Development of a Low-Power Smart Water Meter for Discharges in Indus Basin Irrigation Networks1 ZahoorAhmad , EhsanU.Asad , AbubakrMuhammad , WaqasAhmad , ArifAnwar

1Department of Electrical Engineering, SBA School of Science & Engineering, LUMS, Lahore, Pakistan. {zahoor.ahmad,ehsan.haq,abubakr}@lums.edu.pk

2International Water Management Institute (IWMI) Pakistan Office, 12km Multan

Road, Lahore, Pakistan. {a.anwar,w.ahmad}@cgiar.org Abstract. To improve the sampling frequency of water diversion to distributary canals and to improve equity of distribution and data handling we have developed a smart electronic water meter based on ultrasonic sensors and GPRS modem to frequently record and transmit the water diversion data to a centralized server. The server processes the data to extract useful information for example seasonal cumulative water deliveries and discharge time series. The Wireless Sensor Node (WSN) inspired design is extremely low-power, field deployable and scalable with respect to cost and numbers. This paper, reports the first steps towards practical realization of a smart water grid in the Indus river basin, conceptualized by the authors in pre-vious theoretical studies.

1 Introduction

The Indus River Basin in Pakistan has the world’s largest contiguous irrigation net-work, running over 90,000 Km of watercourses and around 25 million acre irrigated area, the overall irrigation infrastructure accounts for approximately US$ 300 billion. Such a large system cannot be managed with high efficiency without using unprece-dented levels of automation and usage of decision support systems [1],[3],[13]. Quite contrary, most of the operation is manual which combined with poor management practices have contributed towards acute problems of water scarcity and distribution inequity. See [2] and [6] for an overview of such problems. Therefore, there is a great need to contribute towards agricultural development in Pakistan through the efficient management of surface water in canal networks to enhance food security, reduce pov-erty and adapt to uncertainties brought about by climate change.

Future irrigation networks as described by [4] and other leading researchers, represent a prime example of cyber physical systems (CPS), i.e. physical infrastruc-tures coupled tightly with distributed networks of sensing, computing and control structures. See [7],[12] and [15] for a discussion on CPS. In the context of Indus river basin, the LUMS affiliated authors have conducted a series of theoretical studies to determine the feasibility of a fully automated CPS, envisioned as a smart water grid for Pakistan. See [9] and [11] as examples of such case studies. Encouraged by these studies, the group has come into several partnerships in recent years with government

1 The authors would like to acknowledge financial support provided by the Embassy of the

Kingdom of Netherlands, Islamabad, Pakistan through Grant #22294 used in part for the partnership between IWMI and LUMS to undertake the research described in this work.

and non-gIWMI aftions in thagement,

Figure 1Irrigation

Twater dismost critisons. Firslarge. Thurementsterpretatioriorating give neararea watening, resomanual oother opedemand bhealth momeasuremtion such

Kfield depldesign is for instalIndus riveBahawalnanalysis oreported per mostldescriptio Astilling wwell is inter using is time-st

government offiliated authohe context of enforcement

(a): Examplesn Department

ed to flow u

This paper foscharges (pinsical requiremest, the scale oe scale enorm poses obviouon. Second, thinfrastructure

r real-time picer boards, farolve conflicts

operation allowerations and sbased deliveronitoring, re-pments will be

as controlled Keeping in mloyable, standinspired fromlation on braner basin. At thnagar, Punjabof the networkafter testing tly reports the ons of the softA high level d

well is construnstalled a batte

the principle tamped and t

organizations rs of this studwider governof water right

s of manual gat in Hakra Brausing a rating

cuses on autosal nawisi) in ents of automof such measu

mity when comus logistical phe fidelity of me and human fcture to basinrmer organizas, ensure transws no quicker services that ary, detection planning undethe first step gates and oth

mind these mod alone and wem wireless sennch canals andhe time of writb in the Hakraked sensing sythe system for

developmenttware backbondiagram of theucted in close ery powered wof ultrasonic

then transmitt

to translate thdy have similance issues rets and account

auge readout s

anch, Bahawalg curve derived

omating the hthe network

mation in the Iurements, eve

mbined with thproblems in dmanual measufactors in gau

n managers (thations) to perfsparency and than daily up

are not yet feof leakages

er changing sctowards insta

her active strucotivations, weeather proof c

nsor network (d distributarieting, 24 nodes

a canal commystem with dar the upcomint of electronicne and suppore proposed auproximity to

wireless sensodistance mea

ted to a cent

he theory intolarly been argelated to partictability.

