An Energy Conservation MAC Protocol in Wireless Sensor Networks

Abstract

Wireless sensor networks use battery-operated computing and sensing devices. Because of the limitation of battery power in
the sensor nodes, energy conservation is a crucial issue in wireless sensor networks. Consequently, there is much literature
presenting energy-efficient MAC protocols based on active/sleep duty cycle mechanisms to conserve energy. Convergecast is
a common communication pattern across many sensor network applications featuring data gathering from many different source
nodes to a single sink node. This leads to high data collision rates, high energy consumption, and low throughput near the
sink node. This paper proposes an efficient slot reservation MAC protocol to reduce energy consumption and to make transmission
more efficient in data gathering wireless sensor networks. The simulation results show that our protocol provides high throughput,
low delivery latency and low energy consumption compared to other methods.

Full text

Page 1

Wireless Pers Commun (2009) 48:261–276
DOI 10.1007/s11277-008-9521-2
An Energy Conservation MAC Protocol in Wireless
Sensor Networks
Yu-Chia Chang · Jang-Ping Sheu
Published online: 23 May 2008
© Springer Science+Business Media, LLC. 2008
Abstract
Becauseofthelimitationofbatterypowerinthesensornodes,energyconservationisacrucial
issue in wireless sensor networks. Consequently, there is much literature presenting energy-
efficient MAC protocols based on active/sleep duty cycle mechanisms to conserve energy.
Convergecast is a common communication pattern across many sensor network applications
featuring data gathering from many different source nodes to a single sink node. This leads
tohighdatacollisionrates,highenergyconsumption,andlowthroughputnearthesinknode.
This paper proposes an efficient slot reservation MAC protocol to reduce energy consump-
tion and to make transmission more efficient in data gathering wireless sensor networks. The
simulation results show that our protocol provides high throughput, low delivery latency and
low energy consumption compared to other methods.
Wireless sensor networks use battery-operated computing and sensing devices.
Keywords
Energy conservation · MAC protocol · Wireless sensor networks
1 Introduction
Wireless sensor network (WSN) is an emerging technology that is expected to be used in
a wide range of applications such as target tracking, environment monitoring, habitat sens-
ing, and home security [1–3]. Usually it is composed of a large number of battery-operated
distributed nodes, making energy conservation one of the most important issues in WSNs.
Energy conservation can be addressed at each layer of the network protocol stack, but our
focus in this paper is the Medium Access Control (MAC) layer. The role of a network is
Y.-C. Chang (B )
Department of Computer Science and Information Engineering, National Central University,
Jhongli 32054, Taiwan
e-mail: jimchang@axp1.csie.ncu.edu.tw
J.-P. Sheu
Department of Computer Science, National Tsing Hua University, Hsinchu 30013, Taiwan
e-mail: sheujp@cs.nthu.edu.tw
123

Page 2

262 Y.-C. Chang, J.-P. Sheu
to ensure that data can be delivered as expected. A general rule for achieving a predictable
operation is to reduce as much as possible the complexity of the applications and their ser-
vices. Thus, optimizing the communication performance of the sensor networks becomes
very important. A good MAC protocol must always consider the following attributes: energy
efficiency, scalability, fairness, latency, and throughput.
The status of a radio transmitter consists of four possible operations with different power
levels: transmitting, receiving, listening, and sleeping. Typically, the power consumption of
listening is the same as that of receiving. The transmitting power consumption depends on
the transmission power level. From the data sheet of the MICAz Mote sensors radio chip
[4,5], the power levels are: 17.4mA for transmission with maximum power level, 8.5mA for
transmission with minimum power level, 18.8mA for receiving data, 426µA for idle mode
(crystal oscillator and voltage regulator on), 20µA for power down (only voltage regulator
on). In practice, turning the radio into sleep mode is the most power conserving method. If
nodes can turn off the radio when there is no data to send or receive and wake up at the right
time to transmit or receive, the available energy of a battery can be used in an optimal way.
This paper presents an energy conservation MAC protocol for WSNs. Our protocol com-
bines contention-based, scheduling-based, and reserving-based schemes to achieve energy
efficiency, reduce transmission delay, and decrease collision probability of data transmis-
sion. Only a rough time synchronization is needed in our protocol. All nodes must record all
its neighbors’ wake-up schedules. While transmitting, nodes use a reserving-bit to inform
the receiver that they still want to transmit data. Thus, the receiver can reserve slots for
those senders in order to decrease the number of sensor nodes that want to contend the same
transmission slot. The main contribution of this work is to conserve energy by reducing the
collision probability of nodes competing for the same time slot. Our protocol improves net-
work throughput and reduces transmission delay while at the same time conserving energy.
This paper is organized as follows. Section2 reviews relevant works in the literature.
Section3 describes our proposed protocol. Section4 presents the simulation results. Finally,
we draw our conclusions in Sect.5.
2 Related Work
Over the past few years several MAC protocols have been developed for WSNs. They can be
categorized into centralized and decentralized MAC protocols. Most of the centralized pro-
tocolsoperateasacluster-basedscheme.Inacluster-basedschemethebasestationorcluster
header will allocate time slots to each member for creating a collision free operation within
the cluster. Therefore, accurate time synchronization protocol is essential in centralized pro-
tocols. The other type of protocol is the decentralized-based MAC protocol. These protocols
can be divided into scheduled and random access schemes. In the scheduled schemes, all
nodes need to periodically broadcast their wake-up schedule and maintain their neighbors’
schedule information. The nodes are then allowed to transmit data during the active periods
of the receivers and save energy according to their own schedules. Only a rough time syn-
chronization is required in the scheduled schemes, such as S-MAC, T-MAC, P-MAC, and
D-MAC.
S-MAC and T-MAC are two well-known MAC protocols in WSNs. In S-MAC [6] the
four major sources of energy waste are: collision, overhearing, control packet overhead, and
idle listening. Therefore, S-MAC tries to reduce waste by putting sensor nodes into a peri-
odical sleeping mode at a low and fixed duty cycle. T-MAC [7] improves on S-MAC by
using an adaptive duty cycle. Sensor nodes go to sleep when there is no activity at time
123

