keep
two idle slots between CCA1 and CCA2 in our proposed scheme to avoid the channel busy due
to acknowledgement. Also, we find that the analytical result well match with the simulation one.
Sensors 2017,17, 168 19 of 25
The reliability of data transmission with respect to different numbers of devices is compared as
shown in Figure 10. It is observed that the reliability decreases as the number of devices attached
to a coordinator increases irrespective of the beacon order. When the data payload of 100 bytes
and 500 number of devices are considered, the reliability becomes 0.99 when beacon order is 10.
However, it decreases when the value of beacon order becomes 9, 8 and 7.
! "
!#
!#"
!$
!$"
!%
# $ % & " ' ( ) * #
!"#$%!&$'(
)$*+,"(#-( !" #$%&!'()!
+++++) $!#"!&,+-./01/21
+++++324546,1+scheme
Figure 9. Comparison of throughput of our proposed scheme with the IEEE 802.15.4e standard.
!"
!#
!$
!%
!&
!'
(
( ) * " # $ % & '
(
!"#$%#"#&'(
)*+%!,(-.(sensor devices
!"#$%&'()*+,-().
!"#$%/,01)(+,2'.
!"#3%&'()*+,-().
!"#3%/,01)(+,2'.
!"#4%&'()*+,-().
!"#4%/,01)(+,2'.
!"#56%&'()*+,-().
!"#56%/,01)(+,2'.
Figure 10. Validation of reliability for different beacon order.
As shown in Figure 11, our work is fairly compared with the IEEE 802.15.4e standard maintaining
equal beacon and superframe order. It is observed that the reliability of our scheme is higher and the
simulation result matches with the analytic one. Figure 12, demonstrates the delay corresponding
to the number of devices. It is realized that the delay increases as the number of devices increases
Sensors 2017,17, 168 20 of 25
irrespective of the beacon order. When the data payload of 100 bytes and 500 sensor devices are
considered, the delay is nearly 74 ms, if the beacon order is 10.
!"
!#
!$
!%
!&
!'
!(
)
) * " # $ % & ' ( )
!"#$%#"#&'(
)*+%!,(-.( !" #$%&!'()! (
+++++' *!)$!#,+-./01/21
+++++324546,1+scheme
Figure 11. Comparison of reliability of our proposed scheme with the IEEE 802.15.4e standard.
!
"
#!
#"
$!
$"
%!
%"
&!!
&!"
&&!
&!! '!! (!! )!! "!! !! #!! $!! %!!
&!!!
!"#$%&'"'(!)*+,-.
/0&1!2.*3.sensor devices
!"#$%&'()*+,-().
!"#$%/,01)(+,2'.
!"#3%&'()*+,-().
!"#3%/,01)(+,2'.
!"#4%&'()*+,-().
!"#4%/,01)(+,2'.
!"#56%&'()*+,-().
!"#56%/,01)(+,2'.
Figure 12. Validation of packet delay for different beacon order.
We have compared our work with IEEE 802.15.4e standard setting beacon and superframe orders
as 2 as shown in Figure 13. It is noticed that delay of our work is significantly less and the simulation
result matches with the analytic one. This happens as our proposed method avoids the unnecessary
backoff durations. The delay suddenly increases in case of IEEE 802.15.4e when the number of devices
reaches to 70. This situation occurs due to channel access failure or transmission failure by a device and
Sensors 2017,17, 168 21 of 25
the device has to wait for the start of the next superframe. It is to be noted that the device has to wait for
the CFP period before accessing the channel again. However, in our protocol, the chance of accessing
the channel in the next superframe is significantly less due to the high transmission success rate as
compared to the IEEE 802.15.4e. Figure 14, shows the minimum GTS packet delay with respect to
different numbers of beacon order. It is observed that the minimum GTS retransmission delay remain
constant irrespective of the beacon order. This happens as the MAC scheme of our scheme allows the
failed devices to retransmit data during the CAP without waiting for the next allotted retransmission
slots. As shown in Figure 15, we compare our scheme with IEEE 802.15.4e standard taking beacon
order from 1 through 5 and superframe order as 1. It is found that coordinator discovery delay of our
scheme remains constant though it exponentially increases in case of IEEE 802.15.4e. This happens as
the coordinator broadcasts the beacon sequentially in every channel in our proposed scheme.
!
"!
#!
$!
%!
&!
'!!
'! (! )! ! "! #! $! %! &! '!!
!"#$%&'("()!*+,-.%
/0'1!2%+3% !" #$%&!'()!