settings for palnagar. The wd from channe

ourly-to-daily(See Figure 1

Indus basin duen at a local che manual natata collectionurements is quuge reading. The provincial form situationmaintain equ

pdates. Fourth,easible such aand non-techcenarios etc. alling even hictures. e have develocanal flow me(WSN) technoes at irrigations are being ins

mand area off tata analysis anng Rabi and Kcs for the wirerting civil infrautomation is s the canal. A

or node that saasurement. Thtral office at

o practical sysguing for suchcipatory irriga

ansal nawisi bater levels areel geometry.

y measuremen1(a)). This is oue to the follocanal commanure of daily fl

n, disseminatiouestionable du

Third, there is irrigation dep

n assessment uity amongst u, automation cas volumetric hnical losses, Lastly, automigher levels o

oped a fully feasurement syology and mon canal netwostalled by the the River Sutnd interpretatiKharif seasonseless sensor nastructure. hown in Figut the top of thamples the lev

he measured wregular interv

stems. The h automa-ation man-

by Punjab e convert-

nt of canal one of the owing rea-nd is very low meas-on and in-ue to dete-a need to

partments, and plan-

users. The can enable metering, structural

mated flow of automa-

functional, ystem. The st suitable

orks of the authors in

tlej. A full on will be

s. This pa-node, brief

ure 1(b). A he stilling vel of wa-

water level vals using

GPRS/GSdata, caliba predefinmade avavarious st

Tlow-poweextremesand a solcantly to tem onlydemonstrtional merate, weawithin a sound in have pacbriefly tocally tacka laborato

2

In this2(b)). It clevel andperature sare plannsystem isalso has aof the wa

SM based serbrate them, coned rating curailable for distandard web a

The contributier battery pow: A direct AClar-powered sthe cost. Our

y requiring lowrated usage ofechanical floaather proof anstilling well. weather expokaged, tested

ouched in this kled in softwaory prototype.

Wireless S

s section, we wconsists of an d flow. It also sensor immer

ning to add ots operated wita temperature

ater. The WSN

rvices. Here,onvert them inrve and then ssemination inand mobile ba

Figure 1(b): ions of this arwered design

C mains powerystem which

r design guaraw-cost annualf maintenanceats or pressurend narrow be

We have takosed condition

and calibratearticle, real-w

are and civil in.

Sensor Nod

will mainly foultrasonic ranhas capabilitised into the wther sensors toth 3.6V, 1400 sensor imme

N calculates th

a computer nto discharge file them in an the form o

ased services.

System data frticle are as fo

n providing a red system whis difficult to

antees a long-ll battery repl

e-free ultrasoue transducerseam sensors tken care to cns. Thirdly, thed the sensor world deploymnfrastructure,

de Design

ocus on Wirelnge finder as ies to interfac

water to calculo measure the0mAH batteri

ersed into the he water level

server receive(in cubic fee

a database. Suf graphs, tim

flow diagram.ollows. Firstly

practical comhich is infeasio secure physlife completelacement. Sec

und based sen. We have seto overcome compensate fohe unit is field

for all extremment challenge

making our s

less Sensor Nthe main sense to auxiliarylate its tempere turbidity andies with surgewater and calin the canal a

es the raw wt per second, ubsequently, te series and t

y, we achieve mpromise betible at remoteically and addly battery pow

condly, we susing instead o

elected extreminstallation d

or changes ind deployable ime conditionses have been ssystem much m

Node (WSN) (Ssor for measur

y sensors suchrature. In the d salinity of we current capalculates the temand sends the d

water level cfs) using

the data is text using

extremely tween two e locations ds signifi-wered sys-uccessfully of conven-mely accu-difficulties n speed of in that we s. Though systemati-more than

See figure ring water

h as a tem-future, we

water. The abilities. It mperature data to the

server thrthe blockgiven bel

2.1 Ul

The Mplicationscost and reading, iline featuimprovedpower reland GNDdegree cesensor options. Altused the oly and co

MB73of the decone. Beabeam angneed narrthe cone d

rough GPRS/k diagram showlow.

