Energy Efficient Medium Access Control Protocols for
Wireless Sensor Networks and Its State-of-Art*
Ruizhong Lin, Zhi Wang, and Youxian Sun
National Laboratorv of Industrial Control Technology, Institute ofModem Control Engineering
Zhejiang Universitv. Hangzhou. 310027.China
e-mail: rzlin "wangzhii
Abstract-Recent advances in Wireless Sensor Netoiiorks
(WSN) have led to many new pr-otocols specifically designed for
different kinds of applications where energ, efficienc3
essential consideration. Most of the attention, however, has
been given to the Medium Access Control (MAC) protocols
since they pay an important role in wireless communications
and traditional MAC protocols are not suitable for WSN. This
paper tries to survey recent engergy efficient MAC protocols
for WSN and
approaches pursued. The four main categories explor-ed in this
paper are scheduling based, collision firee, contention based and
hvbrid Schemes. Seveeral energ-y-efficient MAC protocols are
described and discussed under the appropriate categorx.
Moreover, requirements for the MAC layer in WSN are also
discussed. The paper concludes with open research issues in the
Index Terms-wireless sensor networks, Medium Access
Control protocol, energy efficiency.
In recent years. we hav-e seen a proliferation of wireless
devices, including cellular phones. pagers. laptops and per-
sonal digital assistants (PDAs). Advances in energy efficient
desi'gn xireless technolog-. sensing technology and MEMS
low -power, multifunctional WSN. WSN will have a sig-
nificant impact on our lives in the 21 st century. Microsensor
devices, ranging in size from cubic inches to cubic millime-
ters, x'ill each have multiple on board miniaturized sensors
(such as for light. temperature. humidity, acoustics, imaging.
etc.). with considerable processing power. in each geo-
graphic position abilitv through global positioning systems
(GPSs) or local positioning methods, and short range radio or
optical communication links. These devices. which will be
cheap and smart, can be deployed in small or very large
numbers to instrument homes and highvwavs, buildings and
bodies, cities and infrastructures, as well as for monitoring
and controlling defense applications. WSN forms a new kind
of wireless network-s with a new set of characteristics and
Despite the diversity of the applications in WSN. the
following requirements are true for all of them: low power,
low cost, wireless and ad -hoc. Among these requirements.
low power is the most important. In most applications, the
nodes are unattended and live only. as long as their batteries
have enabled the dev7elopment of lox -cost.
*This paper is supported by ARPFC-PRA S103-02 and NNSFC-60304018
can support. The node has limited resources. like limited
processing capabilit. limited memory and limited batten',
enerp- and etc. Thus the sensor node must use these limited
resources effectivel7 and manage the energx to extend the
lifetime of the netwATork as much as it can. Therefore. energy
management is a challenging problem in desigling a WSN.
Man>- ofthe research challenges facing WSN reside in the
communication protocol stack, including the application
laver, transport later. netw-ork laver. data lih1k laver and
phy'sical later. Nox'el communication protocols must be
developed to suppoir higher lev'el services inW SN. There are
some kev challenges that cross cut all la> ers of the commu-
Scale. Unpredictable Wolok-rloads. .Vounlifornl NVode Disti-i-
butioni, High Fault Rates and EnergJy Constrain7t. Hoxvever,
scholars have developed many new protocols specificall>T
designed for different kinds of applications where energx
efficiency is an essential consideration. Among them, most of
the attention has been given to the MAC protocols since they
pa> an important role in w ireless networking and like in all
shared-medium networks. MAC is an importalnt technique
that enables the successful operation of the network. How-
ever the existing MAC protocols can not be used in WSN for
MAC schemes in other wireless netw orks hav'e been de-
signed for different goals from WSN and can not meet the
requirements in WSN. e.g.. providing high Quality of Service
(QoS) and bandwvidth efficiently in cellular networks. and
prox'ision of high QoS under mobile conditions w7ithout or
more less considering ofpower consumption in the mobile ad
hoc networks (MANET).
