WISENET

(WIRELESS SENSOR NETWORK)

ABSTRACT
WISENET is a wireless sensor network that monitors the environmental
conditions such as light, temperature, and humidit! This network is comprised o" nodes
called #motes$ that "orm an ad%hoc network to transmit this data to a computer that
"unction as a server! The server stores the data in a data&ase where it can later &e
retrieved and anal'ed via a we&%&ased inter"ace! The network works success"ull with
an implementation o" one sensor mote!

Introduction:
The "irst goal o" WISENET is to create a new hardware plat"orm to
take advantage o" newer microcontrollers with greater "unctionalit and more "eatures!
This involves selecting the hardware, designing the motes, and porting Tin(S! (nce the
plat"orm is completed and Tin(S was ported to it, the ne)t stage is to use this plat"orm
to create a small%scale sstem o" wireless networked sensors!
Wireless sensor%actuator networks can provide the a&ilit to continuousl monitor
the integrit o" structures in real%time, detect damage at an earl stage, and provide
ro&ustness in the case o" catastrophic "ailures with a "raction o" cost associated with
toda*s wired networks! +owever, sensor%actuator networks re,uire a new paradigm o"
computing-one, which e)plicitl addresses less capa&le hardware, unrelia&le
communication with, limited &andwidth, and severe energ constraints! The algorithms
and so"tware tools will "acilitate monitoring and protection o" civil structures using such
networks!
Smart environments represent the ne)t evolutionar development step in &uilding,
utilities, industrial, home, ship&oard, and transportation sstems automation! .ike an
sentient organism, the smart environment relies "irst and "oremost on sensor data "rom
the real world! Sensor data comes "rom multiple sensors o" di""erent modalities in
distri&uted locations! The smart environment needs in"ormation a&out its surroundings as
well as a&out its internal workings/ this is captured in &iological sstems & the
distinction &etween exteroceptors and proprioceptors.
The challenges in the hierarch o"0 detecting the relevant ,uantities, monitoring
and collecting the data, assessing and evaluating the in"ormation, "ormulating meaning"ul
user displas, and per"orming decision%making and alarm "unctions are enormous! The
in"ormation needed & smart environments is provided & 1istri&uted Wireless Sensor
Networks, which are responsi&le "or sensing as well as "or the "irst stages o" the
processing hierarch! The importance o" sensor networks is highlighted & the num&er o"
recent "unding initiatives, including the 1AR2A SENSIT program, militar programs,
and NS3 2rogram Announcements!
The "igure shows the comple)it o" wireless sensor networks, which generall
consist o" a data ac,uisition network and a data distri&ution network, monitored and
controlled & a management center! The plethora o" availa&le technologies makes even
the selection o" components di""icult, let alone the design o" a consistent, relia&le, ro&ust
overall sstem!
The stud o" wireless sensor networks is challenging in that it re,uires an enormous
&readth o" knowledge "rom an enormous variet o" disciplines! In this chapter we outline
communication networks, wireless sensor networks and smart sensors, phsical
transduction principles, commerciall availa&le wireless sensor sstems, sel"organi'ation,
signal processing and decision%making, and "inall some concepts "or home automation!
COMMUNICATION NETWORKS
The stud o" communication networks can encompass several ears at the college or
universit level! To understand and &e a&le to implement sensor networks, however,
several &asic primar concepts are su""icient!
Network Topolo!
The &asic issue in communication networks is the transmission o" messages to achieve a
prescri&ed message throughput 45uantit o" Service6 and 5ualit o" Service 45oS6! 5oS
can &e speci"ied in terms o" message dela, message due dates, &it error rates, packet
loss, economic cost o" transmission, transmission power, etc! 1epending on 5oS, the
installation environment, economic considerations, and the application, one o" several
&asic network topologies ma &e used!
A communication network is composed o" nodes, each o" which has computing
power and can transmit and receive messages over communication links, wireless or
ca&led! The &asic network topologies are shown in the "igure and include "ull connected,
mesh, star, ring, tree, &us! A single network ma onsist o" several interconnected su&nets
o" di""erent topologies! Networks are "urther classi"ied as .ocal Area Networks 4.AN6,
e!g! inside one &uilding, or Wide Area Networks 4WAN6, e!g! &etween &uildings!
"ull! connected network# su""er "rom pro&lems o" N2%comple)it 78are 9:;:</ as
additional nodes are added, the num&er o" links increases e)ponentiall! There"ore, "or
large networks, the routing pro&lem is computationall intracta&le even with the
availa&ilit o" large amounts o" computing power!
Me#$ network# are regularl distri&uted networks that generall allow transmission onl
to a node*s nearest neigh&ors! The nodes in these networks are generall identical, so that
mesh nets are also re"erred to as peer%to%peer 4see &elow6 nets! =esh nets can &e good
models "or large%scale networks o" wireless sensors that are distri&uted over a geographic
region, e!g! personnel or vehicle securit surveillance sstems! Note that the regular
structure re"lects the communications topolog/ the actual geographic distri&ution o" the
nodes need not &e a regular mesh! Since there are generall multiple routing paths
&etween nodes, these nets are ro&ust to "ailure o" individual nodes or links! An advantage
o" mesh nets is that, although all nodes ma &e identical and have the same computing
and transmission capa&ilities, certain nodes can &e designated as >group leaders* that take
on additional "unctions! I" a group leader is disa&led, another node can then take over
these duties!
All nodes o" the #t%r topolo! are connected to a single hu& node! The hu&
re,uires greater message handling, routing, and decision%making capa&ilities than the
other nodes! I" a communication link is cut, it onl a""ects one node! +owever, i" the hu&
is incapacitated the network is destroed! In the rin topolo! all nodes per"orm the
same "unction and there is no leader node! =essages generall travel around the ring in a
single direction! +owever, i" the ring is cut, all communication is lost! The #el&'$e%lin
rin network 4S+R6 shown has two rings and is more "ault tolerant!