Figure

Figure 2

ltra Sonic Sen

Maxbotix MB7s where preciIP67 weather it calibrates itsures 1-mm red accuracy andlated noise iss

D pins. The spentigrade. If tperation, the sthough the MBoptional extermpensations a80 sensor has

etection area oam pattern is gle for that tarrower beam adoes not inter

GSM at a prewn in Figure

e 2(a): Block d

2(b): Photogra

nsor

7380 Ultra Soision range-finresistance rat

self. The sensesolution, targd internal speesues we addedpeed of soundthe temperatu

sensor will appB7380 has anrnal temperatuare applied fos a calibrated bof the sensorused for a sp

rget at the speangle. Our exprfere with the w

edefined but c2(a). The deta

diagram of W

aphs of the de

onic Sensor [8nding, low voting is neededsor then uses tget-size and ed-of-sound ted a 100uF cap

d in air increasure, humidity,ply necessaryinternal temp

ure sensor MBr temperature beam pattern.

r. The beam pecific target a

ecific distancepected distancwalls of the st

configurable iails of each im

ireless Sensor

eveloped Smar

8] is a cost-efoltage operatid. Each time tthis data to ranoperating-volemperature copacitor at the ses about 0.6 , or applied v

y temperature perature sensoB7955 which i

variations in Beam patternpattern is actuat any given de. Generally, fce to be meastilling well.

interval. Pleasmportant comp

r Node.

rt Water Mete

ffective solutioion, space savhe sensor taknge objects. Tltage compenompensation. Tsensor betwemeters per se

voltage changand voltage cr; for best accis detected auacoustic rangn is a 2D reprually shaped distance to calfor shorter dissured is 1-2 m

se refer to ponent are

er.

on for ap-ving, low-es a range

This sensor sation for To correct en the V+ econd, per ges during compensa-curacy, we tomatical-ing path.

resentation like a 3D lculate the stances we meters. So

2.2 M

Microcused in omemory memory We save records inhave a todata in ca

2.3 GP

We uSIM900DThe moduare develtransfer oto 4.8V, bthe netwo

2.4 Ex

As ousystem toapplicationology, otem such The desigcircuitry shows typtantalum Amperes 300mA (cause the

Microcontrolle

chip’s PIC18Four embeddedaddressing, 1addressing. Ethe most recen the volatile

otal of 48 recoase of GPRS/G

PRS/GSM M

used SIMCOD [14] is a quule is integratloped for reliaof 42.8 kbps. but practicallyork below 3.4V

xtreme Low P

ur application o last for up toons like theseoffering the lo

that it does ngned hardwarwith 3.6V, 1pical dischargcapacitor is p(few millisec

2 to 3 secondey have lower

Figur

er

F26K20 flash d design. It is 024 bytes of

Each data sament 110 record

SRAM. We ords daily. ThGSM transmis

Module

OM SIM900Duad-band GPRted with the Table data transThe manufacty we found thV.

Power Design

require proloo 5 to 10 year

e, we selected owest currentsnot require anyre is spending4,000mAH Lge characterisplaced across conds wide) dds wide). Wesurge currents

re 2.4(a): Batt

microcontrollfeatured withEEPROM da

mple is time stds in non-volattake the samp

hus, with this ssion problem

D Version 1RS/GSM engiTCP/IP protocsfer applicatioturer claims thhat the modul

n

onged standalrs, while runnproducts with

s for run and y maintenanceg 98% of itsLithium Thionstics at variouthe battery, w

during transmi cannot run its.

tery Discharge

ler with extremh up to 64 kbata and up to amped whichtile EEPROMple after 30 mmuch memor

ms.