As mentioned above, one of the most important consid-
erations ofMAC protocols in WSN is energ' efficiency and
the MAC of WSN has to be optimized for power consump-
tion. In WSN. the MAC performance has been predomi-
nantlv measured in terms of bandwidth requirement, power
consumption. contention mitigation, and support to maintain
deliverv' has not been a metric to be optimized. but is likely to
become increasingly important as sensor netwvorks are de-
ployed in critical applications. Timeliness is perhaps the
most difficult requirement to meet since it brings to the fore
the tradeoffs between power consumption, interference
mitigation, and scheduling and routing efficiency.
Up to now., there are man>' different MAC protocols
specifically' designed for WSN, and in this paper they are
classified into four categories: 1) scheduling based; 2) colli-
sion free; 3) contention based: and 4) hybrid schemes. In the
 Data-cenitr-ic. Locationi-based. Large
. The latency incurred in message
0-7803-8304-4/04/$20.00 Uc2004 IEEE
followving sections;, we survey the recent energ
MAC protocols for WSN and discuss several main protocols
under the appropriate category. We also summarize the
state-of-art and specifically identify the special requirements
of a MAC laver in WSN in section VI. And we conclude the
discussions with open research issues in the end.
H. SCHEDULING BASED MAC PROTOCOLS
In scheduling-based MAC protocols, the time at w-hich a
node can transmit is determrined bx a scheduling algorithmi
so that multiple nodes can transmit simultaneouslx' wxithout
inteiference on the wireless channel. The time is usually di-
vided intoslots,and slots are further organized into frames.
Within each frame. a node is assigned at least one slot to
transmit. A scheduling algorithm usually finds the shortest
possible frame so as to achieve high spatial reuse and lowv
Most of work- has been focused on time dix'ision multiple
access (TDMA) scheduling, like Self-Organizing Medium
Access Control for Sensor Networks (SMACS) and Eaves-
drop-And-Register (EAR) algorithm
ergy-Aware MAC pi-otocol (DE-MAC) . and Energy Ef-
ficient MAC Protocol for Sensor Netw-orks (EMACS) .
Most ofthem are distributed (so as to be time s'nchronized)
and do not require global connectivit-
result, they can adapt adequately and keep the optimalitx'
propertvx in highly dy'namic environments (such astopology
. Distributed En-
information. As a
A. SMA CS andE.4R algorithm
The SMACS  protocol achieves netxxork start-up and
link layer organization. and the EAR algorithm enables
seamless connection ofmobile nodes in a WSN. SMACS is a
distributed infrastructure-building protocol which enables
nodes to discover their neighbors and establish transmis-
sion/reception schedules for communication without the need
for any' local or global master nodes. In this protocol, the
and chalnel assignment phases are
combined so that by the time nodes hear all their neighbors.
they would have formed a connected network. A communi-
cation link consists of a pair of time slots operating at a
randomly chosen. but fixed frequency (or frequency hopping
sequence). This is a feasible option in sensor networks, since
the available bandwidth can be expected to be much higher
than the maximum data rate for sensor nodes. Such a scheme
avoids the necessitv for network-wide sv'nchronization, al-
though communicatimg neighbors in a subnet need to be time
synchronized. Power conservation is achieved by using a
random wake-up schedule during the connection phase and
by turning the radio off during idle time slots.
TheEAR protocol  attempts to offer continuous service
to the mobile nodes under both mobile and stationariy con-
ditions. Here, the mobile nodes assume full control of the
connection process and also decide when to drop connections,
thereby minimizing messaging overhead.
transparent to the SMACS, so that the SMACS is functional
until the introduction of mobile nodes into the network. In
this model. the netwNork is assumed to be mainlx static. i.e..
anv mobile node has a number of stationarx nodes in its vi-
cinitx. A drawback of such a time-slot assignment scheme is
the possibilitx that members alreadx belonging to diffei-ent
subnets might nev7er get connected.
B. DE-,ALLAC Protocol
In DE-MAC protocol . each node knows xhlich slots its
neighbors will use to transmit packets. The main idea of
DE-MAC is to let the nodes exchange information about their
enerpg levels. Based on thatenergy level information, each
node decides to use one or two of the slots for transnission.