In the (u# topolo!, messages are &roadcast on the &us to all nodes! Each node
checks the destination address in the message header, and processes the messages
addressed to it! The &us topolog is passive in that each node simpl listens "or messages
and is not responsi&le "or retransmitting an messages!
Co))unic%tion *rotocol# %nd Routin
The topics o" communication protocols and routing are comple) and re,uire much stud!
Some &asics use"ul "or understanding sensor nets are presented here! +e%der#, Each
message generall has a header identi"ing its source node, destination node, length o"
the data "ield, and other in"ormation! This is used & the nodes in proper routing o" the
message! In encoded messages, parit &its ma &e included! In packet routing networks,
each message is &roken into packets o" "i)ed length! The packets are transmitted
separatel through the network and then reassem&led at the destination! The "i)ed packet
length makes "or easier routing and satis"action o" 5oS! 8enerall, voice
communications use circuit switching, while data transmissions use packet routing!
separatel through the network and then reassem&led at the destination! The "i)ed packet
length makes "or easier routing and satis"action o" 5oS! 8enerall, voice
communications use circuit switching, while data transmissions use packet routing!
In addition to the in"ormation content messages, in some protocols 4e!g! 311I%
see &elow6 the nodes transmit special frames to report and identi" "ault conditions! This
can allow network recon"iguration "or "ault recover! (ther special "rames might include
route discover packets or ferrets that "low through the network, e!g! to identi" shortest
paths, "ailed links, or transmission cost in"ormation! In some schemes, the "erret returns
to the source and reports the &est path "or message transmission!
When a node desires to transmit a message, handshaking protocols with the
destination node are used to improve relia&ilit! The source and destination might
transmit alternatel as "ollows0 re,uest to send, read to receive, send message, message
received! +andshaking is used to guarantee 5oS and to retransmit messages that were not
properl received!
Switc$in
=ost computer networks use a store-and-forward switching techni,ue to control the
"low o" in"ormation! Then, each time a packet reaches a node, it is completel &u""ered in
local memor, and transmitted as a whole! =ore sophisticated switching techni,ues
include wormhole, which splits the message into smaller units known as "low control
units or "lits! The header "lit determines the route! As the header is routed, the remaining
"lits "ollow it in pipeline "ashion! This techni,ue currentl achieves the lowest message
latenc! Another popular switching scheme is virtual-cut-through! +ere, when the header
arrives at a node, it is routed without waiting "or the rest o" the packet! 2ackets are
&u""ered either in so"tware &u""ers in memor or in hardware &u""ers, and various sorts o"
&u""ers are used including edge &u""ers, central &u""ers, etc!
Multiple Acce## *rotocol#! When multiple nodes desire to transmit, protocols are
needed to avoid collisions and lost data! In the A.(+A scheme, "irst used in the 9:;?*s
at the @niversit o" +awaii, a node simpl transmits a message when it desires! I" it
receives an acknowledgement, all is well! I" not, the node waits a random time and
retransmits the message!
In Frequency Division Multiple Access 431=A6, di""erent nodes have di""erent
carrier "re,uencies! Since "re,uenc resources are divided, this decreases the &andwidth
availa&le "or each node! 31=A also re,uires additional hardware and intelligence at each
node! In Code Division Multiple Access 4C1=A6, a uni,ue code is used & each node to
encode its messages! This increases the comple)it o" the transmitter and the receiver! In
ime Division Multiple Access 4T1=A6, the R3 link is divided on a time a)is, with each
node &eing given a predetermined time slot it can use "or communication! This decreases
the sweep rate, &ut a maAor advantage is that T1=A can &e implemented in so"tware! All
nodes re,uire accurate, snchroni'ed clocks "or T1=A!
Open S!#te)# Interconnection Re&erence Model (OSI-RM),
The International Standards (rgani'ation 4IS(6 (SIBR= architecture speci"ies the
relation &etween messages transmitted in a communication network and applications
programs run & the users! The development o" this open standard has encouraged the
adoption & di""erent developers o" standardi'ed compati&le sstems inter"aces! The
"igure shows the seven laers o" (SIBR=! Each laer is sel"%contained, so that it can &e
modi"ied without undul a""ecting other laers! The Transport .aer provides error
detection and correction! Routing and "low control are per"ormed in the Network .aer!
The 2hsical .aer represents the actual hardware communication link interconnections!
The Applications .aer represents programs run & users!
Routin, Since a distri&uted network has multiple nodes and services man messages,
and each node is a shared resource, man decisions must &e made! There ma &e multiple
paths "rom the source to the destination! There"ore, message routing is an important topic!
The main per"ormance measures a""ected & the routing scheme are throughput 4,uantit
o" service6 and average packet dela 4,ualit o" service6! Routing schemes should also
avoid &oth deadlock and livelock 4see &elow6!
Routing methods can &e "i)ed 4i!e! pre%planned6, adaptive, centrali'ed, distri&uted,
&roadcast, etc! 2erhaps the simplest routing scheme is the token ring 7Smthe 9:::<!
+ere, a simple topolog and a straight"orward "i)ed protocol result in ver good
relia&ilit and precomputa&le 5oS! A token passes continuousl around a ring topolog!
When a node desires to transmit, it captures the token and attaches the message! As the
token passes, the destination reads the header, and captures the message! In some
schemes, it attaches a >message received* signal to the token, which is then received &
the original source node! Then, the token is released and can accept "urther messages!
The token ring is a completel decentrali'ed scheme that e""ectivel uses T1=A!
Though this scheme is ver relia&le, one can see that it results in a waste o" network
capacit! The token must pass once around the ring "or each message! There"ore, there
are various modi"ications o" this scheme, including using several tokens, etc!
