Int. J. Communications, Network and System Sciences, 2009, 2, 695-703
doi:10.4236/ijcns.2009.28080 Published Online November 2009 (http://www.SciRP.org/journal/ijcns/).
A Comparative Study of Medium Access Control Protocols
for Wireless Sensor Networks
Meghan GUNN, Simon G. M. KOO
Department of Mathematics and Computer Science, University of San Diego, San Diego, USA
Received July 7, 2009; revised August 12, 2009; accepted September 21, 2009
Abstract
One of the major constraints in wireless sensor networks (WSNs) is power consumption. In recent years, a
lot of efforts have been put into the design of medium access control (MAC) protocols for WSN, in order to
reduce energy consumption and enhance the network’s lifetime. In this paper, we surveyed some MAC pro-
tocols for WSN and compared their design tradeoffs. The goal is to provide a foundation for future MAC
design, and to identify important design issues that allow us to improve the overall performances.
Keywords: Wireless Networks, Sensor Networks, Performance Evaluation
1. Introduction: Wireless Sensor Networks
A wireless sensor network (WSN) is a wireless network
consisting of spatially distributed autonomous devices
that use sensors to cooperatively monitor physical or
environmental conditions such as temperature, sound,
vibration, pressure, motion, or pollutants at different lo-
cations. The purpose of a WSN is to collect and process
data from a target domain and transmit information back
to specific sites. WSN technology is an emerging tech-
nology that can be utilized in a wide range of potential
applications including but not limited to, biomedical
treatment, military applications, traffic surveillance, fire
detection, structural and earthquake monitoring, industri-
al control, and rescue operations.
Such a network usually consists of a number of wire-
less sensor nodes that arrange themselves into a multi-
hop network. Each node consists of one or more sensors,
a low power radio transceiver or other wireless commu-
nication device, an embedded processor, and an energy
source, usually a battery. The size of a wireless node can
vary from the size of a shoebox, down to size of a grain
of dust, and cost varies depending on size. These size
and cost constraints result in corresponding constraints
on nodes resources, including energy, memory, compu-
tational speed and bandwidth.
2. Factors for Design of a MAC Layer Protocol
Considering that sensor nodes are likely to be battery
powered, and because they are often implemented in
environments where it proves to be difficult to change or
recharge batteries, prolonging the lifetime of nodes is a
critical issue for a successful wireless sensor network.
Not only does the transmission of data cost energy, but
receiving, and scanning for data also use a significant
amount of energy. In addition to being energy efficient,
WSN should be scalable and adaptable to change.
Change can come in the form of network size, node den-
sity, or topology. Additionally, nodes may die over time,
new nodes may join, or nodes may move to a different
location. A good MAC protocol should gracefully ac-
commodate such network changes. Lastly fairness, la-
tency, throughput, bandwidth utilization are also con-
cerns for WSN. However, these goals may be primary
concerns in traditional wireless networks, but they prove
to be secondary for WSN. This is due to the fact that in a
traditional wireless network, usually a number of differ-
ent applications may be competing for use of the com-
munication channel; however in a WSN, the nodes are
typically working for the same application.
3. Sources of Energy Waste
In a sensor, the Radio Frequency (RF) module, which
consumes most of the energy, becomes the crucial entity
to be optimized. Therefore, designing an energy-efficient
Medium Access Control (MAC) protocol is significant
factor in reducing energy consumption based on its direct
control over RF module [2]. There are four distinctive
sources of energy waste for wireless sensor nodes, colli-

M. GUNN ET AL.
696
sions, overhearing, control packet overhead, and idle
listening. Collisions are caused by contention, when two
nearby sensor nodes both attempt to access the commu-
nication channel at the same time. Overhearing is a result
of a node picking up packets that are destined to other
nodes. Common control packets used in WSN include
Ready-to-Send (RTS), Clear-to-Send (CTS), and Ac-
knowledge (ACK). The transmission of these packets
contributes to energy consumption, therefore a minimal
number of control packets should be used to make a data
transmission. Idle listening has proved to be one of the
major sources of energy waste. Given that a node does
not know when it will be the receiver of a message from
one of its neighbors, it must keep its radio in receive
mode at all times, resulting in idle listening. Studies have
shown that idle listening can consume 50-100% of the
power required for receiving [1].