1.03 for GPRine that workscol; extended ons. The modhe module to e is not gettin

one operationning from a sih Microchip’ssleep. We woe over the lifetime in sleep

nyl Chloride Bus current levewhich causes tission burst tot with normal

e Characterist

me low powerbytes of linear

3936 bytes lh makes it 8 b

M and the remaminutes intervary we can sav

RS/GSM tras on GSM freTCP/IP AT c

dule has GPRSbe working f

ng itself regist

n. We are plaingle battery. s nanoWatt X

orked to desigspan of the ap. We are powBatteries. Figels. A 100µF the current spo smooth dow Lithium Ion

tics[5]

r (XLP) is r program inear data

bytes long. aining 500 als, so we

ve 12 days

ansactions. equencies. commands S data and from 3.1V tered with

anning our To enable

XLP Tech-gn the sys-pplication. wering the ure 2.4(a) low ESR

pike of 1-2 wn to 200-

Cells, be-

GSM/Gmodule osensors aious extrecontrollerof sampleple. We ebest optiorunning mneeds som

Fwould bevaries grecurrent dshut dowseconds aonly consGSM moseconds. seconds iwireless cIt makes waits for rents conboast sleedown to main focu

Figu At sleeshown inwe have a

Thus expthe belowstrength wresults th

25710121517

mA

GPRS transacoff after everyre turned on oeme low-powr might have nes. Therefore,evaluated the on is to operatmode during me delays to gFigure 2.4(c) e worth mentieatly with RF

during the tranwn for about o

and take not msider currents

odule turns on At 16th seconit waits for anconnection witwo tries to gan incoming

nsumed by thep currents be100 μA/MHzus of our desig

ure 2.4(b): Cuep mode, WS

n figure 2.4 (b)a total of 28A=

pected battery w results for was varied by

he current sam

025507500255075

0 400

ctions consumy transaction only at samplier control featnothing else t, the microcontime×current

te the MCU fotransaction fo

get registered wshows curre

ioning here thF signal strengnsmission burne hour it takmore than 2mconsumed byat 7th second

nd WSN regisn incoming teith GPRS. Afget connecteddata from the

he microcontrelow 100 nA, z. This essentgn from the po

urrent vs. timeN consumes ) for an hour

AH charge. . ∑life = (28*60burst transmi placing cond

mples are taken

800 12

me most of thwhich takes

ing time. In thtures of an emto do until thentroller enterst elements of or longer at lofor most of thwith the netw

ent consumedhat currents cogth. At low sist (See Figure

kes 1-2 mA cumA current, soy GSM part od and after doisters itself witext message. After 74 secondd. At 85th secoe server and troller PIC18FWatch-dog T

tially means toint of view o

e plot for one t0.5-0.6 mA ointerval. We a.t = 60*60)/6.077=ission during

ductor materialn at 0.5 secon

00 1600 2Time(Sec

he battery pow30uA at shu

his section, wembedded micre peripheral cos "sleep" in bthe energy eq

ow frequency, he time. Duri

work. d for a singleonsumed duriignal level We 2.4(d)). Whurrent. The seo we are ignoof WSN. The ing the transacth the GSM neAt 65th seconds, WSN startsond it sends tthen shuts dow

F26K20 (at 2.Timer down tothat the GSMof power consu

transmission cof current. Letare using two 6077 = 616,587Hours=various sign

ls around WSnds interval. T

2000 2400conds)

wer. We turn utdown mode.e will highlighrocontroller. Tollects a certabetween each quation and dbecause the M

ing that time

e GPRS transing GPRS traSN will consu

hile the GSM ensors are on ring those curgraph specifiection turns offetwork. For th

nd the system s up a TCP cothe data succewn the GSM .9V) in vario

o 1μA and RuM module shou

umption.

cycle (~one hot us integrate batteries in p.077 b

= 691 days. Pnal strength. T

N. In each of The below res

2800 3200

the GSM . Also the ht the var-

The micro-in amount data sam-

determined MCU is in the GSM

saction. It ansactions, ume more module is for only 2 rrents and es that the f after 100 he next 40 brings up

onnection. essfully. It part. Cur-

ous modes un-currents uld be the

our). the curve

parallel, so

b. Please find The signal the below

sults show

0 3600

that at stburst.