"i.ed routin #c$e)e# o"ten use Routing Ta&les that dictate the ne)t node to &e
routed to, given the current message location and the destination node! Routing ta&les can
&e ver large "or large networks, and cannot take into account real%time e""ects such as
"ailed links, nodes with &acked up ,ueues, or congested links!
Ad%pti/e routin #c$e)e# depend on the current network status and can take
into account various per"ormance measures, including cost o" transmission over a given
link, congestion o" a given link, relia&ilit o" a path, and time o" transmission! The can
also account "or link or node "ailures!
Routing algorithms can &e &ased on various network analsis and graph theoretic
concepts in Computer Science 4e!g! A%star tree search6, or in (perations Research
7Bronson 9::;< including shortest%route, ma)imal "low, and minimum%span pro&lems!
Routing is closel associated with dnamic programming and the optimal control
pro&lem in "eed&ack control theor! Shortest 2ath routing schemes "ind the shortest path
"rom a given node to the destination node! I" the cost, instead o" the link length, is
associated with each link, these algorithms can also compute minimum cost routes! These
algorithms can &e centrali'ed 4"ind the shortest path "rom a given node to all other nodes6
or decentrali'ed 4"ind the shortest path "rom all nodes to a given node6! There are certain
well%de"ined algorithms "or shortest path routing, including the e""icient 1iAkstra
algorithm , which has polnomial comple)it! The Bellman%3ord algorithm "inds the path
with the least num&er o" hops! Routing schemes &ased on competitive game theoretic
notions have also &een developed !
0e%dlock %nd Li/elock, .arge%scale communication networks contain ccles 4circular
paths6 o" nodes! =oreover, each node is a shared resource that can handle multiple
messages "lowing along di""erent paths! There"ore, communication nets are suscepti&le to
deadlock, wherein all nodes in a speci"ic ccle have "ull &u""ers and are waiting "or each
other! Then, no node can transmit &ecause no node can get "ree &u""er space, so all
transmission in that ccle comes to a halt! !ivelock, on the other hand, is the condition
wherein a message is continuall transmitted around the network and never reaches its
destination! .ivelock is a de"icienc o" some routing schemes that route the message to
alternate links when the desired links are congested, without taking into account that the
message should &e routed closer to its "inal destination! =an routing schemes are
availa&le "or routing with deadlock and livelock avoidance 7e!g! 1uato 9::C<!
"low Control, In ,ueuing networks, each node has an associated ,ueue or &u""er that can
stack messages! In such networks, "low control and resource assignment are important!
The o&Aectives o" "low control are to protect the network "rom pro&lems related to
overload and speed mismatches, and to maintain 5oS, e""icienc, "airness, and "reedom
"rom deadlock! I" a given node A has high priorit, its messages might &e pre"erentiall
routed in ever case, so that competing nodes are choked o"" as the tra""ic o" A increases!
3air routing schemes avoid this! There are several techni,ues "or "low control0 In "uffer
management, certain portions o" the &u""er space are assigned "or certain purposes! In
choke packet schemes, an node sensing congestion sends choke packets to other nodes
telling them to reduce their transmissions! #sarithmic schemes have a "i)ed num&er o"
>permits* "or the network! A message can &e sent onl i" a permit is availa&le! In window
or kan"an schemes, the receiver grants >credits* to the sender onl i" it has "ree &u""er
space! @pon receiving a credit, the sender can transmit a message! In ransmission
Control $rotocol %C$& schemes 4Tahoe and Reno6 a source linearl increases its
transmission rate as long as all its sent messages are acknowledged "or! When it detects a
lost packet, it e)ponentiall decreases its transmission rate! Since lost packets depend on
congestion, TC2 automaticall decreases transmissions when congestion is etected!
*ower M%n%e)ent
With the advent o" ad hoc networks o" geographicall distri&uted sensors in remote site
environments 4e!g! sensors dropped "rom aircra"t "or personnelBvehicle surveillance6,
there is a "ocus on increasing the li"etimes o" sensor nodes through power generation,
power conservation, and power management! Current research is in designing small
=E=S 4microelectromechanical sstems6 R3 components "or transceivers, including
capacitors, inductors, etc! The limiting "actor now is in "a&ricating microsi'ed inductors!
Another thrust is in designing =E=S power generators using technologies including
solar, vi&ration 4electromagnetic and electrostatic6, thermal, etc!
R3%I1 4R3 identi"ication6 devices are transponder microcircuits having an .%C
tank circuit that stores power "rom received interrogation signals, and then uses that
power to transmit a response! 2assive tags have no on&oard power source and limited
on&oard data storage, while active tags have a &atter and up to 9=& o" data storage! R3%
I1 operates in a low "re,uenc range o" 9??k+'%9!D=+' or a high "re,uenc range o"
:?? =+'%E!F8+', which has an operating range up to G?m! R3%I1 tags are ver
ine)pensive, and are used in manu"acturing and sales inventor control, container
shipping control, etc! R3%I1 tags are installed on water meters in some cities, allowing a
metering vehicle to simpl drive & and remotel read the current readings! The are also
&e used in automo&iles "or automatic toll collection!
=eanwhile, so"tware power management techni,ues can greatl decrease the
power consumed & R3 sensor nodes! T1=A is especiall use"ul "or power conservation,
since a node can power down or >sleep* &etween its assigned time slots, waking up in
time to receive and transmit messages!
The re,uired transmission power increases as the s,uare o" the distance &etween
source and destination! There"ore, multiple short message transmission hops re,uire less
power than one long hop! In "act, i" the distance &etween source and destination is ', the
power re,uired "or single%hop transmission is proportional to '(! I" nodes &etween source
and destination are taken advantage o" to transmit n short hops instead, the power
re,uired & each node is proportional to '()n(. This is a strong argument in "avor o"
distri&uted networks with multiple nodes, i!e! nets o" the mesh variet!