4. Proposed MAC Protocols
There are a considerable number of MAC protocols that
have been designed and implemented for WSN. This
section will discuss a few of these protocols and their
essential behaviors.
4.1. Sensor MAC (S-MAC)
The key idea behind S-MAC [1] is the utilization of ma-
naged synchronized duty cycles. A duty cycle utilizes a
periodic awake and sleep schedule, allowing nodes in
sleep mode turn off their radio [1]. A duty cycle is
represented as a ratio of wake time to total cycle time, S-
MAC limits it duty cycles to about 10%, reducing energy
waste by attempting to minimize idle listening. Sleep and
listen periods are predefined and constant in S-M AC .
Additionally, nodes in S-MAC create virtual clusters
by periodically exchanging sleep schedules with their
neighboring nodes [1]. This exchange is implemented by
sending a SYNC packet, which is very short, and in-
cludes the address of the sender and the time of its next
sleep. Nodes that receive the SYNC packet will adjust
their timers immediately after they receive the SYNC
packet and will go to sleep when the timer fires. Thus the
schedules are updates and the nodes are synchronized.
Nodes that reside in two virtual clusters wake up for
the listen phases for both clusters. This however is one of
the drawbacks of the S-MAC algorithm, the possibility
of a node following two different schedules resulting in
more energy consumption via idle listening and over-
hear in g.
Lastly, S-MACs design includes the utilization of
adaptive listen, overhearing avoidance techniques, and
message passing. With adaptive listen, neighboring
nodes wake up for a short period of time at the end of
each transmission to listen for possible data transmis-
sions. To avoid overhearing, all immediate neighbors of
sender and receiver are put to sleep upon receiving
RTS/CTS. Resultantly neighbors do not overhear data
packets and following ACKS. The nodes use the duration
field in the packet, which indicates how long to sleep.
Message passing is a technique in which long messages
are divided into frames and sent in a burst. With this
method, nodes may achieve energy savings by minimiz-
ing communication overhead and latency at the expense
of unfairness in medium access.
4.2. Timeout MAC (T-MAC)
T-MAC is similar to S-MAC in that it utilizes an ac-
tive/sleep duty cycle. However, TMAC improves upon
the design of S-MAC by introducing an adaptive duty
cycle in which the active part is dynamically ended, in-
creasing the efficiency of the algorithm for variable traf-
fic loads. The idea behind the design of T-MAC is as
follows. While latency requirements and buffers space
are generally fixed, the message rate will usually vary.
Therefore, the nodes must be implemented with an active
time that can handle the highest expected load. Whenev-
er the load is lower than that which is expected, the ac-
tive time is not optimally used and energy will be wasted
on idle listening. To solve this inefficiency, the T-MAC
protocol implementation reduces idle listening by trans-
mitting all messages in bursts of variable length, and
sleeping between bursts. To maintain an optimal active
time under variable load, the length of the active time is
dynamically determined, ending in an intuitive way by
timing out when the node hears nothing [4].
Every node periodically wakes up to communicate
with its neighbors during active time periods. The nodes
communicate using a modified RTS-CT S-D ATA-ACK
four-step exchange to deliver messages, which provides
both collision avoidance and reliable transmission [4].
A node will keep listening and potentially transmitting,
as long as it is in an active period. An active period ends
when no activation event has occurred for a time TA. An
activation event includes but is not limited to, the recep-
tion of any data on the radio, the sensing of communica-
tion of the radio, the end-of-transmission of a node’s own
data packet or an ACK data packet. If no activation event
is sensed, the node then goes to sleep again until the next
frame. During the sleep mode, new messages are queued.
An important aspect of T-MAC is determining TA, the
time that a node must wait before it times out, and goes
to sleep. The idea is that a node should not go to sleep
while its neighbors are still communicating, since it may
be the receiver of a subsequent message [4]. Receiving
the start of the RTS or CTS packet from a neighbor is

M. GUNN ET AL.
697
enough to trigger a renewed interval TA. Additionally, a
node may be out of range, and therefore may not hear the
RTS that starts a communication with its neighbor, so the
interval TA must be long enough to receive at least the
start of the CTS packet.