Page 3

An Energy Conservation MAC Protocol in Wireless Sensor Networks 263
Fig. 1 The schedules of S-MAC and T-MAC protocols
TA= (C + R + T) × 1.5), where C is the length of the contention interval, R is the length
of an RTS packet, and T is a short time between the end of the RTS packet and the beginning
of the CTS packet. Figure1 shows the difference between S-MAC and T-MAC protocols.
T-MAC provides a better throughput than S-MAC under variable traffic. When the traffic
load is heavy, the throughput of T-MAC performs more efficiently than S-MAC. However,
both the throughputs of S-MAC and T-MAC are influenced by packet collisions. Thus, an
efficient collision avoidance method is needed to decrease the waste of battery energy of
sensor nodes and to improve the overall network performance.
P-MAC [8] is a time-slotted and pattern-based scheduling protocol. Each sensor node
determines its sleep/wake-up schedule based on its own traffic and the traffic patterns of its
neighbors. In P-MAC, time is divided into super time frames (STF) which have two sub-
frames: PRTF and PETF. PRTF is a data transmission sub-frame. Each sensor node decides
to stay awake for transmitting and receiving data or going to sleep for saving energy based
upon the patterns of its neighboring nodes. In the PETF sub-frame, nodes exchange their
traffic patterns with their neighbors. The purpose of this pattern-exchange is to ensure that
the schedules of the sensor nodes are adapted to the current traffic load. Although P-MAC
can conserve energy by using the traffic-pattern information, it needs to spend extra energy
and bandwidth to maintain the pattern information.
D-MAC[9]presentsacontinuouspacketforwardingschemealongthedatagatheringtree
to solve the data forwarding interruption problem. A node skews its wake-up schedule dt
ahead of the schedule of the sink (d is the depth of the tree and t is the period of sending or
receiving a packet). D-MAC staggers the active/sleep schedule of nodes in the data gathering
tree. By allowing packets to be forwarded on the multi-hop path, D-MAC can decrease the
transmissiondelay.D-MACisanon-flexibilityprotocol.Ifthenetworktopologyischanging,
allthenodesneedtore-constructtheiractive/sleepschedules.Anothercategoryofdecentral-
ized MAC protocol is the random-access based scheme. In this classification, nodes contend
the channel “on-demand” without any schedule or synchronization. Wise-MAC, STEM, and
B-MAC are examples of this category.
Wise-MAC [10] is anunslotted MAC protocol.Eachnodeonlymaintains the sleep/wake-
uptimesofitsneighbors.Whenasensornodehaspacketstosend,thenodewillholdthedata
packetsuntilthereceiverisactive.Thenodefirstsendsthedurationofthepreambletoinform
the receiver, then transmits the data and finally receives an ACK from the receiver. In Wise-
MAC, each node will periodically wake up for a short interval. If there is no message from
other nodes, the node will go to sleep to conserve energy. This is a simple and easy protocol
for saving a node’s energy requirements, however, the transmission latency will increase.
STEM [11] is a two-radio architecture. One radio is for transmitting data and the other
one is for waking up sensor nodes. Sensor nodes can go to sleep until communication is
desired. STEM can conserve energy but it requires a higher hardware cost for the two radio
transmitters.
123