A current topic o" research is active power control* where& each node cooperates
with all other nodes in selecting its individual transmission power level 7Humar E??9<!
This is a decentrali'ed "eed&ack control pro&lem! Congestion is increased i" an node
uses too much power, &ut each node must select a large enough transmission range that
the network remains connected! 3or n nodes randoml distri&uted in a disk, the network
is asmptoticall connected with pro&a&ilit one i" the transmission range r o" all nodes
is selected using
1,1,2, Network Structure %nd +ier%rc$ic%l Network#
Routing ta&les "or distri&uted networks increase e)ponentiall as nodes are added! An n )
m mesh network has nm links, and there are multiple paths "rom each source to each
destination! +ierarchical network structures simpli" routing, and also are amena&le to
distri&uted signal processing and decision%making, since some processing can &e done at
each hierarchical laer!
It has &een shown 7.ewis and A&dallah 9::G< that a "ull connected network has
,$%hard comple)it, while imposing routing protocols & restricting the allowed paths to
o&tain a reentrant flow topolog results in polnomial comple)it! Such streamlined
protocols are natural "or hierarchical networks!
Multic%#t S!#te)# in mesh networks use a hierarchical leader%&ased scheme "or message
transmission 7Chen et al! E???<! Each group o" nodes has a designated leader that is
responsi&le "or receiving messages "rom and transmitting to nodes outside the group! 2art
4a6 o" the "igure shows messages routed in a mesh net using standard peer%to%peer
protocols! The link lengths o" the transmission paths are shown! 2arts 4&6 and 4c6 show
the same two messages &eing routed using a multicast protocol! Note that the total
transmission paths are signi"icantl shorter! =ulticast has &een implemented using tree
&ased and path%&ased schemes!
+ier%rc$ic%l Network#, =uch work has &een done on "ormal hierarchical structures "or
distri&uted networks! Cao 79:::< studies how to determine optimal con"igurations "or
hierarchical routing! Shi 79::D< anal'es hierarchical sel"healing rings! Shah%+edari
7E??9< shows the importance o" a consistent num&ering scheme in hierarchical sstems,
which allows "or a simpli"ied tree&ased routing scheme!
The "igure shows a &asic F% element ring element consisting o" "our nodes and
"our links! It shows
two was o" connecting these two rings, which results in two mesh networks o" di""erent
structures! The "irst network consists o" alternating one%wa streets, while the second
consists o" alternatingdirection vortices! It is interesting to anal'e these two structures
"rom the point o" view o" the notions o" "low "ield divergence and curl!
In an network, the phenomenon o" edge "inding means that much o" the routing
power o" peripheral stations is wasted &ecause peripheral links are unused! Thus,
messages tend to re"lect o"" the &oundar into the interior or to move parallel to the
peripher 7I!W! Smith, Rand Corp! 9:CF<! To avoid this, the =anhattan geometr
connects the nodes at one edge o" the network to nodes at the opposite edge! The "igure
shows the standard =anhattan geometr as well as a =anhattan net &uilt "rom the
alternating one%wa street mesh Aust constructed!
As nodes are added, the num&er o" links increases e)ponentiall! This makes "or
N2%comple)it pro&lems in routing and "ailure recover! To simpli" network structure,
we can use hierarchical clustering techni,ues! The hierarchical structure must &e
consistent, that is, it must have the same structure at each level! The "igure shows a F)F
mesh net and also a clustering into "our groups! Note that the clustered structure has a
dual ring S+R topolog! To reduce the routing comple)it, we can disa&le one o"
the rings and o&tain a ring structure!
The ne)t "igure shows an J)J mesh net! Shown "irst are all the links, and then the
hierarchical clustering with some links disa&led to reduce comple)it! We have chosen to
keep the outer ring at each level! Note that the clockwise ring structure is the same at
each level, resulting in a regular hierarch!
Routing is ver eas in this hierarchical network 7Swam E??G<! 3irst, one selects
a consistent num"ering scheme! 3or e)ample num&er the groups as 9,E,G,F &eginning in
the top le"t and going clockwise! This is done at each level! Then, re"erring to the J)J
mesh net in the "igure, node 9FG, shown in the "igure, is in the top le"t F)F group, within
which it is in the "ourth E)E group, within which it is the third node! @sing this num&er
scheme one ma construct a simple routing scheme wherein the same &asic routing
algorithm is repeated at each level o" the hierarch! This is not unlike ,uadtree routing in
mo&ile ro&ot path planning! 3ailure recover is also straight"orward! I" a link "ails, one
ma simpl switch in one o" the disa&led links to take over! Code "or this is ver eas to
write!
0i#tri(uted Routin3 0eci#ion'M%kin3 %nd 0S*, It is natural in routing and "ailure
recover "or these hierarchical networks to designate the entr node "or each group as a
group leader. This node must make additional decisions &eond those o" the other nodes,
including resource availa&ilit "or deadlock voidance, disa&led link activation "or "ailure
recover, and so on! This las a ver natural "ramework "or distri&uted decision%making
and digital signal processing 41S26, wherein a group leader processes the data "rom the
group prior to transmitting it! The group leader "or communications should &e the entry
node o" each group, while the group leader "or 1S2 should &e the exit node "or each
group!
! +i#toric%l 0e/elop)ent %nd St%nd%rd#
=uch o" this in"ormation is taken "rom 72C Tech 8uide<, which contains a thorough
summar o" communication network standards, topologies, and components! See also
Iordan and A&dallah 7E??E<!
Et$ernet! The Ethernet was developed in the mid 9:;?*s & Kero), 1EC, and Intel, and
was standardi'ed in 9:;:! The Institute o" Electrical and Electronics Engineers 4IEEE6
released the o""icial Ethernet standard IEEE J?E!G in 9:JG! The 3ast Ethernet operates at
ten times the speed o" the regular Ethernet, and was o""iciall adopted in 9::D! It
introduces new "eatures such as "ull%duple) operation and auto%negotiation! Both these
standards use IEEE J?E!G varia&le%length "rames having &etween CF and 9D9F%&te
packets!