4.3. S-MACL, a Global Sleeping Schedule
As mentioned the S-MAC protocol creates virtual clus-
ters in which the clustered nodes follow a common
sleeping schedule. In order to connect these virtual clus-
ters, nodes residing between clusters have to adopt mul-
tiple schedules. These nodes, known as border nodes,
constitute nearly 50 percent of the nodes in some net-
works and may have to adopt up to 4 different schedules.
These border nodes have to stay in active mode longer
than other node, which means that they waste more
energy than non-border nodes. Resultantly, these nodes
will die sooner, and the network coverage rate is reduced.
A more serious problem happens in multi-hop sensor
networks, in which border nodes have to act as interme-
diate outers to relay packets. The death of these border
nodes may increase the routing difficulty, even segment
a network. Some nodes will not be able to communicate
to the rest of the network [6].
To resolve the problem of multiple sleeping schedules,
S-MACL attempts to merge all the virtual synchro ni za-
tion clusters into one cluster to ensure that only one
sleeping schedule will be used in a fully connected net-
work. To do this, S-MACL utilizes the node id, a unique
identifier that is mounted on each sensor node. More
specifically it uses the id of the synchronizing sender
node and applies it as a schedule id. The scheduling
process in S-MACL is presented as follows. When a
node does not receive any SYNC frame after its first
listening period, it will arbitrarily choose one schedule
and announce this schedule and assign its own id as the
schedule id. We call such a node a synchro ni ze r , since it
chooses its schedule independently and other nodes will
synchronize with it. Otherwise, the node will receives a
schedule from a neighbor SYNC frame before having a
chance to choose its own schedule, and will follow that
schedule by setting its schedule as the same, and an-
nouncing this schedule to its neighbors. We call such a
node a follower. When a node receives a different sche-
dule from its neighbors’ SYNC frame, it will compare
the current schedule id and the new schedule id. Then it
will start following the schedule with the higher id. If the
new schedule in the incoming SYNC frame has a lower
id, this node will announce its own schedule during the
listening time of the new schedule. This operation en-
sures that nodes will always use the schedule with high-
est id. The authors show through various scenarios, with
different numbers of nodes and different topologies that
S-MACL performs better than S-MAC in most cases [6].
4.4. Patten MAC (P-MAC)
P-MAC [14] is unique in that instead of having fixed
sleep and awake schedules as with S-MAC, the sleep-
wakeup schedules of the sensor nodes are adaptively
determined, based on a node’s own traffic and that of its
neighbors. This improves throughput under heavy traffic
and reduces unwanted energy consumption while the
networks is performing under light loads when compared
to the performance S-MAC.
Similar to S-MAC, P-MAC is a time-slotted protocol,
however unlike S-MAC in which a node sleeps for a
duration of the time slot and is awake for the remainder,
in P-MAC, the node must either be awake or asleep for
the entire duration of the time slot. With P-MAC, a sen-
sor node gets information about the activity in its neigh-
borhood before sending communication packets through
patterns. Based on these patterns, a sensor node can put
itself into a long sleep for several time frames when there
is no traffic in the network. If there is any activity in the
neighborhood, a node will know this through the patterns
and will wake up when required. Thus P-MAC saves
more power than S-MAC as well as T-MAC, without
compromising on the throughput.
A sleep-wakeup pattern is a stream of bits indicating
the tentative sleep-wakeup plan for a sensor node over
several slot times [14]. A 1 in the stream indicates that
the node intends to stay awake during a slot time, while a
0 indicates that the node intends to sleep. Since the pat-
tern is only a tentative plan, it is subject to change. This
pattern stream of 1s and 0s is generated for each indi-
vidual node. These patterns are used to convey activity
from one node to its neighbors. Thus, the schedule for a
node is derived from its own pattern and, as well as the
patterns of its neighboring nodes, resulting in a schedule
for the network.
Pattern generation based on the binary strings that are
associated with a node over some number of time slots,
this is referred to as a period [14]. The nodes’ pattern is
updated during each period using local traffic informa-
tion available at the node and exchanged between the
neighboring nodes at the end of each period. When the
network is activated, the pattern at every node has just
one bit for the first period, which is 1. If there is no data
for a node to send at the first time slot of bit 1, then it
indicates that the traffic load is light, and the node can
afford to go to sleep. Consequently, the node updates its
pattern to 01, and so on. If during the next time slot, the
node still has no data to send, the node is encouraged to
sleep longer by doubling the number of 0 bits, ie. 001.