Page 4

264 Y.-C. Chang, J.-P. Sheu
B-MAC [12] uses in-channel signaling to wake up the destination node. It operates by
periodicallylisteningtothechannelforactivity.Nodeswillturnontheirreceiversifachannel
issensedtobebusy,andtheyturnoffafteradatapacketisreceivedorafteracertaintimeout.
While transmitting, a sender will send a long preamble time period to inform the destination
to receive a data packet. However, a receiver needs to wake up first and listen to the channel
until a data packet is received. Thus, the receiver spends time on idle listening which results
in the wasting of energy.
In the literature, many MAC protocols use a periodic sleep/wake-up schedule technique
to reduce energy waste from idle listening and overhearing. However, collision problems
occur if the convergecast [13] routing (many-to-one) or many-to-many communication is
not well-solved. If there are many senders who want to all transmit data to a destination
simultaneously in its active interval, then these senders must share the same communication
channel with each other. Most of these MAC protocols use the CSMA/CA and RTS/CTS
schemetoreducetheeffectofcollision.However,ifthetrafficloadisheavy,thecollisionrate
is high. This situation wastes the energy of the sensor nodes. Therefore, we propose a novel
MACprotocoltoconserveenergyandatthesametimeimprovethenetworkthroughput.The
details of this protocol are described in the next section.
3 ESR-MAC Protocol Design
As mentioned previously, the goal of our protocol is to reduce the collision problem and
to improve throughput without reducing the energy efficiency. Many researches indicated
that the most significant source of energy waste is idle listening. In order to decrease energy
waste from idle listening, nodes will periodically wake up and listen to the channel. If nodes
have packets to send, they have to share the time slot with other competitors. Although some
protocols present CSMA/CA-liked schemes to lessen the effect from this congestion, the
probability of a collision taking place will still increase during periods of heavy traffic. How-
ever, traffic density may vary both in time and location for different applications. When we
look at a data gathering tree, the traffic pattern is like a ripple spreading from the sink node.
The closer to the sink, the heavier the traffic. Nodes need to report their sensed data and
forward their data packet to other nodes. As a result the region near the sink node becomes
a traffic bottleneck and results in a significant amount of collisions. Previous sleep/wake
up MAC protocols could not solve this contention and collision problem. Therefore, this
contention problem will increase the delay and collision problem and at the same time waste
a significant amount of energy. In this paper, we propose an Efficient Slot Reservation MAC
(ESR-MAC)protocolwhichcanreducethenumberofnodescontendingthiscontentionslot.
We present a slot reservation scheme that will lessen the effect from both contention and
collision. ESR-MAC can decrease the transmission delay and improve throughput without
reducing the energy efficiency.
3.1 Network and Application Assumptions
Sensornetworksarecomposedofmanysmallnodeswhicharedeployedinanadhocmanner,
usingashortrangeandmulti-hopcommunicationmechanisms.Nodesinthenetworkarebat-
tery-operated and have no mobility. We implement a rough time synchronization scheme as
per [14,15]. Nodes in WSNs do not need a high overhead for global synchronization. Each
node only records its neighbors’ sleep/wake-up times. The neighbors’ time offset can be
123

Page 5

An Energy Conservation MAC Protocol in Wireless Sensor Networks 265
achieved by collecting the “hello” messages which are periodically broadcast by each node.
With the neighbors’ time offset, senders can awaken and transmit data packets to receivers at
the right time. Each node must also maintain its own sleep/wake-up schedule. When there is
no data to be transmitted, sensor nodes can follow the sleep/wake-up schedule and turn the
radio into sleep mode and save energy.
We used the surveillance or environment measurement as an example of applications. For
such applications, nodes need to periodically report their collected data to a sink. However,
nomatterhowfastthereportingrate,theapplicationwillcausecontentionandcollisionprob-
lems. This has a strong influence on our protocol design. The main idea of the ESR-MAC
protocol is to reduce the waste of energy from collisions and to improve the overall network
performance.
3.2 Slots Reservation Scheme
Our proposed protocol, ESR-MAC, is a rough slot-based scheme. We allow each node to
transmit only one data packet per slot. A duty cycle consists of one contention slot, one con-
trol slot, and at most n reservation slots. For a sensor node, the duty cycle is determined at
the initial state and equals to (1+1+n) slots. However, different nodes can have a different
number of reservation slots n. In the ESR-MAC protocol, nodes periodically wake up at a
contentionslotinordertoreceivedatapackets.Sensornodescontendthecontentionslotwith
each other. In order to avoid collisions, the CSMA/CA-like protocol is used. If a node misses
or fails to access the contention slot, it needs to wait until next duty cycle. If it obtains access
to the contention slot, it can then successfully transmit a data packet in the contention slot.
In our protocol, we permit the node (winner of accessing the contention slot) to piggyback
a reservation bit in the data frame to notify that it still has data to transmit or not in the
next duty cycle. Consequently, the receiver can schedule the reservation slot according to the
reservation bit in the data frame. The schedule will then be broadcast at the control slot to all
the neighbors. This process guarantees that nodes that are listed in the schedule can transmit
data at the reservation slot in the next duty cycle. This method reduces the number of nodes
at the contention slot and decreases the collision probability. If the reservation slots are not
reserved for any senders, then the node turns the radio into sleep mode to save energy and
wakes up until the next cycle.
The framework of ESR-MAC is shown in Fig.2. There may be no reservation slots when
the traffic is light or there may be no sleep slots if the traffic is heavy. According to the spec-
ification of MICAz [4], we can precisely calculate the length of the contention, control, and
Fig. 2 The framework of S-MAC, T-MAC and ESR-MAC protocols
123