*****%!(+'"+ ,*-./01/21
*****324546,1*scheme
Figure 13. Comparison of packet delay of our scheme with IEEE 802.15.4e standard.
!
"
"!
#
#!
$
"#$%!
!"#"$%$ &'( )*+,- .$"+"/*01#)2
3*,01# 14)*4
&&&&&' #("!(%)&*+,-.,/.
&&&&&0/1213).&scheme
Figure 14. Comparison of GTS retransmission delay of our scheme with IEEE 802.15.4e standard.
Sensors 2017,17, 168 22 of 25
!
"
"!
#
#!
$
$!
%
%!
" # $ % !
!""#$%&'("# $%)"*+#, $+-', ./%-%0+)"&$1
2+')"& "#$+#
&&&&&' #("!(%)&*+,-.,/.
&&&&&0/1213).&scheme
Figure 15.
Comparison of minimum coordinator discovery delay of our proposed scheme with the
IEEE 802.15.4e standard.
As shown in Figure 16, different numbers of devices are considered to evaluate the corresponding
energy consumption in joule. It is observed that the energy consumption increases as the numbers
of devices attached to a coordinator increase irrespective of the beacon order. When the data
payload of 100 bytes and five hundred number of devices are considered, the energy consumption is
nearly 1.64
×
10
−5
joule, if the beacon order is 10. As shown in Figure 17, our work is compared
with IEEE 802.15.4e standard for equal number of beacon and superframe order. It is found that the
energy consumption of our scheme is less as compared to the IEEE 802.15.4e standard. This happens
as the numbers of sensors in our scheme are restricted to access the channel within one CAP by which
number of collisions is reduced significantly. Besides, the simulation result of our scheme matches
with the analytic one.
0.00E+00
2.00E-05
4.00E-05
6.00E-05
8.00E-05
1.00E-04
1.20E-04
1.40E-04
100 200 300 400 500 600 700 800 900
1000
Energy Consumption (joule)
Number of devices
BO=7(Analytical)
BO=7(Simulation)
BO=8(Analytical)
BO=8(Simulation)
BO=9(Analytical)
BO=9(Simulation)
BO=10(Analytical)
BO=10(Simulation)
1.4×10-4
1.2×10-4
2.0×10-5
8.0×10-5
1.0×10-4
6.0×10-5
4.0×10-5
0.0×100
Figure 16. Validation of energy consumption for different beacon order.
Sensors 2017,17, 168 23 of 25
0.00E+00
2.00E-05
4.00E-05
6.00E-05
8.00E-05
1.00E-04
1.20E-04
1.40E-04
1.60E-04
1.80E-04
10 20 30 40 50 60 70 80 90 100
Energy Consumption (joule)
Number of devices
802.15.4e Standard
Proposed protocol
1.8×10-4
1.6×10-4
1.4×10-4
1.2×10-4
1.0×10-4
8.0×10-5
6.0×10-5
4.0×10-5
2.0×10-5
0.0×100
Figure 17. Comparison of energy consumption of our scheme with IEEE 802.15.4e standard.
7. Conclusions
In this paper, a new channel access scheme and a beacon broadcast scheme are proposed to
avoid the collision in dense wireless network with mobile devices. A new dynamic guaranteed
retransmission slot allocation scheme is proposed to reduce the GTS retransmission delay, which is
reduced significantly as compared to the IEEE 802.15.4e standard. Analytical models are designed and
performance analyses are made for the data transmission reliability, throughput, energy consumption
and success rate probability based on our proposed schemes. Our models are validated with the
simulation and analytical methods, which are well matched and comparisons of our schemes are made
with IEEE 802.15.4e standard. The results obtained from the analytical and simulation show that our
schemes can improve the reliability, throughput, delay and energy consumptions significantly.
Acknowledgments:
This work is partly supported by Ministry of Science and Technology (MOST), Taiwan under
the grant number 105-2221-E-182-050 and partly under the grant number 104-2221-E-182-032, 104-2745- 8-182-001,
105-2221-E-182-043.
Author Contributions:
Prasan Kumar Sahoo, Sudhir Ranjan Pattanaik and Shih-Lin Wu conceived and designed
the protocol and experiments, analyzed the data, and wrote the paper. Sudhir Ranjan Pattanaik performed
the experiments.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
IEEE Standard. 802.15.4-2011: Wireless Medium Access Control (MAC) and Physical Layer (PHY) Specifications
for Low Rate Wireless Personal Area Networks (LR-WPANs); IEEE Standard: New York, NY, USA, 2011.