Figure

2.5 H

The(dust tighAluminumis not coenvironmcircuitry. will be ping. Foursensors (emade of A

RoHS trical andMercury,diphenylerequireme

2.6 Re

Implemto put a ti

025071001215017

mA

112223

trong signal s

Figure 2.4(

e 2.4(d): Curre

ardware Enc

e electronic Ciht and protecm Box of 180rroded even i

mental conditioIf the device

rotected. As Gr IP67 standardenvironment aABS plastic an(Restriction o

d electronic p Hexavalent ethers. All coent of RoHS.

eal Time Clo

mentation of Rime stamp wh

05050505

0 10

04080120160200240280320

0 20

strength, WSN

(c): Current v

ents consumed

closure

ircuitry is procted against i0mmx80mmxif immersed ions. The moi

e is immersed GSM/GPRS sd connectors aand water) andnd are corrosiof Hazardous

products put oChromium, P

omponents and

ck

Real Time Clhile take the re

20 30 4

40 60

N consumes l

s. time plot fo

d at -74dBm a

tected againstmmersion) In

x60mm. Die cin water. It histure and dusinto the wate

signals do notare provided fd turbidity senion free with aSubstances) D

on the marketPoly brominad packing ma

ock (RTC) weading. Dallas

40 50 60Seconds

80 100

low current d

or one transmi

and -56dBm si

t the environmngress Protectcast Aluminumhas the capabist cannot ente

er in case of flt work properfor antenna, twnsor in the futa special seal fDirective esset shall not co

ated biphenylsaterials used i

was necessary s DS1302 was

0 70 80

120 140

during the tra

ission burst.

ignal levels

mental factors tion standard

m is selected bility to withster inside the

flooding, the erly inside a mwo external temture. The connfilled inside.entially states ontain Lead, Cs and Poly brin the design

for Smart Was selected for

90 100

160

(‐74 (‐56d

ansmission

with IP67 s die cast because it tand harsh electronic

electronics metal hous-

mperature nectors are

that elec-Cadmium, rominated fulfill the

ater Meter its ease of

110

dBm)dBm)

availability and reliability. It consumes less than 500nA in battery backup and is op-erated with separate 3.3V 40mAH cell. Thus, RTC cell backup time= (40mAH/500nA=80,000 hours=9 years). The RTC time can be checked and reset remotely from server side. RTC counts seconds, minutes, hours, date of the month, month, day of the week, and year with leap-year compensation valid up to 2100. It has three wire SPI interface and 8-pin SOIC pack-age is used in our embedded design.

2.7 Design Problems and their Solutions

Since Wireless Sensor Node is likely to be installed at locations with limited GSM/GPRS coverage. It is provided with an external antenna mounted at highest point inside the Stilling well. The device sends data on GPRS and in case of low RF signals it will initiate and SMS to the server, as SMS requires less signal strength. Being operated from 3.6V battery, we are quite close to the minimum voltage rat-ings of SIM900D i.e., 3.4V.The copper traces from battery to the GSM module are 3.2 mm wide in the PCB design to reduce the voltage drop for situations, where the circuit consumes high currents like 1-2A during transmission burst. Note that = ( ∗ ℎ )/ . Also, wide tracks sink heat to the environment instantly. Narrow tracks result in high energy dissipation, which further increases the resistance in the absence of heat sink. Thus for best performance during transmission bursts we need to reduce the length of wire as well as widen cop-per traces on PCB from battery terminal to the GSM module. To appreciate the im-portance of this, note that SIM900D does not work below 3.4V. At high currents the voltage across the battery terminal decreases due to the internal resistance of the bat-tery as shown in the Figure 2.4(a) above. If the resistance from battery terminal to the GSM module is 0.3 Ohm and we have a transmission burst of say 1.0A, then voltage drop across the wire, = = 1 ∗ .3 = 0.3 .So voltage at SIM900D will be 3.6V-0.3V=3.3V and will be powered down due to these drops.

The GSM part of WSN is protected against electro static discharge (ESD) with SMF05C, a 5 line transient voltage suppressor array, having a peak power dissipation of 100W (8 x 20 µS waveform) [10]. It has ESD rating of class 3B (exceeding 8 kV) per human body model and Class C (exceeding 400 V) per machine Model. Human body interaction with the circuitry is possible during SIM replacement. During as-sembling, we have ensured proper handling against electrostatic discharge. Anti-static Wrist Straps, are used to prevent ESD by safely grounding the technician working with electronic equipment. It consists of a band of fabric with fine conductive fibers woven into it.