DE-MAC uses tow types of control packets. Vote packet and
Radio-powler-mode packet, to exchange theenergy informa-
tion. Each node maintains a local state variableRadio-mode
for each of its neighbor indicating the Radio-poNwer-mode of
the neighbor i. This information about the neighbor is used to
set its receivrer to listen for packets from its neighbors. When
a node falls below- the previous election winner'is energ
value, it enters the voting phase. The node in voting phase
sends its current energy value to all of its neighbors and
collects all the votes from them. If each of the neighbors'
energy value is bigger than this node's. it will receix'e posi-
tive votes from all of the neighbors. and it owns the cuiTent
slot to send data or put the radio to sleep if it has nothing to
DE-MAC uses the concept of periodic listen and sleep. A
sensor node sxxitches off its radio and goes into a sleep mode
onlyT xvhen it is in its owxn time slot and does not hav7e anything
to transmit. Thus a low energy node sleeps more than higher
energ- nodes. thus balancing theenergy among the nodes and
thus increasing the energy savings and therebxT increasing the
lifetime of the network. However, a node has to keep the
radio awvake in the slots assigned to its neighbors in order to
receivTe packets from them even if the node with current slot
has nothing to transmit.
EMACS  is also a TDMA based MAC scheme. Time is
divided into so called frames and each frame is divided into
timeslots and each slot contains three sections: commumica-
tion request (CR), traffic control (TC)., and data section. Each
timeslot can be oxvned bxT only one network- node. This
network node decides w hat communication should take place
in its timeslot. Other nodes can ask for data or notifx' the
avTailability of data for the owner of the timeslot in the CR
section. The owner of the slot transmits its schedule for its
data section and broadcasts a table in the TC section. which
tells to x,vhich other TC sections the node is listening. After
the TC section, the transmission of the actual data packet
follows. The Node goes into a standby sleep mode when at a
certain time no transmissions are expected. It releases its slot
and starts periodically listening to a TC section of a frame.
When the node has to transmit some data (event driven
sensor node), it can just fill up a CR section of another net-
workl node and agree on the data transmission, complete it
and go back to sleep. EMACS minimizes the utilization of
the transceivers of the nodes to save power. Latency in the
network is reduced by allowing transmissions in not owned
or released data sections. The traffic control section can be
deployed to make wvake-up calls for sleeping nodes.
II1. COLLISION FREE MAC PROTOCOLS
An important perfonnance criterion in data-centric WSN
is timeliness. There are some collision fi'ee MAC protocols
devzeloped for WSN, e.g.. Spatial TDMA , the Implicit
Prioritized Access Protocol
medium access protocol (TRAMA) .
The collision free multihop charnel access protocol:
Spatial TDMA is used for the transport of periodic and fre-
quent traffic in WSN . and the power consumption of
Spatial TDMA with background preamble sampling
computed. taking into account the power cost of keeping
synchronized two nodes having imprecise clocks. With the
hardwvare parameters considered (1 .8mW in reception. 9mW
in transmission. reversal time of Ims. 1 LR6 Alkaline battein')
the netwvork- can last more than 5 vears rumling Spatial
TDMA w ith background preamble scarning. wvith a delay of
10 seconds per hop.
. and the traffic-adaptive
A. the Imtiplicit P7rioritizedAccess Pr-otocol
Bv exploiting the periodic nature of sensor network- traffic,
Caccamo et al. realize collision-free real-time scheduling as
follows : frequency division multiplexing (FDM) is used
among adj acent cells to allow for coneurrent communica-
tions in different cells. Implicit earliest deadline first (EDF)
schedulina is used inside each cell. There is a router located
in the center area of each cell. Router nodes are equipped
with two transceivers so thev can transmit and receix'e at the
same time using twvo different fi'equencs' channels.
intracell commnumication: The key idea for conflict free
real-time scheduling is to replicate the EDF schedule at each
node for packet transmission. If the schedules are kept iden-
CIBAA |AlBCIB CIAIAIAlB
C |A JA
(a) Example of implicit contention using EDF
4,recholi | dlrechor. F:
4/ tCH5 /
(b) Exalmiple ofthe inter-cell communication mechanism using TDM
Fig. 1 the Implicit Prioritized Access Protocol
tical. each node w ill knoN which one has the message w ith
the shortest deadline and has the right to transmit next. FoI-
instance- suppose each node is given a message table as
shown in Fig. 1(a), the same schedule is derived bh
node in the cell according to EDF (deadline ties are broken in
favor of the node with the highest address ID). Due to the
identical ordering ofthe schedule at each node, a node knows
wi-hich node should transmit next. In addition. when a node is
listening to the channel, it is also able to know the completion
of a node's transmission and. thus, update its scheduling
queue for the next round of communication.