2.
IEEE Standard. 802.15.4e-2012: Wireless Medium Access Control (MAC) and Physical Layer (PHY) Specifications
for Low Rate Wireless Personal Area Networks (LR-WPANs); IEEE Standard: New York, NY, USA, 2012.
3.
Park, P.; Marco, P.D.; Soldati, P.; Fischione, C.; Johansson, K.H. A Generalized Markov Chain Model for
Effective Analysis of Slotted IEEE802.15.4. In Proceedings of the IEEE 6th International Conference on
Mobile Adhoc and Sensor Systems, Macau, China, 12–15 October 2009; pp. 130–139.
4.
Pollin, S.; Ergen, M.; Ergen, S.C.; Bougard, B.; Perre, L.V.D.; Moerman, I.; Bahai, A.; Varaiya, P.; Catthoor, F.
Performance analysis of slotted carrier sense IEEE802.15.4 medium access layer. IEEE Trans. Wirel. Commun.
2008,7, 3359–3371.
5.
Recas, J.; Khaled, N.; Barrio, A.A.D.; Hermida, R. Generic Markov model of the contention access period of
IEEE 802.15.4 MAC layer. Digit. Signal Process. 2014,33, 191–205.
Sensors 2017,17, 168 24 of 25
6.
DiMarco, P.; Fischione, C.; Santucci, F.; Johansson, K.H. Modeling IEEE 802.15.4 Networks over Fading
Channels. IEEE Trans. Wirel. Commun. 2014,13, 5366–5381.
7.
Guglielmo, D.D.; Restuccia, F.; Anastasi, G.; Conti, M.; Das, S. Accurate and Efficient Modeling of 802.15.4
Unslotted CSMA/CA through Event Chains Computation. IEEE Trans. Mob. Comput. 2016,15, 2954–2968.
8.
Yu, J.; Park, L.; Park, J.; Cho, S.; Keum, C. CoR-MAC: Contention over Reservation MAC Protocol for
Time-Critical Services in Wireless Body Area Sensor Networks. Sensors 2016,16, 656.
9.
Bhandari, S.; Moh, S. A Priority-Based Adaptive MAC Protocol for Wireless Body Area Networks. Sensors
2016,16, 401.
10.
Park, P.; Fischione, C.; Johansson, K.H. Performance Analysis of GTS Allocation in Beacon Enabled
IEEE802.15.4. In Proceedings of the 6th Annual IEEE Communications Society Conference on Sensor,
Mesh and AdHoc Communications and Networks, Rome, Italy, 22–26 June 2009; pp. 1–9.
11.
Zhan, Y.; Xia, Y.; Anwar, M. GTS size adaptation algorithm for IEEE 802.15.4 wireless networks. Ad Hoc Netw.
2016,37, 486–498.
12.
Chen, C.; Pei, Q.; Li, X. A GTS Allocation Scheme to Improve Multiple-Access Performance in Vehicular
Sensor Networks. IEEE Trans. Veh.Technol. 2016,65, 1549–1563.
13.
Hayashi, N.; Takai, S. GTS-based communication task scheduling for quantized output consensus over
IEEE 802.15.4 wireless networks. Automatica 2015,55, 6–11.
14.
Doudou, M.; Djenouri, D.; Badache, N.; Bouabdallah, A. Synchronous contention-based MAC protocols for
delay-sensitive wireless sensor networks: A review and taxonomy. J. Netw. Comput. Appl.
2014
,38, 172–184.
15.
Wagh, S.S.; More, A.; Kharote, P.R. Performance Evaluation of IEEE 802.15.4 Protocol Under Coexistence of
WiFi 802.11b. Procedia Comput. Sci. 2015,57, 745–751.
16.
Elshabrawy, T. Network throughput analysis of IEEE 802.15.4 enabled wireless sensor networks with FEC
coding under external interference. AEU Int. J. Electron. Commun. 2015,69, 1641–1649.
17.
Ghayvat, H.; Mukhopadhyay, S.C.; Gui, X. Issues and mitigation of interference, attenuation and direction of
arrival in IEEE 802.15.4/ZigBee to wireless sensors and networks based smart building. Measurement
2016
,
86, 209–226.
18.
Ma, X.; Trivedi, K.S. Reliability and performance of general two-dimensional broadcast wireless network.
Perform. Eval. 2016,95, 41–59.
19.
Chiplunkar, N.N.; Nayak, K.P.; Prasad, D. Performance Analysis of the Physical and Medium Access Control
Layer Parameters with Effect of Varying Transmission Power Using IEEE 802.15.4 Standard for Wireless
Body Sensor Networks. Procedia Comput. Sci. 2016,93, 639–646.