Page 6

266 Y.-C. Chang, J.-P. Sheu
Fig. 3 An example data
gathering tree
reservation slots. For the contention slot, a random back-off process is first operated in order
toavoidcollision.Then,thewinnerofcontenderscantransmitdatapacketsimmediately.The
lengthofthecontentionslotincludesarandomback-offperiodandadatatransmissionperiod.
OneunitofCSMAback-offperiodisequalto0.32ms,andthemaximumcontentionwindow
size is set to 2kwhere k is an integer. Therefore, the length of a random back-off period is
2.56ms if k = 3(0.32ms × 2k). Because the data rate of a CC2420 radio chip is 250kbps,
the data transmission period is 1.408ms if the packet size is 44bytes (44bytes×8/250ms).
Thus, the length of a contention slot is 3.968ms with a fixed contention window size
of 8, and a maximum packet size of 44bytes, in a MICAz device. For a control slot, the
first two bytes are used for synchronization and the next 2n bytes are the schedule list which
indicates the access right of n reservation slots at the next duty cycle (size of node ID is
2bytes). So, the length of a control slot is 2(1 + n) × 8/250ms. If n = 10, the length of
a control slot is 0.704ms. For the reservation slots, the sensor nodes that were listed in the
broadcast schedule can transmit data packets in sequence. Because the reservation scheme is
acollisionfreemethod,senderscantransmitdatapacketsdirectlywithoutarandomback-off
mechanism. The length of a reservation slot is 1.408ms. Thus, a duty cycle of a node is equal
to 3.968ms+0.704ms+14.08ms=18.752ms, where CW=8, the packet size=44bytes,
and n =10.
In our slot reservation scheme, if there is no traffic in the network, then nodes must wake
upatthecontentionslotforreceivingadatapacketandatthecontrolslotforbroadcastingthe
schedule of the reservation slots. Contrary to S-MAC and T-MAC, our protocol is a slotted
MAC protocol. Nodes can handle their traffic more efficiently based on the slotted architec-
ture. With the time slotted architecture, receiver can reserve slots to the senders and perform
a collision free transmission. Therefore, our protocol can improve the network throughput
and reduce transmission delays without impacting on the conservation of energy.
Inthe following, wewill presentanexample ofour proposedslots reservation scheme.As
showninFig.3,adatagatheringtreeisalreadyconstructedandallthenodeswillperiodically
report their sensing data to sink node S.
First, each node determines its active/sleep schedule and exchanges its schedule infor-
mation with its neighbors. After collecting its neighbors’ active/sleep schedules, each node
begins its scheduled duty cycle. If the contention slot period of a node overlaps with the
contention slot period of its parent, it may decrease the performance of our ESR-MAC. For
example,assumethecontentionslotperiodofnode Aoverlapswithitsparentnode S inFig.3.
Node A cannot receive data packets from node C, D, or E at its own contention slot if it tries
to transmit data packets to node S during its contention slot. To reduce this conflict, a node
cannot determine its active/sleep schedule until it receives the active/sleep schedule from its
parent. Thus, each node can choose a proper active/sleep schedule such that its control slot
period does not overlap with that of its parent. Since a node has only one transceiver, we
123

Page 7

An Energy Conservation MAC Protocol in Wireless Sensor Networks267
Fig. 4 An example of ESR-MAC protocols
assume that packet sending has a higher priority than packet receiving. This assumption is
considered about the validity of a packet.
In this example, node A must handle its own traffic pattern and forward three data packets
fromitschildrentosinknode S.First,wediscussthereceivingpartofnode A.Node A wakes
up at its contention slot and its three children, C, D, and E, will contend with each other at
this slot. A random back-off scheme is used to avoid collision. Assuming that node C wins
the access to the contention slot, then it will transmit data to node A. We assume that nodes
always have packets to send in the next duty cycle, so the reserved bit in the data frame is set
to 1 as shown in Fig.4.
While receiving a packet, node A will preserve a reservation slot to node C for the next
transmission.Inthefirstrun,therearenoreservationslotsinnode A’sschedule,insteadthere
are sleep slots to save more energy, as shown in Fig.5a. Nodes D and E fail to contend the
slot, so they need to contend for being able to transmit in the next run. After the sleeping
slots, node A wakes up at its contention slot again. This time only nodes D and E participate
in the contention process. We assume that node E wins this contention. Node E transmits
a data packet at the contention slot and its reserved bit in the data frame is set to 1. In the
control slot, node A will broadcast a message to notify the scheduled list of the reservation
slots. The information in this message contains a reservation slot for node C. In this duty
cycle, node C will transmit at the reservation slot which was preserved in the last cycle,
and the reserved bit in the data frame is still set to 1 to reserve for the next transmission, as
shown in Fig.5b. In the third run, only node D will contend the contention slot. After the
contention slot, node A broadcasts its reservation schedule to its neighbors. Nodes C and E
are included in the schedule. Then, nodes C and E will transmit at the reservation slots as
shown in Fig.5c.
Next, we describe the transmission scenario of node A. Node A needs to forward the
packets of its children and its own data to sink node S. Node A will wake up while node
S’s contention slot is coming up and it contends the slot with node B. During the time
at the contention slot, node A operates a random back-off scheme to access the conten-
tion channel. If it wins the channel, it can immediately transmit a data packet, otherwise
it contends the channel in the next duty cycle. The proposed protocol is summarized as
follows.
123