Take Fig. 1(a) as an example: the scheduling table re-
sen'es the worst case message transmission time for each
periodic message stream. Suppose that node A in its first
round uses onlyT one of three reseneTd frames. Since all nodes
are listening, they' know. that Node A has finished earlil and
Node B is the next one to transmit. Instead of transmitting its
resein'ed periodic message early', Node B maxy use the twNo
frames left by Node A to send best effort a periodic messages.
This is the obsernation that prompted the development of the
FRAme SHaring (FRASH) teclmique  designed to sv=s-
tematicallv and reliably exploit resen'ed. but unused. frames.
linter-cell communiaftlcationi: Each router node transmits in-
tercell messages using the channel of the cell it belongs to.
and receives intercell messages using the channel ofthe cell it
expects to receive from. Intercell messages are ordered by
earliest deadline bi each router, and each of them is able to
reach only its six neighboring cells w'ithin one hop. When-
ever an intercell frame occurs s\'nchronouslh in all cells, each
router transmits and receives intercell messages according to
a predetermined direction, which is the same for all cells.
Note that there are six possible directions that are assigled
staticallv to the intercell frames followTing a periodic scheme
as Fig. 1(b) shows.
Take frame 2 as an example: notice that router R is re-
ceiving a message from router R1 using chaimel four and is
transmitting a message to routerR3using channel six. Dur-
ing the same frame, R; is receiving from
and is transmitting toR,on channel one: in short, each
roteiS transmittingan eeRn
the same time. After the routing path is set, the end-to-end
delav is simply the sum of cell delays along the message path.
The interference due to the intercell frames can be taken
into account in the cell schedulabilitv analNsis as blocking
tem-s. In fact, let ,,, be the message transmission time, T be
message period.andr81c8 (Tbhc
tercell frames, the schedulability of intracell messages can be
using the approach proposed in : all the
messages are sorted bv increasing relative deadlines, so that
D,<D,onlv if i c<
time of each message is equal to the maximum number of
intercell frames that can occur during the message period
, on channel six
n the same direction at
.2) be theperiodof the in-
It is worth noting that the blocking
Due to the contention free nature of this metliod, implicit
EDF not only provides guaranteed schedulabilitx. but also
delivers higher throughput, especiall
load as compared wvith commonhi used ad-hoe network
during heavy work-
TRAMA  is an energx -efficient collision-free channel
access protocol for WSN. TRAMA reduces energy con-
sumption by ensuring that unicast, multicast. and broadcast
transmissions have no collisions, and by allowing nodes to
switch to a low-power, idle state wvhenever thev are not
transmitting or receiving. TRAMA assumes that time is
slotted and uses a distributed election scheme based on in-
fonnation about the traffic at each node to determine w-hich
node can transmit at a particular time slot. TRAMA avoids
the assignment oftime slots to nodes wivith no traffic to send.
and also allows nodes to deteirmine when thex can become
idle and not listen to the channel using traffic inffonmation.
TRAMA is shown in  to be fair and correct. in that no idle
node is an intended receiver and no receiver suffers collisions.
The performance of TRAMA is evaluated through extensive
simulations using both synthetic- as well as sensor- netweork
scenanos. The results indicate that TRAMA out peiforms
contention-based protocols (e.g.. carrier sense multiple ac-
cess (CSMA). IEEE 802.11. Sensor-MAC ) as well as
scheduling-based protocols with significant energy savings
IV. CONTENTION BASED MAC PROTOCOLS
Most of the distributed MAC protocols are based on car-
rier sensing and/or collision avoidance mechanisms and max
emplo- additional signaling control messages to deal with
hidden and exposed node problems. Such signaling messages
mav be delivered in two waxvs: in-band handshak-ing (e.g..