By exponentially increasing the sleep time during light
traffic the node is able to save a considerable amount of
energy. On the other hand, if a node has any data to
transmit at any time slot, regardless of the pattern bit at
that time slot, the next bit in the pattern becomes a 1.

M. GUNN ET AL.
698
These patterns are not the decisive sleep schedule for
the nodes; they are only a tentative sleep-wakeup plan
[14]. As mentioned P-MAC obtains it schedule based on
the node’s pattern, and the pattern of its neighbors. The
nodes broadcast newly generated patterns at the end of
the current period. As a result, the time is divided into
time frames, referred to as super time frames (STF).
Each STF has two sub-frames. In the first, the Pattern
Repeat Time Frame (PRTF), each node repeats its cur-
rent pattern. The second time frame, the Pattern Ex-
change Time Frame (PETF), is used for the exchange of
new patterns between neighbors. To obtain the actual
sleep-wakeup itinerary, the strings of bits are compared
between neighbors at each time slot as well as looking
for data packets in the buffer of the neighboring nodes at
each time slot. Based on a series of rules, the bits are
compared and a new pattern is created and followed. In
addition to the use of 1s and 0s, 1- bit is introduced. A 1-
bit will be used when the nodes pattern bit is 1 and either
the receivers bit is 0, or the node has no packets to be
sent. Therefore, 1- implies that the node should wakeup
at the beginning of the time slot and listen for a short
amount of time. If it hears no communication from its
neighbors, then it goes back to sleep. The reason for this
is that since the pattern bit for the node is 1, the node is a
candidate to be a receiver of communication and its
neighbors may try to send data to it. Thus, if the node
goes to sleep, the packet destined to it will be lost, and
energy is wasted.
4.5. Traffic-Adaptive MAC (TRAMA)
As a traffic load increase, the probability of collisions of
control or data packets occurring in any contention-based
scheme increases. This degrades channel utilization and
further reduces battery life [7]. TRAMA implementation
attempts to provide energy-efficient conflict free channel
access in wireless sensor networks by creating transmis-
sion schedules that are adaptive to changes, prolongs the
battery life of each node, and is robust to wireless loses
[7]. The protocol consists of three components: the
Neighbor Protocol (NP), the Schedule Exchange Proto-
col (SEP) and the Adaptive Election Algorithm (AEA).
Additionally, TRAMA uses single, time-slotted channel
access that is divided up into random and scheduled
access periods.
The main function of the Neighbor Protocol is to
gather two-hop neighborhood information by using sig-
naling packets. This protocol operates periodically dur-
ing random access periods. Schedule Exchange Protocol
utilizes a schedule consisting of intended receivers for
future transmission slots. Schedules are established based
on the current traffic information at the node, and are
periodically propagated to the neighboring node. SEP
mai ntains consistent schedules for the one-hop neighbors
of each node. The Adaptive Election Algorithm uses the
schedule information from SEP and the neighborhood
information to elect a transmitter, receiver and stand-by
nodes for the current time slot. Nodes that are not se-
lected to transmit or receive data for a particular time slot
are removed from the election process, allowing them to
switch to sleep mode and improving the channel utiliza-
tion. As a result, the sleep schedule of a node is a direct
function of the traffic going through the node and its
neighbors, and is synchronized automatically when
nodes exchange information about their identifiers and
their traffic [7].
TRAMA organizes access to the communication
channel into time slots allowing random and scheduled
access. Random Access periods are used for signaling,
synchronization, and updating two-hop neighbor infor-
mation. The scheduled access periods are used for con-
tention free data exchange between nodes.
4.6. B-MAC, a Versatile Low Power MAC
B-MAC is a carrier sense media access (CSMA) protocol
that utilizes low power listening and an extended pream-
ble to achieve low power communication [10]. Furthe r-
more, B-MAC is designed for duty cycled WSN, so
nodes have an awake and a sleep period, and each node
can have an independent schedule.
Periodic channel sampling or low-power listening
(LPL) is the primary technique that B-MAC employs.