Page 8

268Y.-C. Chang, J.-P. Sheu
Fig. 5 The sleep/wake up schedule of node A
ESR-MAC Protocol:
Initial:
Each node determines its own duty cycle and broadcasts it to its neighbors if it receives
the active/sleep schedule from its parent.
Begin:
(1)Whenanodehaspacketstosenditwaitsuntilthereceiver’scontentionslothasarrived.
Then, the node contends the slot with other senders. If a node has more than one packet
to transmit to the receiver, it will set the reservation bit in the data frame to preserve a
reservation slot in the next duty cycle of the receiver. If a node is listed in the schedule list
of a receiver, then that node will transmit its data packet in the indicated reservation slot.
(2) When a node has no data to send, it will execute its own active/sleep schedule as
follows.
In the contention slot: Listen to the channel and receive data packets.
In the control slot: The node broadcasts its time for time synchronization. According to
the reservation bit, a node preserves the reservation slots to the senders and broadcasts the
scheduled list to inform the senders.
In the reservation slot: If the reservation slot is reserved by the senders, the node wakes
up to receive data packets, otherwise it goes into sleep mode in order to save energy.
In the sleeping slot: The node turns off the radio in order to save energy.
4 Simulation Results
Inoursimulations,wecomparetheperformanceofourproposedprotocolwithtwoschedule-
based MAC protocols S-MAC and T-MAC. We use GloMoSim [16], a network simulator, to
simulate the nodes’ behavior for each MAC protocol. We present two network topologies for
investigating the performance among the three protocols. The first scenario is a single-hop
123

Page 9

An Energy Conservation MAC Protocol in Wireless Sensor Networks269
Table 1 The energy consumption defined by the CC2420 radio chip
Term Description UnitValue
Packet (Data)
Packet (RTSorCTS)
Packet (ACK)
Power (Rx)
Power (Tx)
Size of a data packet
Size of RTS or CTS control packets
Size of an acknowledge packet
Power required for receiving a packet
Power required for transmitting a packet
byte
byte
byte
mW
mW
44
17
11
62.1
28.1 (−25dbm)
57.4 (0dbm)
62.1
1.41
250
Power (idle)
Power (sleep)
Txrate
Power required for idle listening
Power required for sleep mode
Transmit bit rate
mW
mW
bps
networkenvironmentinwhichs sendersreporttheirsenseddatatothesinkperiodically.This
case displays the affect of the traffic bottleneck. In the second scenario we place 100 nodes
in a 100m×100m 2-dimensional area in a random distribution with a transmission range
set to 20m. In this network, there is a sink node, and all the sensor nodes will report data to
the sink node. Since the nodes have no mobility, we chose the Bellman-ford algorithm [17]
as the routing protocol and a CSMA-like scheme is used to avoid collision in the simulation.
Each simulation lasts 1000s and each result is obtained from the average of 100 simulation
results. The type of traffic loads are CBR and set to 1, 10, 20, 30, 40 and 50 packets per sec-
ond. As a result, the energy consumption of a sensor node includes receiving, transmitting,
idle listening and sleeping states. Table1 lists the constants and variables, respectively, for
the calculation of the energy consumption in the radio chip of the CC2420 data sheet [4] and
the documents of Mica Mote [5].
4.1 Single-hop Network Environment
In this subsection, three performance metrics are investigated. The successful rate is defined
as the total received packets over the total transmitted packets. The energy consumption
means the average energy consumption among sensor nodes. The throughput is defined as
thetotalnumberofbitsreceivedpersecondatthesinknode.First,weincreasethenumberof
senders and the traffic load to demonstrate the impact on the collision rates and the success
rates. In our ESR-MAC protocol, we set a fixed contention window size, CW=8, and the
maximum number of reservation slots, n = 10. Then the maximum duty cycle of a node is
18.752ms. For a fair comparison, we set the same length of duty cycles for both S-MAC and
T-MAC protocols. In addition, we take S-MAC with 50% active period as the competition
and the value of TAin T-MAC is set to 4.658ms [7].
Figures6 and 7 show the simulation results of the success rate of S-MAC, T-MAC, and
ESR-MACprotocolsindifferentdensityenvironments.Theresultsshowthatwhilethenum-
berofsenders,s,increasesfrom5to20,thecollisionprobabilityincreasesforbothprotocols.
For the traffic load, the success rate decreases as the traffic load increases. Both the S-MAC
andT-MACprotocolswereoperatedwithaCSMA-likescheme.Theonlydifferencebetween
these two protocols is that T-MAC uses a short duration TAto reduce the effect of idle lis-
tening. This process can save more energy but cannot reduce the collision probability if the
number of senders is increased or if the traffic load becomes heavy. Consequently, T-MAC
provides a higher success rate than S-MAC because T-MAC can adjust its nodes’ active
period dynamically to receive more packets when the traffic is heavy. In our simulations,
123