(T-MAC) ) or out-of-band signaling. There are also some
MAC protocols, like CSMA based MAC combined with
spread spectrum . and the Adaptive Rate Control Based
Protocol , do not use anv MAC control packet. Other
contention based MAC protocols are as follows, the conten-
tion window,7 based CSMA protocol - Sift , Radio pro-
tocol , WiseMAC , Aloha . extended IEEE
802.11 based RAP MAC , and non-persistent CSMA
with preamble sample (NP-SCMA-PS) .
S-MAC  uses three novel techniques to reduce energy
consumption and support self-configuration. To reduce en-
ergv consumption in listening to an idle channel, nodes pe-
riodically sleep. Neighboring nodes form virtual clusters to
auto-synchronize on sleep schedules. S-MAC also sets the
radio to sleep during transmissions of other nodes. And it
uses in-band signaling. Finallv, S-MAC applies message
passing to reduce contention latency for sensor-network ap-
plications that require store-and-forward processing as data
move through the network. By evaluating the implementation
ofS-MAC over a sample sensornode. the Mote. developed at
Universitx of California. BerkLelev, the experiment results
showv that. on a source node. an 80f2. 11 -like MAC consumes
2-6 times more energx than S-MAC for traffic load wvith
messages sent everv 1-lOs .
To handle load xvariations in time and location. T-MAC 
introduces an adaptive duty cx cle in a novel wax: bx' dx-
namicallx ending the active part of it. It dynamically adapts a
listen/sleep dutv cycle through fine-grained timeouts, xxhile
having minimum complexity. This reduces the amount of
energy wasted on idle listening, in xvhich nodes wait for po-
tentiallv incoming messages. xvhile still maintaining a rea-
sonable thr-oughput. ByT the head-to-head comparison with
classic CSMA (no dutx cxvcle) and S-MAC (fixed duty cycle)
through extensive simulations in . it can be found that
under homogeneous load. T-MAC and S-MAC  achieve
similar reductions in energy consumption (up to 98 %)
compared to CSMA. And in a sample scenario xvith xvariable
load, however. T-MAC outpelforms S-MAC bx a factor of 5.
C. 4daptve Rate Contrt;ol Based Pr-otocol
As reported and based on simulations in  the constant
listen periods are energy efficient and the introduction of
random delax proxvides robustness against repeated collisions.
Fixed window and binanT exponential decrease backoff
schemes are recommended to maintain proporlional fairness
in the netwxxork. A phase change at the application levlel is also
adv7ocated to get ovxer anx capturing effects. It is proposed in
this work that the energx consumed per unit of successful
communication can serxe as a good indicator of energy ef-
An adaptixve transmission rate control (ARC) scheme that
achieves medium access faimess by balancing the rates of
originating and route-thru traffic is also discussed here. This
ensures that nodes closer to the access point are not favored
over those deep down into the network. The ARC conltrols
the data origination rate of a node in order to allov the
route-thru traffic to propagate. A progressive signaling
mechanism is used to inform the nodes to lower tlheir data
originating rate. The computational nature of this scheme
makes it more energy efficient than handshakling and mes-
saging schemes using the radio. The ARC also attempts to
reduce the problem ofhidden nodes in a multihop network- by
constantly tuning the transmission rate and performing phase
changes, so that periodic streams are less likely to repeatedlxv
V. HYBRIDMAC PROTOCOLS
Several MAC protocols, such as Physical Layer Driven
, Hybrid TDM-FDM MAC
channel assignment table (CAT) based MAC  can be
viewed as hybrid schemes.
, and the
A. Physical Laver Driven Protocol
In , the effect of non-ideal physical layer electronics
on the design of MAC protocols for WSN is investigated.
The system is assumed to be made up of energy constrained
sensor nodes that colmmunicate to a single. nearby, high
powered base station (<10 m). Specificalix. the machine
monitoring application of sensor networks. with strict data
TDMA-FDMA medium access scheme is proposed. While a
pure TDMA scheme dedicates the full bandxxidth to a single
sensor node. a pureFDMA scheme allocates minimum signal
bandwidth per node. Despite the fact that a pure TDMA
scheme minimizes the transmit on-time, it is not alwax's
preferred due to the associated time synchronization costs.