LPL is carried out as follows. A node wakes up every
check-interval; it turns on the radio and samples the
channel. If activity (a preamble) is detected, the node
remains awake for the time required to receive the in-
coming data packet. After reception, the node returns to
sleep. However, if no packet is received, a timeout forces
the node back to sleep.
If a node wishes to transmit, it precedes the data pack-
et with a preamble that is slightly longer than the sleep
period of the receiver. The preamble is predefined data
automatically appended at the beginning of transmitted
data. By using an extended preamble, that is at least as
long as the sleep period, a sender is assured that at some
point during the transmission of the preamble, the re-
ceiver will wake up and detect the preamble, and remain
awake to receive the data packet.
A key challenge of B-MAC is implementing check in-
tervals that are very short which then ensure a reasonable
length for the preamble. Carrier sense duration also has
to be very short so that receiver does not have to spend
too much energy listening to the communication channel.
A carrier sense must be accurate to reduce latency of
transmission and energy consumption at sender.
B-MAC additionally utilizes software automatic gain
control as a method of Clear Channel Assessment (CCA),
which accurately determines if the channel is clear, thus

M. GUNN ET AL.
699
effectively avoiding collisions. This is a necessity so that
the node can determine what is a noise and what is a
signal, due to the fact that ambient noise is prone to en-
vironmental changes. This is achieved by taking signal
strength samples when the channel is assumed to be free,
such as immediately after transmitting a packet. These
samples are stored in a FIFO queue and the median of
the queue is added to an exponentially weighted moving
average with decay. This value gives a fairly accurate
estimate of the noise floor of the channel. Effectively, a
node, before transmission, takes a sample of the channel;
if the noise is below the noise floor, the channel is clear
and it can send immediately [10].
4.7. X-MAC, a Short Preamble MAC
While being simple and improving energy efficiency, the
low power listening approach used by B-MAC which
employs a long preamble is suboptimal in terms of ener-
gy consumption, is subject to overhearing, as well as
introducing excess latency at each hop [11]. This issue is
threefold. First, the receiver typically has to wait the full
period until the preamble is finished before the DATA/
ACK exchange can begin, even if the receiver has woken
up at the start of the preamble. Second, LPL suffers from
the overhearing problem, where receivers who are not
the target of the sender also wake up during the long
preamble and have to stay awake until the end of the
preamble to find out if the packet is destined for them.
This wastes energy at all non-target receivers within
transmission range of the sender. Third, because the tar-
get receiver has to wait for the full preamble before re-
ceiving the data packet, the per-hop latency is lower
bounded by the preamble length. Over a multi-hop path,
latency can accumulate to become substantial [11].
X-MAC is a low power MAC protocol that strives to
overcome these shortcomings by employing a shortened
preamble approach. The ideas behind this approach is to
embed address information of the target node in the
preamble so that non-target receivers can realize that
they are not the receiver and quickly go back to sleep.
This solution addresses the overhearing problem. Fur-
thermore, X-MAC introduces the strobed preamble. This
approach allows the target receiver to interrupt the long
preamble as soon as it wakes up and determines that it is
the target receiver. This is accomplished by dividing the
one long preamble into a series of short preamble packets,
each containing the id of the target node. Accordingly,
instead of sending a constant stream of preamble packets,
the protocol inserts small pauses between the series of
short preamble packets, during which time the transmit-
ting node pauses to listen to the medium. These gaps
enable the receiver to send an early ACK packet back to
the sender by transmitting the ACK during the short
pause between preamble packets. When a sender rece-
ives an ACK from the intended receiver, it stops sending
preambles and sends the data packet. This allows the
receiver to cut short the excessive preamble, which re-
duces per-hop latency and energy spent unnecessarily
waiting and transmitting [11].
Medium Access Protocols
Centralized
Distributed
Schedule Based Contention Based
Fixed Assignment Demand Assignment Slotted Access Random Access
TRAMA
P-MAC
S-MAC T-MAC
B-MAC
Shortened Preamble
S-MACL
Global Schedule
Clustered Schedule
Fixed Duty Cycle Adaptive Duty Cycle X-MAC
Figure 1. MAC design option s .