Token Rin! In 9:JF IB= introduced the F=&itBs token ring network! The sstem was o"
high ,ualit and ro&ust, &ut its cost caused it to "all &ehind the Ethernet in popularit!
IEEE standardi'ed the token ring with the IEEE J?E!D speci"ication! The 3i&er
1istri&uted 1ata Inter"ace 4311I6 speci"ies a 9??=&itBs token%passing, dual%ring .AN
that uses "i&er optic ca&le! It was developed & the American National Standards Institute
4ANSI6 in the mid 9:J?s, and its speed "ar e)ceeded current capa&ilities o" &oth Ethernet
and IEEE J?E!D!
4i%(it Et$ernet! The 8iga&it Ethernet Alliance was "ounded in 9::C, and the 8iga&it
Ethernet standards were rati"ied in 9:::, speci"ing a phsical laer that uses a mi)ture
o" technologies "rom the original Ethernet and "i&er optic ca&le technologies "rom 311I!
Client'Ser/er networks &ecame popular in the late 9:J?*s with the replacement o" large
main"rame computers & networks o" personal computers! Application programs "or
distri&uted computing environments are essentiall divided into two parts0 the client or
"ront end, and the server or &ack end! The user*s 2C is the client and more power"ul
server machines inter"ace to the network!
*eer'to'*eer networkin architectures have all machines with e,uivalent capa&ilities
and responsi&ilities! There is no server, and computers connect to each other, usuall
using a &us topolog, to share "iles, printers, Internet access, and other resources!
*eer'to'*eer Co)putin is a signi"icant ne)t evolutionar step over 2E2 networking!
+ere, computing tasks are split &etween multiple computers, with the result &eing
assem&led "or "urther consumption! 2E2 computing has sparked a revolution "or the
Internet Age and has o&tained considera&le success in a ver short time! The Napster
=2G music "ile sharing application went live in Septem&er 9:::, and attracted more than
E? million users & mid E???!
561,77 Wirele## Loc%l Are% Network, IEEE rati"ied the IEEE J?E!99 speci"ication in
9::; as a standard "or W.AN! Current versions o" J?E!99 4i!e! J?E!99&6 support
transmission up to 99=&itBs! Wi3i, as it is known, is use"ul "or "ast and eas networking
o" 2Cs, printers, and other devices in a local environment, e!g! the home! Current 2Cs and
laptops as purchased have the hardware to support Wi3i! 2urchasing and installing a
Wi3i router and receivers is within the &udget and capa&ilit o" home 2C enthusiasts!
8luetoot$ was initiated in 9::J and standardi'ed & the IEEE as Wireless 2ersonal Area
Network 4W2AN6 speci"ication IEEE J?E!9D! Bluetooth is a short range R3 technolog
aimed at "acilitating communication o" electronic devices &etween each other and with
the Internet, allowing "or data snchroni'ation that is transparent to the user! Supported
devices include 2Cs, laptops, printers, Aosticks, ke&oards, mice, cell phones, 21As, and
consumer products! =o&ile devices are also supported! 1iscover protocols allow new
devices to &e hooked up easil to the network! Bluetooth uses the unlicensed E!F 8+'
&and and can transmit data up to 9=&itBs, can penetrate solid non%metal &arriers, and has
a nominal range o" 9?m that can &e e)tended to 9??m! A master station can service up to
; simultaneous slave links! 3orming a network o" these networks, e!g! a piconet, can
allow one master to service up to E?? slaves!
Currentl, Bluetooth development kits can &e purchased "rom a variet o"
suppliers, &ut the sstems generall re,uire a great deal o" time, e""ort, and knowledge
"or programming and de&ugging! 3orming piconets has not et &een streamlined and is
undul di""icult!
+o)e R" was initiated in 9::J and has similar goals to Bluetooth "or W2AN! Its goal is
shared dataBvoice transmission! It inter"aces with the Internet as well as the 2u&lic
Switched Telephone Network! It uses the E!F 8+' &and and has a range o" D? m, suita&le
"or home and ard! A ma)imum o" 9E; nodes can &e accommodated in a single network!
Ir0A is a W2AN technolog that has a short%range, narrow%transmission%angle &eam
suita&le "or aiming and selective reception o" signals!
, WIRELESS SENSOR NETWORKS
Sensor networks are the ke to gathering the in"ormation needed & smart environments,
whether in &uildings, utilities, industrial, home, ship&oard, transportation sstems
automation, or elsewhere! Recent terrorist and guerilla war"are countermeasures re,uire
distri&uted networks o" sensors that can &e deploed using, e!g! aircra"t, and have sel"%
organi'ing capa&ilities! In such applications, running wires or ca&ling is usuall
impractical! A sensor network is re,uired that is "ast and eas to install and maintain!
IEEE 7297 %nd S)%rt Sen#or#
Wireless sensor networks satis" these re,uirements! 1esira&le "unctions "or sensor nodes
include0 ease o"
installation, sel"%identi"ication, sel"%diagnosis, relia&ilit, time awareness "or coordination
with other nodes, some so"tware "unctions and 1S2, and standard control protocols and
network inter"aces 7IEEE 9FD9 E)po, E??9<! There are man sensor manu"acturers and
man networks on the market toda! It is too costl "or manu"acturers to make special
transducers "or ever network on the
There are man sensor manu"acturers and man networks on the market toda! It
is too costl "or manu"acturers to make special transducers "or ever network on the
market! 1i""erent components made & di""erent manu"acturers should &e compati&le!