An analyTtical fonnula is derived in  to find the optimum
number of channels w7hich gives the lowest sx stem power
consumption. This determines the hybrid TDMA-FDMA
scheme to be used. The optimum number of channels is
found to depend on the ratio ofthe power consumption of the
transmitter to that ofthe receiver. Ifthe transmitter consumes
morepower,a TDMA scheme is favored, wrhile the scheme
leans toward FDMA xwhen the receiveer consumes greater
B. Hvbrid TDs,AI-FDAIAL4C
A hybrid TDM-FDM MAC is proposed in . wiuhere
both time and frequency are divided into transmission slots.
It is a vrariable bandwidth allocation scheme. xvith consider-
ing the inefficienex of the startup energy s dominating the
overall energy consumption of the sensors for short packet
sizes. to reduce the energy consumption forWSN wvIhich have
large spatial variation in sensor distribution.
Instead of all the nodes broadcasting on a single channel.
the given frequencv band in CAT-based MAC  is parti-
tioned ilnto multiple channels and each node is assigned a
CAT-based MAC is as follows: a node listens to a commion
control channel (CCC) and periodically broadcasts a channel
assignment packet (CAP) to its neighbors using the same
channel. Evern node keeps a channel assignment table (CAT)
to record the chanmel usage by its one-hop and twvo-hop
neighbors. and make sure, its own channel is different from
all its two-hop neighbors. A novel wake-up radio scheme is
also incorporated for a node to send a packet to a sleeping
node. Combined with the multi-channel scheme and wake-up
radio. the mobilitv aware and adaptive protocol can reduce
network- maintenance overhead and shows much higher
power efficiency for typical sensor network- applications.
VI. CHALLENGES FORMAC TECHNOLOGY INWSN
WSN provide a different computation and communication
infras-tructure from those for traditional xxireless networks.
Those differences originate not only from their physical
characteristics, but also from their typical applications. For
example, physical characteristics include the large scale of
deployment, limited computing capability, and constraints on
power consumption. Typical applications include tracking
objects or detecting evTents, which are seldom emphasized in
mainstream traditional wireless netxorks.
As aresult, therequirements for the MAC lax er of a WSN
are noticeablx different from those for traditional networks.
The major requirements for the MAC lax er in a WSN are as
WSN are often deployed in a physical environment and are
expected to interact with the environment. Therefore, the
timelx detection. processing. and deliverx of infomiation are
often indispensable requirements in a WSN application. As
the base of the communication stack, the MAC laxer should
support real-time guarantees or QoS features .
Decentralized: Most algorithms running in WSN need to
be decentralized. This is due to both the large scale of the
network- and the intrinsic unreliabilitv of any single node in
the network. Consequently. the MAC layer needs to run
Pow,ei aware: In the design of a MAC protocol for WSN.
the poweer limitations need to be taken into consideration.
This has two direct implications. One is that the MAC pIo-
tocol needs tobemindful that the power maz not alwaxTs be
available. This could be because the power management
service has put the node to sleep to save powver. or the node
has actuallv run out of pow'er. The other is that the MAC
protocol needs to save powver consumption as much as pos-
sible. For example. the MAC protocol may want to axoid
excessixe collisions, continuous listening, and long-range
Flexibility: WSN are often application specific. While
there are tvpical applications for WSN. different applications
still exhibit peculiarities on their usage pattern ofthe network.
As a result, the MAC laver of a sensor netxvork- needs to be
flexible enough to accommodate a variety of network- traffic
pattems-rate-based or bursty. reliable or best effort, and so
Balance among muiltiple met7ics: The MAC design for
WSN needs to accomplish a balance among a number of
metrics. This balance might be more important than the
perfonnance on anv indixvidual metric. In an unbalanced de-
sign. a protocol that performs excellent on one performance
metric in lab experiments could observe surprising per-
formnance degradation in real enxvironnments. For example, a
protocol can use a smart scheme to save power. HowevTer, if
this scheme does not consider other metrics. such as the
real-time guarantee or reliability of the packet deliverv. it
could not only hinder the performance on other metncs, but
also degrade on the performance of power sax'ing. For ex-
ample. if the node tums off the radio component too often.
some packets max be lost and more retransmissions could
happen, which result in an even greater power consumption.