M. GUNN ET AL.
700
Table 1. Tradeoff analysis.
Energy Fairness Latency Throughput
S-MAC
(+) Periodic Sleep
(+) Message Passing
(-) Idle Listening
(-) Overhearing
(-) Message Passing
(-) Periodic Sleep
(+) Adaptive listen
(+) Message Passing
(-) Periodic Sleep
T-MAC (+) Adaptive Duty Cycle (-) Adaptive Duty Cycle (-) Adaptive Duty Cycle
S-MACL (+) Global Sleep Schedule
P-MAC (+) Adaptive Sleep Schedules (+) Adaptive Sleep Schedules
TRAMA (+) Transmission Schedules
(-) Overhearing (+) Transmitter Electron Algorithm (+) Transmitter Electron Algo- rithm
B-MAC
(+) LPL
(-) Long Preamble
(-) Overhearing
X-MAC (+) Shortened Preamble (+) Strobed Preamble
5. A Comparison
5.1. Comparison of Design
Centralized MAC protocol design can be divided into
two sub sections, schedule- based, and contention-based.
A schedule based design schedules nodes into different
sub -channels. Schedules protocols are successful in that
they avoid collisions thus promoting energy efficiency.
However, they tend to have poor scalability and adapta-
bility. On the other hand in a contention-based schedule,
nodes compete in a probabilistic coordination for access
to the communication channel. Contention-based proto-
cols have proved to be more scalable and flexible to to-
pology change. However , when compared with schedule -
based designs, they are not as energy efficient. Figure 1
illustrates which of the previous protocols discussed use
each of the schemes and partitions the design of the pro-
tocols into more detailed subsets.
5.2. Analysis of Tradeoffs
Table 1 represents comparison of the tradeoffs in proto-
col design based on the statistics available. Due to re-
source constraints, the table is not complete. A (+) indi-
cates a positive outcome of the subsequent design me-
thod. A (-) indicates a tradeoff of a network performance
metric as a result of the design techni que.
5.3. Protocol Comparison
This section offers a more detailed discussion of the ad-
vantages and disadvantages of the protocol design. It also
offers a direct comparison between some of the protocols.
Again, for some of the protocols, the information is mi-
nimal due to a lack of available information.
5.3.1. S-MAC
S-MAC reduces the amount of energy wasted by idle
listening, which is accomplished by introducing sleep
schedules. Its implementation is simple, and time syn-
chronization overhead is prevented with sleep schedule
announcements. Lastly, adaptive listening is used to re-
duce multi-hop latency due to periodic sleep modes and
nodes waiting until the subsequent listen period of the
intended receiver. Adaptive listen saves more energy for
heavy loads by reducing latency by at least half.
On the other hand, the S-MAC protocol essentially
trades energy efficiency for reduced throughput and in-
creased latency. Throughput is reduced because only the
active part of the frame is used for communication. La-
tency increases because a message-generating event may
occur during sleep time. Additionally, adaptive listening
incurs overhearing or idle listening if the packet is not
destined to the listening node. Lastly, sleep and listen
periods are predefined and constant, which decreases the
efficiency of the algorithm under variable traffic load.
For light traffics loads S-MAC offers significant
energy efficiency over always listening MAC protocols.
Simul ation experiments have shown that the S-MAC
protocol reduces the energy used by the radio with up to
30%, after optimal tuning. The energy savings and in-
creased throughput of S-MAC as compared with tradi-
tional protocols without sleep cycles such as CSMA and
IEEE 802.11 without duty cycle control is shown in Fig-
ures 2 and 3.

M. GUNN ET AL.
701
Figure 2. Energy consumption at different traffic loads.
Figure 3. Effective through put under highest traffic load.
Figure 2 shows that at light traffic load, periodic sleep-
ing has significant energy savings over fully active mode
and adaptive listen saves more at heavy load by reducing
latency. In Figure 3 one can see that adaptive listen sig-
nificantly increases throughput.
5.3.2. T-MAC
Simulation experiments have shown that the T-MAC
protocol reduces the energy used by the radio with as
much as 80% in a typical scenario when compared to
classical protocols like CSMA. The S-MAC protocol
saves only 30% in this scenario, after optimal tuning.