There"ore, in 9::G the IEEE and the National Institute o" Standards and Technolog
4NIST6 &egan work on a standard "or Smart Sensor Networks! IEEE 9FD9, the Standard
"or Smart Sensor Networks was the result! The o&Aective o" this standard is to make it
easier "or di""erent manu"acturers to develop smart sensors and to inter"ace those devices
to networks!
S)%rt Sen#or3 :irtu%l Sen#or, =aAor components include STI=, TE1S, TII, and
NCA2 as detailed in the "igure! A maAor outcome o" IEEE 9FD9 studies is the "ormali'ed
concept o" a -mart -ensor! A smart sensor is a sensor that provides e)tra "unctions
&eond those necessar "or generating a correct representation o" the sensed ,uantit
73rank E???<! Included might &e signal conditioning, signal processing, and decision%
makingBalarm "unctions! A general model o" a smart sensor is shown in the "igure!
(&Aectives "or smart sensors include moving the intelligence closer to the point o"
measurement/ making it cost e""ective to integrate and maintain distri&uted sensor
sstems/ creating a con"luence o" transducers, control, computation, and communications
towards a common goal/ and seamlesslinter"acing numerous sensors o" di""erent tpes!
The concept o" a Lirtual -ensor is also depicted! A virtual sensor is the phsical
sensorBtransducer, plus the associated signal conditioning and digital signal processing
41S26 re,uired to o&tain relia&le estimates o" the re,uired sensor in"ormation! The
virtual sensor is a component o" the smart sensor!
Tr%n#ducer# %nd *$!#ic%l Tr%n#duction *rinciple#
A transducer is a device that converts energ "rom one domain to another! In our
application, it converts the ,uantit to &e sensed into a use"ul signal that can &e directl
measured and processed! Since much signal conditioning 4SC6 and digital signal
processing 41S26 is carried out & electronic circuits, the outputs o" transducers that are
use"ul "or sensor networks are generall voltages or currents! Sensor transduction ma
&e carried out using physical principles* some o" which we review here!
=icroelectromechanical Sstems 4=E=S6 sensors are & now ver well developed and
are availa&le "or most sensing applications in wireless networks! Re"erences "or this
section include 3rank 7E???<, Hovacs 79::J<, =adou 79::;<, de Silva 79:::<!
Mec$%nic%l Sen#or# include those that rel on direct phsical contact!
T$e *ie;ore#i#ti/e E&&ect converts an applied strain to a change in resistance that can &e
sensed using electronic circuits such as the Wheatstone Bridge 4discussed later6!
1iscovered & .ord Helvin in 9JDC, the relationship is , . - ' ' M M B with ' the
resistance, M the strain, and - the gauge "actor which depends on ,uantities such as the
resistivit and the 2oisson* ratio o" the material! There ma &e a ,uadratic term in M "or
some materials! =etals and semiconductors e)hi&it pie'oresistivit! The pie'oresistive
e""ect in silicon is enhanced & doping with &oron 4ptpe silicon can have a gauge "actor
up to E??6! With semiconductor strain gauges, temperature compensation is important!
T$e *ie;oelectric E&&ect, discovered & the Curies in 9JJ?, converts an applied
stress 4"orce6 to a charge separation or potential di""erence! 2ie'oelectric materials
include &arium titanate, 2NT, and single%crstal ,uart'! The relation &etween the change
in "orce F and the change in voltage / is given & M , where k is proportional to the
material charge sensitivit coe""icients and the crstal thickness, and inversel
proportional to the crstal area and the material relative permittivit! The pie'oelectric
e""ect is reversi&le, so that a change in voltage also generates a "orce and a corresponding
change in thickness! Thus the same device can &e "oth a sensor and an actuator.
Com&ined sensorBactuators are an intriguing topic o" current research! F k / M M
Tunnelin Sen#in depends on the e)ponential relationship &etween the
tunneling current # and the tipBsur"ace separation 0 given & , where k depends on the
tunnel &arrier height in ev! Tunneling is an e)tremel accurate method o" sensing
nanometer%scale displacements, &ut its highl nonlinear nature re,uires the use o"
"eed&ack control to make it use"ul!
S!#te) 0e#cription:
There are two primar su&sstems 41ata Analsis and 1ata
Ac,uisition6 comprised o" three maAor components 4Client, Server, Sensor =ote
Network6!
*ri)%r! Su(#!#te)#:
There are two top%level su&sstems O
Data Analysis
Data Acquisition.
0%t% An%l!#i#:
This su&sstem is so"tware%onl 4relative to WISENET6! It relied on e)isting
Internet and we& 4+TT26 in"rastructure to provide communications &etween the Client and Server
components! The "ocus o" this su&sstem was to selectivel present the collected environmental data
to the end user in a graphical manner!
0%t% Ac<ui#ition:
The purpose o" this su&sstem is to collect and store environmental data "or later
processing & the 1ata Analsis su&sstem! This is a mi) o" &oth 2C P em&edded sstem so"tware, as
well as em&edded sstem hardware! It is composed o" &oth the Server and Sensor =ote Network
components!
S!#te) Co)ponent#:
Sstem components are Client* -erver* and -ensor Mote ,etwork.
CLIENT SER:ER SENSOR MOTE NETWORK

TC2BI2
Internet
Sstem
We&
Browser
+TT2
RSEGE SERIA.

TC2BI2
TC2BI2
:J?=+N R3 Comm!
+TT2
Server
We&
2rogram
S5.
1ata&ase
Wise
1B
8atewa
.a& A .a& B
(""ice9
(""iceE
0%t% An%l!#i# Su(#!#te) 0%t% Ac<ui#ition Su(#!#te)
"iure 7: WISENET S!#te) 8lock 0i%r%)
Client:
The Client component is necessar &ut e)ternal to the development o" WISENET!
That is, an computer with a we& &rowser and Internet access could &e a Client! It served onl as a
user inter"ace to the 1ata Analsis su&sstem!