With these requirements in mind, we can evaluate current
MAC layer technologies and consider whether they are
suitable for sensor networks.
TDMA is a promising technology because it provides fair
usage of the channel and, if equipped with an adequate
scheduling algorithm, could also avoid collisions. But many
TDMA protocols use global ilnformation to do scheduling,
wvhich render those protocols to be impractical in genernal
sensor netxxorks. Besides. some of the protocols still hax'e
collisions, and it is quite difficult to control the collisions to
the degree that does not hurt the guarantee of timeliness.
These issues mak-e it difficult for existing TDMA protocols
to be broadl1 used in sensor networks.
Collision-free protocols are surely' notew7orthv because
they save power bv eliminating collisions. A good colli-
throughput., reduce the delav, and provide real-time guaran-
tee. A problem in a large class of current collision-free pro-
tocols is the use of multiple channels . This imposes a
nontrixial requirement on the hardware of the nodes in a
sensor network. Further studyT is needed to tell whether the
peiformance gain would overcome the increased cost of
hardxxare. Another concern is the complexity' of the protocol.
Normally, simple protocols are preferred because of the
limited computinpg capabilitv of nodes in the netxx ork.
Contention-based protocols often havTe difficulty in pro-
viding real-time guarantees. As mentioned above, collisions
also waste energ. How ever, there have been some adx'ances
in this area w-hich can largelv mitigate chances of collisions
and reduce power consumption . This could be useful in
some applications w-here predictabilitx- is less critical and
power consumption is the main concern. On the othei- hand.
for the collision-based protocols to be successfullv used in
sensor netxxorks. a xx7ell-defined statistical bounld is still
needed. And the hvbrid protocols. integrating more than one
approach's adx'antages, may be useful in meeting the re-
quirements in WSN.
In summary, existing wireless MAC protocols focus on
optimizing system energx wvith considering the MAC per-
fonnances. vet still do not adequately consider all of the
requirements of sensor netwvorks. The key challenge remains
to provide predictable delay and/or prioritization guarantees
w'hile minimizing ox'erhead packets and energx' consumaiption.
MAC protocols in WSN hav13e attracted a lot ofattention in
the recent years and introduced unique challenges compared
with traditional MAC in other wireless networks. In this
paper. we have summarized recent research on MAC in WSN
and discussed the open research issues. Though many MAC
schemes have been proposed for WSN, the area is still
largely' open to research. And in some cases. QoS supported
MAC may be needed in WSN . Along with the current
research of MAC protocols in WSN, we encourage more
insight into the problems and more development in solutions
to the open research issues as described in this paper.
John A. Stankovic, Tarek F. Abdelzaher, Chenyang Lu. Lui Sha and
Jennifer C. Hou. "Real-Time Communication and Coordination in
Embedded Sensor Networks".Proceedings ofthe IEEE, Vol.91, No.7.
Comnunnications, October 2000. pages 16-27.
R. Kalidindi. L.Ray. R. Kiannlani. S. Ivengar. "Distributed Energx
.Aware MAC Layer Protocol For Wireless Sensor Networks", inl In-
ternational Conference on WTJireless Aets-'ork-s
Conmmunicationi Protocols fol W7ireless Sensor Networks", European
Research on Aliddleware and Architectures for Coniplex and Em-
beddedS)ystemsWforkshop. Pisa, Italy. April 2003.
Anre El-Hoiydi. "SpatialTDMAanid CSMA with Preamble Sam-
pling for Low Power Ad-hoc Wireless Sensor Networks-, IEEE
Symposium on Computers and Comnmnication.s (ISCC). Taormina.
Italv. July 2002.
Marco Caccamo. Lynn Y. Zhanig.
hnplicit Pnroritized Access Protocol for Wireless Sensor Networks".
Proceedings of the 23rd IEEE Real-Timie Systens Symposium
tRTSS'02). December 2002.