Implementation of the T-MAC protocol on real wireless
sensor hardware has shown that, in an idle situation, the
radio can be turned off for as much as 97.5% of the time,
reducing the total energy used more than 96%. In a situa-
tion with high message rates, the T-MAC protocol does
Figure 4. Energy consumption based on event triggered
reporti ng.
not increase the latency, since nodes do not sleep in that
case. Furthermore, the authors show that, for variable
workloads, T-MAC uses one fifth of the energy used by
S-MAC. While this adaptive duty cycling reduces energy
usage for variable workloads, these gains come at the
cost of reduced throughput and increased latency. Re-
sults of simulations are illustrated in Figure 4, which
compares the amount of energy used for CSMA, S-MAC,
and T-MAC in a typical scenario.
5.3.3. S-MACL
With S-MACL, all nodes consume less energy, especial-
ly the border nodes that act as intermediate routers,
greatly increasing the lifetime of these nodes. Addition-
ally, as a result of the global synchronization schedule,
the number of collisions is reduced, which also reduces
the amount of energy wasted. The contrastive simulation
of S-MAC with S-MACL results showed that S-MACL
achieves a great level of energy efficiency compared
with S-MAC.
5.3.4. P-MAC
Based on simulations done by the authors, in comparison
to S-MAC under light traffic loads, P-MAC consumes
less energy, though throughput remains the same. How-
ever, under heavy traffic loads, P-MAC consumes less
energy and achieves a higher throughput. This is due to
the fact that with S-MAC, sensor nodes must periodically
go to sleep, even if the traffic load is high. On the other
hand, the implementation of P-MAC allows the nodes to
stay awake due to the varying schedule patterns. Because
PMAC is able to adaptively schedule sleep and awake
periods, it offers more energy savings under light loads,
and higher throughput under heavy loads as compared to
S-MAC.

M. GUNN ET AL.
702
5.3.5. TRAMA
TRAMA is able to improve energy efficiency by utiliz-
ing transmission schedules that avoid collisions of data
packets at the receiving nodes. Additionally nodes switch
to low power radio mode when there are no data packets
intended for those nodes. Furthermore, TRAMA achie-
ves conflict-free transmission by scheduling access among
two-hop neighboring nodes during a particular time slot
and by allowing nodes to switch to sleep mode if they are
not selected to transmit or are not the intended receivers
of traffic for a particular time slot. Lastly, adequate
throughput and fairness is achieved based on the trans-
mitter-election algorithm that is inherently fair and pro-
motes channel reuse as a function of the competing traf-
fic around any given source or receiver. On the other
hand, TRAMAs efficiency is limited by its complex
election algorithm and data structure. Moreover, it incurs
overhead due to explicit schedule propagation as well as
higher queuing delays [7].
TRAMA implementation results in a higher percen-
tage of sleep time and less collision probability when
compared to CSMA based protocols, which greatly im-
proves energy savings. TRAMA has a higher delay but
higher maximum throughput than contention-based S-
MAC. Through extensive simulations, TRAMAs per-
formance is compared against a number of contention
and a scheduled based MACs. It is evident from the si-
mulation results that significant energy savings can be
achieved by TRAMA depending on the offered load.
TRAMA also achieves higher throughput (around 40%
over S-MAC and CSMA and around 20% over 802.11)
when compared to contention-based protocols because it
avoids collisions due to hidden terminals [7].
5.3.6. B-MAC
The authors have show that testing the communication
channel for activity is about 10x less expensive than lis-
tening for a full contention period. Idle listening is re-
duced in the B-MAC protocols by shifting the burden of
synchronization to the sender: when a sender has data,
the sender transmits a preamble that is at least as long as
the sleep period of the receiver; thus, the sender and re-
ceiver can be completely decoupled in their duty cycles
[10]. This removes the need for, and the overhead intro-
duced by, synchronized wake/sleep schedules.
The authors show that B-MAC surpasses existing
protocols in terms of throughput, latency, and for most
cases energy consumption. It is simple in both design
and implementation. While B-MAC performs quite well,
it suffers from the overhearing problem, and the long
preamble dominates the energy usage. Additionally,
while, unscheduled sleep reduces control overhead, con-
seque ntially, the sender incurs greater overhead to wake
up the unsynchronized receiver from sleep.