Page 10

270 Y.-C. Chang, J.-P. Sheu
Successful Rate
0.7
0.75
0.8
0.85
0.9
0.95
1
1020 3040 50
CBR Traffic(packets per second)
Successful Rate (%)
ESR (s=5)
SMAC (s=5)
TMAC (s=5)
ESR (s=10)
SMAC (s=10)
TMAC (s=10)
Fig. 6 The success rate of the S-MAC, T-MAC, and ESR-MAC protocols with s = 5 and s = 10
Successful Rate
0.5
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
1020 304050
CBR Traffic (packets per secnod)
Successful Rate (%)
ESR (s=15)
SMAC (s=15)
TMAC (s=15)
ESR (s=20)
SMAC (s=20)
TMAC (s=20)
Fig. 7 The success rate of the S-MAC, T-MAC and ESR-MAC protocols with s = 15 and s = 20
the worst success rates of S-MAC and T-MAC are 53.34% and 62.69% when the number of
senders is 20 and the traffic load is 50 packets per second.
As the number of senders is equal to 5, both the ESR-MAC and T-MAC have the same
successful rate. When the traffic load is heavy or when the number of senders is increased,
the ESR-MAC protocol has a better successful rate than the S-MAC and T-MAC protocols.
Our reservation scheme can reduce the quantity of the contenders and consequently decrease
the collision rate. Although the reservation slots are used to avoid collision during free trans-
mission, collisions still happen at the contention slots. Therefore, the successful rate of our
protocol decreases when the traffic load becomes heavy or when the number of senders
increases. The worst case scenario for the successful rate of our protocol is 70.59% when the
number of senders is equal to 20 and the traffic load is 50 packets per second. This result is
better than the S-MAC and T-MAC protocols.
In the following, we compare the energy consumption of the three protocols. The energy
consumption of each node is the summation of energy consumptions of a node in trans-
mission, receiving, idle, and sleeping modes. In our simulations, the result of S-MAC is
affected by the ratio of the active/sleep periods. The longer the sleep period is, the larger
the energy saving. In Figs.8 and 9, the energy consumption of the S-MAC protocol is not
123

Page 11

An Energy Conservation MAC Protocol in Wireless Sensor Networks 271
Energy Consumptions
0
20
40
60
80
100
120
140
10 2030 4050
CBR Traffic (packets per second)
Energy Consumption (J)
ERS (s=5)
SMAC (s=5)
TMAC (s=5)
ESR (s=10)
SMAC (s=10)
TMAC (s=10)
Fig. 8 The energy consumption of the S-MAC, T-MAC, and ESR-MAC protocols with s = 5 and s = 10
Energy Consumptions
0
20
40
60
80
100
120
140
1020 30 4050
CBR Traffic (packets per second)
Energy Consumption (J)
ESR (s=15)
SMAC (s=15)
TMAC (s=15)
ESR (s=20)
SMAC (s=20)
TMAC (s=20)
Fig. 9 The energy consumption of the S-MAC, T-MAC, and ESR-MAC protocols with s = 15 and s = 20
affected by the traffic or by the number of senders. The results remain stable because of the
fixed active/sleep periods. Nodes wake up to receive during their active period and go into
sleep during their sleep period. Although no traffic might have arrived, nodes still wake up
to listen to the channel. This idle listening wastes energy and does not improve the network
throughput.
T-MAC can dynamically adjust the active/sleep periods. As a result, the throughput of
T-MAC is better than that of S-MAC. But, the energy consumption of T-MAC increases
when the traffic becomes heavyor the number of senders increases. The energy consumption
of our ESR-MAC protocol also increases in the same scenario. In Figs.8 and 9, the energy
consumptions increase if the traffic load becomes heavy. However, ESR-MAC has better
energy consumption than T-MAC. ESR-MAC uses the reservation scheme to cope with the
heavy traffic or the increase in senders. The reservation slots of the ESR-MAC protocol can
efficiently reduce the number of contenders, thereby decreasing the collision probability. In
addition,thereservationschemeprovidescollisionfreeslots,guaranteeingdatatransmission
and improving throughput. An investigation of the T-MAC protocol shows that the energy
consumption of T-MAC rapidly achieves saturation when the number of senders equals to
20. The increase in contenders leads to a high collision rate, with the result that nodes cannot
123