2.8 Wireless Sensor Node Budget

The cost breakdown of the wireless sensor node has been given below from current list prices. Note, mainly that the absence of the solar panels have significantly reduced the unit price. Moreover, the maintenance is only annual replacement of batteries. We

believe that this (low) cost has made the deployment of these units feasible at practi-cal scales of deployment. See table 2.8 below for budgeting details. S.No. ITEM PRICE (USD) 1 MB7380 with temperature Sensor 114 2 SIM900D 20 3 Batteries 16 4 Die cast Aluminum Enclosure 26 5 Miscellaneous parts 14 Total $190 / PKR. 19,000

Table 2.8: WSN Budget

3 Software Design

3.1 Working

The WSNs installed at different canals send Water level readings on server. After the server receives the strings of data, it processes the data string and extracts the required parameters i.e., date, time, temperature, and level readings. These readings are then stored in the database with their respective date and time, which are then used to compute the seasonal cumulative water deliveries and discharge in average cubic feet per second (cfs). The data is displayed as CSV files with graphs depicting seasonal and yearly variations on a website. The whole software design consists of Server and Client side designs.

3.2 Server-Side Software design

The Server side design comprises of Web Server, Server-side script, Operating System and a Database for storing the required data. Our choice for Server is the Mi-crosoft Internet Information Services (IIS). We are using Microsoft server-side Web application framework (ASP.NET). For Relational Database Management System (RDBMS), we are using Microsoft SQL in web applications. The server is running on the Microsoft Window Server 2008 R2 Operating System.

The important aspects of the software design include: 1. Security. Common types of software flaws that lead to vulnerabilities in-clude: SQL injection, Cross-site scripting and some others. MSSQL and Asp.net provides sufficient extensions and options to avoid such type of vulnerabilities. 2. Replication and Back Up. Replication of MSSQL database can be a solu-tion to various problems like Scale-out problems, Data Security, Analytics, Long-distance data distribution and the Backup taken from slaves rather than from mas-ter, which result in no load on the production Master machine for this task. If the Slave node is down, this is not a problem since replication is performed asyn-chronously and when the Slave Node is up and Live after a downtime, it contin-ues replication from the point it has been paused.

3.3 Cl

The deto check different 3.3(a)). Tcaded styimplemenand additthe canalsdisplay thspreadsheings frommitted fro

Fig

Figure 3.3controlled

5:00 PM

lient-Side Sof

esign of the wout the canalcanals with

The website isyle sheet (CSSnted as a parttion of some os, drop down he level readineet programs l

m a test site (Som WSN at an

gure 3.3(a): W

3(b): Water led leakage for

9:40 PM

8:30 PM

7:20 PM

6:10 PM

ftware Design

website is quitl readings wiother parame

s designed usinS) for displayint of a web brother featuresmenu lists etcngs. CSV filelike MS Exce

See Figure 4.1n interval of 1

Website User In

evel measurem19 hours

2:20AM

1:10 AM

12:00 AM

10:50 PM

Tim

n

te user-friendlth the seasoneters like turng the Hypertng and stylingowser, is uses like Google c. Comma Sepes are practicael. Figure 3.3(b) over 19 hou hour.

nterface for ac

ments from sti

700AM

5:50 AM

4:40 AM

3:30 AM

2:20 AM

me

ly and it provnal water discrbidity and tetext markup lag web pages. Jd to create enmap for show

parated Valueal for importinb) shows a pl

urs, sampled a

ccessing the c

lling well at a

10:30 AM

9:20 AM

8:10 AM

7:00 AM

vides users wiharge per uni

emperature (Sanguage (HTMJavaScript, a cnhanced user wing the locates (CSV) files ng level readilot of the receat 10 minutes

canal Informat

a test site with

450480510540570600630

11:40 AM

th an ease it area for See figure ML), Cas-commonly interfaces tion of all is used to ngs into a ived read-and trans-

tion

h a small

Readingsin mms

4

To protecwater surto achievstilling won either conduit. Adielectricreaches threflector Thus, stee

Figure 4.