The performance benchmark has shown that B-MAC
outperforms S-MAC with greater energy savings and
network performance [7].
5.3.7. X-MAC
B-MAC requires more time to transfer packets from the
source to the destination. This is because the entire
preamble has to be always sent, even though the receiver
was already awake. X-MAC saves this time, thus con-
serving energy.
6. Conclusions
When developing a MAC protocol, prolonging lifetime
for nodes is a critical issue to consider in order to pro-
mote for a successful wireless sensor network. Many of
the developed protocols are developed with specific as-
sumptions in mind and for specific applications. In this
article, we surveyed wireless MAC protocols for wireless
sensor networks, and we can conclude that no protocol is
the “best” implementation. However, each of these pro-
tocols addresses different issues that arise from energy
waste in sensor nodes.
7. References
[1] W. Ye, J. Heidemann, and D. Estrin, “An energy-efficient
MAC protocol for wireless sensor network,” In Proceed-
ings of the INFOCOM’02. IEEE Computer Society, San
Franci s co, 2002.
[2] J. Ai, J. F. Kong, and D. Turgut, “An adaptive coordi-
nated medium access control for wireless sensor net-
works,” In Computers and Communications Proceedings,
ISCC’04, 2004.
[3] I. Akyildiz, W. Su, Y. Sankarasubramaniam, and E.
Cayirci, “Wireless sensor networks: A survey,” In Com-
puter Networks, Vol. 38, No. 4, pp. 393–422, 2002.
[4] T. Zheng, S. Radhakrishnan, and V. Sarangan, “PMAC:
An adaptive energy-efficient MAC protocol for Wireless
Sensor Networks,” In Parallel and Distributed Processing
Symposium Proceedings, 19th IEEE International, 2005.
[5] LAN MAN Standards Committee of the IEEE Computer
Society, IEEE Std 802.11-1999, Wireless LAN Medium
Access Control (MAC) and Physical Layer (PHY) speci-
fications, IEEE, 1999.
[6] L. Zhang, G. Somnath, V. Prakash, and S. Samar, “An
energy efficient wireless sensor MAC protocol with
global sleeping schedule,” In Computer Science and its
Applications, CSA’08, International Symposium, pp.
303– 308, October 2008.
[7] V. Rajendran, K. Obraczka, and J. J. Garcia-Luna-Aceves
“Energy-efficient, collision-free medium access control
for wireless sensor networks,” In Wireless Networks, Vol.
12, pp. 63, 2006.
[8] J. M. So and N. Vai dya, “Multi-ch annel MAC for ad hoc
networks: Handling multi-channel hidden terminals using

M. GUNN ET AL.
703
a single transceiver,” Talk at Workshop with Intl. School
on WSN, Dagstuhl, Germany, August 30, 2005.
[9] C. L. Fullmer and J. J. Garcia-Luna-Aceves, “Solutions to
hidden terminal problems in wireless networks,” In Pro-
ceedings ACM SIGCOMM’97, Cannes, France, Septem-
ber 14-18, 1997.
[10] J. Polastre, J. Hill, and D. Culler, “Versatile low power
media access for wireless sensor networks,” In Second
ACM Conference on Embedded Networked Sensor Sys-
tems, 2004.
[11] M. Buettner, G. V. Yee, E. Anderson, and R. Han, “X-
MAC: A short preamble MAC protocol for duty-cycled
wireless sensor networks,” In Proceedings of 2nd ACM
Conference on Embedded Networked Sensor Systems
(SenSys’06), pp. 307–320, 2006.
[12] S. Singh and C. Raghavendra, “PAMAS: Power aware
multi-access protocol with signalling for ad hoc net-
works,” ACM SIGCOMM Computer Communication
Review, Vol. 28, No. 3, pp. 5–26, July 1998.
[13] P. Karn, “MACA - a new channel access method for
packet radio,” In ARRL/CRRL Amateur Radio 9th Com-
puter Networking Conference, pp. 134–140, 1990.
[14] LAN MAN Standards Committee of the IEEE Computer
Society, IEEE Std 802.11-1999, Wireless LAN Medium
Access Control (MAC) and Physical Layer (PHY) speci-
fications, IEEE, 2000.