20.
Khan, Z.A.; Rasheed, M.B.; Javaid, N.; Robertson, B. Effect of Packet Inter-arrival Time on the Energy
Consumption of Beacon Enabled MAC Protocol for Body Area Networks. Procedia Comput. Sci.
2014
,32,
579–586.
21.
Anchora, L.; Capone, A.; Mighali, V.; Patrono, L.; Simone, F. A novel MAC scheduler to minimize the energy
consumption in a Wireless Sensor Network. Ad Hoc Netw. 2014,16, 88–104.
22.
Shuai, J.; Zou, W.; Zhou, Z. Priority Based Adaptive Timeslot Allocation Scheme for Wireless Body Area
Network. In Proceedings of the 13th International Symposium on Communications and Information
Technologies, Surat Thani, Thailand, 4–6 September 2013; pp. 609–614.
23.
Yin, D.; Lee, T.T. Throughput Stability and Energy Consumption of IEEE 802.15.4 Beacon Enabled
Mode. In Proceedings of the 21st Wireless and Optical Communications Conference, Kaohsiung, Taiwan,
19–21 April 2012; pp. 37–42.
24.
Lee, B.H.; Lai, R.L.; Wu, H.K.; Wong, C.M. Study on additional carrier sensing for IEEE 802.15.4 wireless
sensor networks. Sensors 2010,10, 6275–6289.
25.
Son, K.J.; Hong, S.H.; Moon, S.P.; Chang, T.G.; Cho, H. Segmentized Clear Channel Assessment for
IEEE 802.15.4 Networks. Sensors 2016,16, 815.
26.
Guennoun, M.; Khanafer, M.; Mouftah, H.T. Modeling of variable Clear Channel Assessment MAC protocol
for Wireless Sensor Networks. Comput. Commun. 2015,59, 67–83.
27.
Martalo, M.; Buratti, C.; Ferrari, G.; Verdone, R. Clustered IEEE 802.15.4 Sensor Networks with Data
Aggregation: Energy Consumption and Probability of Error. IEEE Wirel. Commun. Lett. 2013,2, 70–73.
28.
Pham, C. Communication performances of IEEE 802.15.4 wireless sensor motes for data-intensive
applications: A comparison of WaspMote, Arduino MEGA, TelosB, MicaZ and iMote2 for image surveillance.
J. Netw. Comput. Appl. 2014,46, 48–59.
Sensors 2017,17, 168 25 of 25
29.
Lee, J.; Jeong, W.C. Performance Analysis of IEEE 802.15.4e DSME MAC Protocol under WLAN Interference.
In Proceedings of the International Conference on ICT Convergence (ICTC), Jeju, Korea, 15–17 October 2012;
pp. 741–746.
30.
Juc, I.; Alphand, O.; Guizzetti, R.; Favre, M.; Duda, A. Energy Consumption and Performance of
IEEE 802.15.4e TSCH and DSME. In Proceedings of the IEEE Wireless Communications and Networking
Conference, Doha, Qatar, 3–6 April 2016; pp. 1–7.
31.
Wijetunge, S.; Gunawardana, U.; Liyanapathirana, R. Throughput Analysis of IEEE 802.15.4 MAC Protocol
in the Presence of Hidden Nodes. In Proceedings of the 11th International Symposium on Communications
and Information Technologies, Hangzhou, China, 12–14 October 2011; pp. 303–308.
32.
Wijetunge, S.; Gunawardana, U.; Liyanapathirana, R. Performance Analysis of IEEE 802.15.4 MAC Protocol
for WSNs in Burst Error Channels. In Proceedings of the 11th International Symposium on Communications
and Information Technologies, Hangzhou, China, 12–14 October 2011; pp. 286–291.
33.
Hao, P.; Liu, W.; Wang, J. Performance Evaluation of IEEE 802.15.4 MAC with Synchronization-Loss Based on
Discrete-Time Markov Chain. In Proceedings of the 14th IEEE International conference on Communication
Technology, PyeongChang, Korea, 19–22 February 2012; pp. 229–234.
34.
Martalo, M.; Busanelli, S.; Ferrari, G. Markov Chain-based performance analysis of multihop IEEE 802.15.4
wireless networks. Perform. Eval. 2009,66, 722–741.
35. OMNeT++ Homepage. Available online: http://www.omnetpp.org (accessed on 6 July 2016).
c
2017 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).