Page 12

272Y.-C. Chang, J.-P. Sheu
Throughput
0
20
40
60
80
100
120
140
160
1020 304050
CBR Traffic (packets per second)
Throughput (Kbps)
ESR (s=5)
SMAC (s=5)
TMAC (s=5)
ESR (s=10)
SMAC (s=10)
TMAC (s=10)
Fig. 10 The throughput of the S-MAC, T-MAC, and ESR-MAC protocols with s = 5 and s = 10
Throughput
0
20
40
60
80
100
120
140
160
102030 4050
CBR Traffic (packets per second)
Throughput (Kbps)
ESR (s=15)
SMAC (s=15)
TMAC (s=15)
ESR (s=20)
SMAC (s=20)
TMAC (s=20)
Fig. 11 The throughput of the S-MAC, T-MAC and ESR-MAC protocols with s = 15 and s = 20
correctly and reliably receive data packets and waste energy when they receive error packets.
Moreover,aheavy-trafficloadandalargenumberofsendersmakesthatT-MACcannotenter
the sleeping mode to save energy. Therefore, T-MAC cannot improve the collision problem
when the traffic is heavy or when the number of senders is large.
Inthefollowingsimulations,weshowthethroughputofS-MAC,T-MAC,andESR-MAC
in Figs.10 and 11, respectively. Because of the similar successful rates of ESR-MAC and
T-MAC protocols, ESR-MAC also shows the similar throughput as T-MAC protocol if the
numberofsendersisequalto5.Whenthetrafficloadisheavyorwhenthenumberofsenders
is increased, ESR-MAC has a higher throughput than S-MAC and T-MAC. Because of the
reservation scheme, ESR-MAC can handle the traffic-bottleneck problem and can support a
highnetworkthroughput.ThethroughputofESR-MACdependsonthelengthofthereserva-
tion slots. In the light-traffic scenario, the reservation slots will become sleep slots. However,
the sensor nodes that lose at the contention slot must spend more time to access the channel
at the next contention slot. In the heavy-traffic pattern, we will reserve more reservation slots
than under a light-traffic load and can thereby improve the network throughput.
123

Page 13

An Energy Conservation MAC Protocol in Wireless Sensor Networks273
4.2 The Multi-hop Network Environment
In a multi-hop network environment, all the nodes periodically report data to the designated
sink node. One hundred sensor nodes with 20meters of transmission range and a sink node
are randomly deployed in an area 100meters square. We consider two performance metrics:
the transmission delay which means the average transmission delay per hop and the energy
utilization which is defined as the average energy cost per byte.
The simulation of the transmission delay is shown in Fig.12. The transmission delay is
definedasthedurationoftheattemptsbyanodetotransmitapacketuntilthepacketsuccess-
fullyreachestothereceiver.Theaveragetransmissiondelayincreaseswiththeincreaseofthe
trafficloadofthethreeMACprotocols.SinceS-MAChasalongeractiveperiodthanT-MAC
and ESR-MAC, the senders have a higher probability to transmit packets immediately. As a
result, S-MAC has the best performance in a light-traffic environment. With T-MAC, light
traffic results in receivers quickly entering the sleep mode after a duration TA. This results
in the transmission delay of T-MAC being longer than that of S-MAC under a light-traffic
load. In ESR-MAC, if the traffic load is light, we have small reservation slots with a sleeping
period that is longer than the active period. So, the waiting time of the ESR-MAC protocol
is longer than that of the S-MAC protocol. However, our ESR-MAC protocol provides a
reservation scheme to reduce the collision probability and to improve the transmission delay
of continuous data packets. Therefore, the transmission delay of ESR-MAC increases only
slightly as the traffic load increases. In S-MAC, the fixed active/sleep periods have a poor
performance in heavy traffic. The transmission delay of S-MAC increases when the traffic
load increases. Although T-MAC proposed a dynamic active/sleep periods that adapts to
the traffic variation, serious collisions remain and still lead to longer transmission delays.
The average transmission delays of S-MAC, T-MAC and ESR-MAC are 4.6, 9.2 and 9.1ms,
respectively at light-traffic load (10 packets per second). However, the average transmission
delays of S-MAC, T-MAC and ESR-MAC are 35.9ms, 32ms and 18.6ms, respectively, in
heavy-traffic load (40 packets per second).
Figure13 shows the results of the transmission delay by hop count. We simulate both
light-traffic and heavy-traffic scenarios denoted as ESR-10 and ESR-40, respectively. The
delay time increases with the increase in the hop count. S-MAC has the lowest transmission
delay in a light-traffic environment. A longer active period makes the S-MAC transmission
Transmission Delay
0
5
10
15
20
25
30
35
40
1020
CBR Traffic(packet per second)
304050
Delay Time (ms)
ESR
S-MAC
T-MAC
Fig. 12 Te average transmission delay per hop
123