\. Rajendran. K. Obraczka. and
ergy-Efficient. Collisioni-Free MIedium Access Control for Wireless
Sensor Networks". The First AI4.11
it'0orked SensorS)y-stems (SenSys 2003).La, s Angeles CA- Nox. 2003.
protocol for wireless sensor networks." in In Proceedings of the 21st
International .-innuacl Joint Confer-ence of the IEEE Computer canid
Commnunications Societies tINFOCOIf '02). New Y_ork. June 2002.
T. van Darn and K. Langenidoen
MI AC Protocol for WNireless Sensor Networks". The First AC'AI Con-
ference on EmbeddedNAeniaorked Sensor SV'stems (SenSxs 2003), Los
Angeles CA, Nov. 2003.
 L. Charlie zhong. R. Shah. C. C(io. J. Rabaev. ".An Ultra-low Power
and Distributed Access Protocol for Broadband Wireless SenIsoI
Networks". IEEEBroadband ITireless Summit, Las Vegas. May 2001
 Alec Woo and David E. Culler. ".A transmission control scheme for
media access in sensor networks". in Proceeding of the seenth an-
nual international conference on M.Iobile coinputing and netwvorking
200]. Rome. Ital. July 2001. pages 221-235.
 K. Jarnieson. H. Balakrishnan. Y. C. Ta. "'Sift: A MAC Protocol for
Event-Driven Wireless Sensor Networks". LCS Technical Reports.
May 1st. 2003.
 G. Stein. K. Kabitzsch, "A Radio Protocol for Loaw! Power Wireless
Sensor Network-s'. In Proceedings of the J01 IRFC Conference on
Fieldbus Systems cand Their Applications tFET2003). .Aveiro. Por-
tugal, July 7-8. 2003. pages 53-58.
Abstract: WiseMAIC. ani Ultra Loaw Power MAC Protocol for the
WiseNET Wireless Sensor Network". The First ACAIf Conference on
Enihedded N`etwiorked Sensor Systems (SenSgys 2003). Los Angeles
CA. Nov. 2003.
 AAmre El-Hoiydi. "Aloha with Preamble Sampling for Sporadic Traffic
in Ad-hoc Wireless Sensor Networks". IEEE Initernational Confer-
ence on Communications (ICC), New York USTA_ April 2002.
 C. Lu, B. NI. Blum. T. F. Abdelzaher. J. A. Stankovic, T. He. "RAP: A
Real-Time Comnmunication Architecture for Large-Scale Wireless
Sensor Networks". IEEE Real-Timie andEmbedded Technology and
Applications Symposiuni (RTAS 2002), San Jose. California, Sep-
tember 25 - 27. 2002. pages 55-66.
Chandrakasan, "'Physical Layer Driven Protocol and Algorithm De-
sign for Energx-Efficient Wireless Sensor Networks"'. InProceedings
ofACAfMobiCom '01. Rome. Italy. Jul 2001, pages 272-286.
 SeongHwani Cho. Anantha P. Chandrakasan, "Energy! Efficient Pro-
tocols for Low Duty Cycle Wireless Microsensor Networks", ICASSP
2001, May 2001.
 C. Guo, L. Charlie Zhong, J. M. Rabaex'. "Loaw Power Distributed
MAC for Ad hoc Sensor Radio Networks", In Proceedings of EEE
GlobeCom2001, San Antonio, November 25-29, 2001.
 M. Younis. K.Akkaya. M. Eltoweissy, .A. Wadaa, "On Handling QoS
Traffic in Wireless Sensor Networks", in the Proceedings of the In-
ternationalConferenceHAUAIIInternational Conference on System
Sciences(HICSS-37), BigIsland, Hawaii, January 2004.
J. Gao. V. Ailawadhi.
of a wireless sensor network". IEEE Personal
Potlie. "Protocols for
Las x'egas. Nevada.
P. Havinga. "Collaborative
Lui Sha. Giorgio Buittazzo, "An
J.J. Garcia-Luna-Aceves. "En-
Conj1rence on Einbedded Net-
J. Heidemaiin and D. Estrin.
"'An .Adaptive Energy-Efficient
A. EI-Hoiydi. J.-D. Decotignie. C. Enz anid E. Le Roux. "Poster
S. Cho. N.
Ickes. R. Min,
A. Siiha. .A. Wang,