USER SER:ER
Re<ue#t# WE8 p%e Re<ue#ted WE8
p%e
USER SER:ER
Re<ue#ted WE8 p%e Re<ue#t# WE8 p%e
"iure 1: Client Co)ponent Input#-Output#
Ser/er:
The Server is a critical component as the link &etween the 1ata Ac,uisition and
1ata Analsis su&sstems! (n the 1ata Analsis side, an we& 4+TT26 server hosting a we&
application! When a page re,uest came in, the we& server e)ecutes the we& application, which
retrieved data "rom the data&ase, processes it, and returns a we& page that the we& server
transmitted to the Client! 3or the 1ata Ac,uisition sstem there is a daemon 4Wise1B6 running to
"acilitate communication with the Sensor =ote Network!
CLIENT SENSOR NETWORK
WE8 p%e Re<ue#t# 0%t% p%cket#
4:i% 4ATEWA= MOTE)
CLIENT SENSOR NETWORK
CLIENT
Input# >
Output#
SER:ER
Input# >
Output#
Re<ue#ted WE8 p%e Co))%nd#
"iure ?: Ser/er Co)ponent# Input#-Output#
This daemon is responsi&le "or collecting raw data packets "rom the Sensor
=ote Network! These packets are then processed to convert the raw data into meaning"ul
environmental data! This processed data is then inserted into the data&ase! Thus the data&ase is
the link &etween the 1ata Analsis and 1ata Ac,uisition su&sstems! The Server also had the
potential to send commands to the Sensor =ote Network 4via the gatewa mote6, although this
"unctionalit was not e)plored in WISENET!
It should &e noted that since the S5. data&ase connections can &e made via
TC2BI2, onl the we& server and we&%program 4see "igure F6 needed to &e located on the same
phsical machine! The we& server, the data&ase, and Wise1B could all &e on di""erent phsical
machines connected via a .AN or the Internet! This allows a "le)i&le Server component
implementation that is use"ul during WISENET development!
CLIENT
SENSOR NETWORK

(4ATEWA= MOTE)
TC2BI2 TC2BI2
S5.
1ata&ase
Tin(S
1aemon
WISE1B
WEB
2rogram
+TT2
Server
"iure 2: Ser/er Co)ponent 8lock 0i%r%)
Sen#or Mote#:
The primar "ocus o" WISENET is the development o" the Sensor =ote
Network component! It is the component responsi&le "or collecting and transmitting raw
environmental data to the Server! There is also the potential "or the motes to receive commands
"rom the Server, although that "unctionalit ma not &e implemented in WISENET! @ses "or this
"eature would include server%&ased snchroni'ation and wireless network reprogramming!

This component consists o" two parts! The "irst is the sensor mote! The primar
purpose o" the sensor mote is to collect and transmit raw environmental data! When not doing this, it
went into a low%power idle mode to conserve energ! Another aspect o" the sensor motes involved ad%
hoc networking and ma &e "or multi%hop routing/
The gatewa mote is the second part o" the Sensor =ote Network! Its purpose is to
serve as the liaison &etween the Server and the Sensor =ote Network and deliver all the data packets
to Wise1B! In theor &oth standard and gatewa motes could &e implemented on the same hardware
2CB and with the same so"tware! 3or WISENET, however, resource and time constraints necessitated
the use o" slightl di""erent hardware and so"tware con"igurations "or gatewa versus standard motes,
as descri&ed &elow!
+%rdw%re 0e#in:
The selection o" components "or the sensor motes is a critical process in
the development o" WISENET! 8reat "unctionalit and low power are two o" the highest
priorities in evaluating the "itness o" &oth the microcontroller and the sensor candidates!
WISENET is introduced to the new state%o"%the%art Chipcon CC9?9? microcontroller
with integrated R3 transceiver! A"ter a little research it was decided the CC9?9? would
make the per"ect microcontroller!
It had the "ollowing "eature list0
9! (ptimi'ed J?D9%core
E! Active 49F!J mA6, Idle 4E: MA6 and sleep 4?!E MA6 power modes
G! GE kB "lash memor
F! E kB Q9EJ &tes SRA=
D! Three channel 9?%&it A1C
C! 3our timers B Two 2W=Rs
;! +ardware 1ES encrptionBdecrption
J! +ardware random &it%generator
:! 3ull integrated @+3 R3 transceiver 4FGG =+' B JCJ =+' nominal6
_ 2rogramma&le output power 4%E? to 9? dBm6
_ .ow current consumption 499!: mA "or RK, 9;!? mA "or TK at ?dBm6
_ RSSI output that can &e sampled & the on%chip A1C

WISENET includes a socketed evaluation &oard 4CC9?9?EB6 and two
evaluation modules 4CC9?9?E=6! The evaluation &oard provided access to all o" the
analog and digital pins on the CC9?9?, as well as two serial ports, a parallel
programming port, R3 network analsis ports, and other peripherals! Each evaluation
module "eatured the CC9?9?, R3 network hardware, an antenna port, and an analog
temperature sensor! The modules connected to the evaluation &oard via two T3=%1
sockets! These sockets also allowed the possi&ilit o" designing a custom e)pansion
&oard!
WISENET is designed to measure light, temperature, and humidit!
There are man digital temperature sensors availa&le, &ut there is a much smaller
selection o" digital humidit and light sensors! A larger selection o" analog sensors are
availa&le/ however, analog sensors tended to re,uire more power and &e less precise than
their digital counterparts, in addition to re,uiring more comple) circuitr! 3or these
reasons, digital sensors are given higher priorit! Two new sensors provided the re,uired
"unctionalit! 3irst, Sensirion released the S+T99, a digital temperature and humidit
sensor with ultra low power consumption 4DD? =icroA while measuring, 9 =icroA when
in sleep mode6, a 9F &it analog to digital converter, and the desired accurac 4SDT
relative humidit, SGUC6! It also "eatured a simple serial inter"ace! The light sensor chosen
was the Te)as Advanced (ptoelectonic Solutions 4TA(S6 TS.EDD? am&ient light sensor
with S=Bus inter"ace! This sensor also "eatured ultra%low power 4C?? =icroA active, 9?