Page 14

274Y.-C. Chang, J.-P. Sheu
Transmission Delay vs. Hop Count
0
100
200
300
400
500
600
1234
Hop Count
Transmission Delay (ms)
ESR-10
ESR-40
SMAC-10
SMAC-40
TMAC-10
TMAC-40
Fig. 13 The relationship between transmission delay and hop count
Energy Utilization
0
5
10
15
20
25
30
35
40
45
10 20 30 4050
CBR Traffic (packets per second)
Energy Cost (J)
ESR
S-MAC
T-MAC
Fig. 14 The average energy consumptions
faster, but it costs more energy for each packet transmission. Under heavy-traffic load, the
ESR-MAC protocol can reserve slots for continuous transmissions. As a result, the delay
time of ESR-MAC increases only slowly with the increase in traffic load. The transmission
delays under light-traffic load with four hops are 88ms for S-MAC, 132ms for T-MAC, and
117ms for ESR-MAC, respectively. Under heavy-traffic load with the same hop count, the
delaytimesofS-MAC,T-MAC,andESR-MACare571ms,449msand255ms,respectively.
Here, we show the total energy utilization of the S-MAC, T-MAC, and ESR-MAC
protocols. In our simulation, each node periodically reports its sensing data to the sink
node. While sink node successfully receives 50,000 packets, we terminate the simulation
and calculate the total energy consumption of the network. In Fig.14, all protocols have the
worst energy utilization under light-traffic loads because most of the energy is wasted in idle
listening. Since ESR-MAC and T-MAC can save energy by turning off the radio if there is no
traffic, their energy costs are better than that of S-MAC. When the traffic load is greater than
30packetspersecond,boththeS-MACandT-MACprotocolsreducetheirenergyutilization
123

Page 15

An Energy Conservation MAC Protocol in Wireless Sensor Networks275
since high collision rates reduce their throughputs. However, our ESR-MAC is not affected
by the increasing in traffic load because our protocol allows senders to transmit continuous
packets at the reservation slots. Therefore, our ESR-MAC protocol has the lowest energy
consumption since the reservation scheme allows for a more efficient packets transmission
than the S-MAC and T-MAC protocols.
5 Conclusion
This paper presents a novel MAC protocol for WSNs. Our protocol combines contention-
based,scheduling-based,andreserving-basedschemestoachieveenergyefficiency,improve
network throughput, and decrease collision probability. Based on the reservation bit in the
data packet, sensor nodes can preserve collision free slots for continuous transmissions.
Therefore, the ESR-MAC protocol is an efficient method for managing the battery energy
for sensor nodes. When traffic is light, ESR-MAC can save additional energy by switching
the radio to sleeping mode. In a heavy-traffic scenario, our proposed protocol can reserve
collision free slots for continuous transmissions. Thus, collision probability is decreased and
network throughput is improved.
Inthesingle-hopsimulation,ESR-MAChasahighersuccessrateandahigherthroughput
than either S-MAC or T-MAC. In the multi-hop environment, S-MAC has a shorter trans-
mission delay than T-MAC and ESR-MAC under a light-traffic load scenario. When the
traffic load is higher than 30 packets per second, ESR-MAC has the smaller transmission
delay compared to S-MAC and T-MAC. The transmission delay increases as the hop count
increases. S-MAC has the lowest transmission delay at light-traffic load and ESR-MAC has
thelowestdelayunderaheavy-trafficload.Thesimulationresultsalsoshowthattheproposed
ESR-MAC protocol has lower energy consumption than the S-MAC and T-MAC protocols.
Toconcludethesimulationresults,theESR-MACprotocolprovidesahigherthroughputand
has lower energy consumption compared to the S-MAC and T-MAC protocols.
References
1. Ahmed, A. A., Shi, H., & Shang, Y. (2003) A survey on network protocols for wireless sensor net-
works.InProceedings ofInternationalConferenceonInformationTechnology:Research andEducation,
pp. 301–305, August 2003.
2. Vieira, M. A. M., Coelho, C. N. Jr., da Silva, D. C. Jr., & da Mata, J. M. (2003). Survey on wireless
sensor network devices. In Proceedings of IEEE International Conference on Emerging Technologies
and Factory Automation, Vol. 1, pp. 537–544, September 2003.
3. Demirkol, I., Ersoy, C., & Alagoz, F. (2006). MAC protocols for wireless sensor networks: A survey.
IEEE Communications Magazine, 44, 115–121.
4. Chipcon, A. S. (2004). SmartRF CC2420 PRELIMINARY Datasheet, revision 1.2, February.
5. http://www.tinyos.net/.
6. Ye, W., Heidemann, J., & Estrin, D. (2002). An energy-efficient MAC protocol for wireless sensor net-
works. In Proceedings of INFOCOM, Vol. 3, pp. 1567–1576, June 2002.
7. Dam, T. V., & Langendoen, K. (2003). An adaptive energy-efficient MAC protocol for wireless sensor
networks. In Proceedings of the 1st International Conference on Embedded Networked Sensor Systems,
pp. 171–180, November 2003.
8. Zheng, T., Radhakrishnan, S., & Sarangan, V. (2005). PMAC: An adaptive energy efficient MAC proto-
col for wireless sensor networks. In Proceedings of the 19th IEEE International Parallel and Distributed
Processing Symposium, 8 p, April 2005.
9. Lu, G., Krishnamachari, B., & Raghavendra, C. S. (2004). An adaptive energy efficient and low-latency
MAC for data gathering in wireless sensor networks. In Proceedings of the 18th IEEE International
Parallel and Distributed Processing Symposium, 224 pp, April 2004.
123