Figure 4. While segmentsstilling wnal and tobeen madment prostrong enhigh strencilitate hstrength 2icant stren

5

A low-poaffordabl

Stilling W

ct WSN againrface to be reave these objectwell (See figur

side of a chaA reinforced c

c slab (concrehe wall and, dappears. FSSel reinforceme

1(a): Cross se

1(b): Instrumedesigning the

s were used inwell is wide eno enable persode for levelingoperly. WSN nough to bear ngth PCC con

handling and 2-part thixotrongth while in

Conclusio

ower water de and ready f

Well and Civ

nst harsh outdoad by the ultratives. In hydrre 4.1) which annel and is hconcrete wall

ete). This strudue to this phS reflector maent is avoided

ection of stillin

ent installatione stilling well, n constructionnough that it conal access tog the WSN; wis installed othe harsh ou

ncrete. The strtransportation

opic epoxy adtension from

ons

discharge meafor wide-scale

vil Infrastr

oor environmasonic sensor

raulic engineenecessarily cohydraulically is composed

ucture offers ahenomenon, anay act as ban

d in the civil st

ng well across

ns, alignment we considereto achieve un

could not intero its base for awith nuts on thon a metal fraut door enviroructure is conn. The segm

dhesive to elimlateral force.

asurement hase deployments

ucture

ment and to pror we have desiring this typeonsists a concconnected toof a square s

a resonant behn FSS (Frequend stop filter tructure.

s canal.

and leveling. ed several impniformity and rfere with theannual cleaninhe side wall wame. The stilnment; theref

nstructed in seents are join

minate seepag

s been develos in irrigation

ovide a relativigned the infr

e of structure icrete lined boxo the channel teel grid immhavior when ency Selectivefor GSM fre

(c): Installedportant matters

cost effectivee cone of ultrang. An arrangwe can align tlling well is afore it is consegments of 1 fned together we and to provi

oped which isn canal netwo

vely stable rastructure is called a x installed through a

mersed in a the signal e Surface) equencies.

d Unit s. Pre-cast eness. The asonic sig-ement has the instru-assembled structed of foot to fa-with high ide signif-

s suitable, orks of the

Indus river basin. GSM/GPRS based transmission is chosen as a suitable technology for communication whereas ultrasound based ranging is found suitable for sensing water levels. Power requirements, packaging, installation and system integration is-sues have been addressed and resolved. References [1] Afzal, M. (1996). Managing water resources for environmentally sustainable irrigated agri-culture in Pakistan. Pakistan Development Review, 35, 977–988. [2] Bandaragoda, D. and ur Rehman, S. (1995). Warabandi in Pakistan’s canal irrigation sys-tems: Widening g2ap between theory and practice. IWMI. [3] Briscoe, J. and Qamar, U. (2009). Pakistan’s Water Economy Running Dry. Oxford Univ Pr. [4] Cantoni, M., Weyer, E., Li, Y., Ooi, S., Mareels, I.,and Ryan, M. (2007). Control of large-scale irrigation networks. Proceedings of the IEEE, 95(1), 75–91. [5] EEMB; (2012), ER34615M. Lithium Thionyl Chloride Battery. http://eemb.com. [6] Latif, M. and Saleem Pomee, M. (2003). Impacts of institutional reforms on irrigated agri-culture in Pakistan. Irrigation and Drainage Systems, 17(3), 195–212. [7] Lee, E. (2008). Cyber physical systems: Design challenges. In 2008 11th IEEE International Symposium on Object Oriented Real-Time Distributed Computing (ISORC), 363–369. [8] Maxbotix; (2012), MB7380 Outdoor Ultrasonic Range Finders. http://www.maxbotix.com. [9] Nasir, H. and Muhammad A. (2011). Feedback Control of Very-Large Scale Irrigation Networks: A CPS Approach in a Developing-World Setting. 18th World Congress of Interna-tional Federation of Automatic Control (IFAC), Milano, Italy. [10] ON Semiconductor; (2005), SMF05C Transient Voltage Suppressors. http://onsemi.com. [11] Tariq M. U., Nasir H., Muhammad A. and Wolf M. (2012) Model-Driven Performance Analysis of Large Scale Irrigation Networks. IEEE/ACM International Conference on Cyber-Physical Systems (ICCPS), Beijing, China. [12] Rajkumar, R., Lee, I., Sha, L., and Stankovic, J. (2010). Cyber-physical systems: the next computing revolution. In Proceedings of the 47th Design Automation Conference, 731–736. ACM. [13] Seckler, D., Barker, R., and Amarasinghe, U. (1999). Water scarcity in the twenty-first century. International Journal of Water Resources Development, 15(1), 29–42. [14] SIMCOM; (2010), SIM900D GSM/GPRS Module Hardware Design Manual V1.03. www.sim.com. [15] Wolf, W. (2009). Cyber-physical systems. Computer, 88–89.