=icroA power down6, a 9E%&it analog to digital converter, and dual photo diodes! The
TS.EDD? uses &oth photo diodes to compensate "or in"rared light and to produce a
measurement that appro)imates the human ee response!
The "inal stage o" hardware design involved creating the Add%on
module! The WISENET Add%(n =odule has the two digital sensors descri&ed a&ove!
The Sensirion S+T%99 humidit and temperature sensor has a E%wire proprietar serial
inter"ace! The TA(S TS.EDD? digital light sensor uses an S=Bus serial inter"ace!
S=Bus is a standardi'ed E%wire serial inter"ace! The laout must &e care"ull designed
such that the light, temperature and humidit sensors does not underneath the evaluation
module when it is plugged into the &oard, which would make them useless!
So&tw%re 0e#in'#$el& product#:
The server using "or WISENET should have "our commercial o"" the shel"
applications installed on it that worked together to create the 1ata Analsis portion o" the
Server component! Apache, =S5., and 2+2 are open%source products "reel availa&le
on the Internet! In addition, Chart%1irector the trial version o" the commercial application
Chart%1irector was used!
Ap%c$e is a standard we&%server, which makes a we& document availa&le on
the Internet!
*+* is a we& programming language, which allows dnamic we&%pages! It
should also &e designed to use along with a data&ase and included man &uilt%in
"unctions "or inter"acing with =S5.!
M!S@L is a data&ase that can contain an tpe o" data and is accessed & a
TC2BI2 4Internet6 call!
C$%rt'0irector is a program that generates a graph "rom raw data! It is
availa&le in man languages such as 2+2, AS2, CQQ, and others!
So&tw%re Co)ponent# A Cu#to):
WISENET is also composed o" three custom so"tware components0
The Web program, WiseDB, and a port of TinyOS.
WISENET*s we& program was written in 2+2 and utili'ed the Chart%
1irector charting so"tware! The we& application ,ueried =S5. data&ase "or the data in
the re,uested date range, then we use a Chart%1irector to generate a graph o" that data!
Wise1B is the custom so"tware component that inter"aced with the
Sensor =ote Network via a serial link to the gatewa mote and with the =S5. data&ase
via a TC2BI2 link to the =S5. server application! Alread we know a&out how
Wise1B interacted with the rest o" the sstem! Wise1B was written in CQQ and utili'ed
two open%source A2I*s 4application programming inter"ace6!
The "inal custom so"tware component involves porting Tin(S to the
CC9?9?%&ased hardware plat"orm descri&ed in the +ardware 1esign section! As
previousl mentioned, Tin(S is a real%time operating sstem designed "or use in sensor
network applications where low%power, limited resources and hard real%time constraints
are critical parameters! A"ter implementing all the so"tware and em&edding in a single
sstem other important goal o" WISENET is to completel replace the lower%laer
"unctionalit to permit e)isting higher%level components and applications to &e
immediatel implemented on the new hardware plat"orm without modi"ication!
"uture Work:
There are a num&er o" "uture e)tensions "or this WISENET! A "ew are0
We can e)pand the sensor mote network & adding more motes! This
would allow the development and testing o" advanced network%laer "unctions, such as
multi%hop routing!
B creating a new 2CB design that integrates the CC9?9?E= design with
the sensors and power hardware on a single%&oard another interesting "eature can &e
developed or adopt a standard e)panda&le plug%in sensor inter"ace in &oth hardware and
so"tware
In researching alternative energ sources to e)tend mote &atter li"e!
2ossi&ilities include solar cells and rechargea&le &atteries!
Conclu#ion#:
Wireless sensor networks are getting smaller and "aster, increasing their
potential applications in commercial, industrial, and residential environments!
WISENET, as implemented, represents one commercial application! +owever, the limit
o" applications depends onl upon the sensors used and the interpretation o" the data
o&tained! As the technolog improves and new low%power digital sensors &ecome more
readil availa&le, motes will increase "unctionalit without increasing power
consumption and will e)pand the wireless sensing market!
Re&erence#:
7!Atkin#on, M!S@LBB: A CBB A*I &or M!S@L3 /er# 7,C,:,
<http://www.mysql.com/downloads/api-mysql++. htmlV!
1,4%! Le/i#, T$e nesC L%nu%e: A +oli#tic Appro%c$ to Network E)(edded
S!#te)s,
<http://today.cs.berkeley.edu/tos/papers/nesc.pdf>.
?,M%inw%rin3 *ol%#tre, et al! Wirele## Sen#or Network# &or +%(it%t Monitoring,
http://www.cs.berkeley.edu/~polastre/papers/wsna!.pdf
2!+ill3 S;ewc;!k, et al! S!#te) %rc$itecture direction# &or network #en#or#3
http://today.cs.berkeley.edu/tos/papers/tos.pdf
9!Tor/)%rk3 Applic%tion Note AN67C: Low *ower S!#te)# U#in t$e CC767?,
http://www.chipcon.com/files/"#_$%_&ow_'ower_(ystems_)sin*_+he_
,,$$_$_$.pdf
D!=e3 +eide)%nn3 et al! An Ener!'E&&icient MAC *rotocol &or Wirele## Sen#or
Networks,
http://www.isi.edu/-%.weiye/pub/smac_infocom.pdf
/or software shelf products downloads0 websites are:
www.apache.or*
www.php.net
www.mysql.com
www!advso"teng!comBinde)!html
http0BBInternetmaster!comBinstalltutorialBinde)!html