wireless sensor network for animal monitoring using both antenna and base-station diversity

Wireless Sensor Network for Animal Monitoringusing both Antenna and Base-station Diversity

Konstantinos Sasloglou, Ian A Glover, Kae-Hsiang Kwong and Ivan AndonovicDepartment of Electronic and Electrical Engineering

University of StrathclydeGlasgow, G1 1XW, United Kingdom

Tel:+44 (0) 141 548 4016 Fax:+44 (0) 141 552 4968Email: [email protected]

Abstract—Wireless sensor networks are widely used for con-dition monitoring applications. Much effort has been invested inimproving the performance of such networks. Diversity is a well-proven technique in this context. Here, we present the practicalapplication of base station and antenna diversity. Empiricalmeasurements of performance in a realistic environment arereported and a statistical analysis of the resulting data ispresented.

Index Terms—antenna diversity; base station diversity; ani-mal monitoring; wireless sensor networks; ricean distribution;rayleigh distribution; fading

I. INTRODUCTION

The term Wireless Sensor Network (WSN) refers to a groupof spatially dispersed sensors that can communicate with eachother via the wireless medium [1].

The associated set of wireless transceivers form a network ofcommunicating nodes. Sensor data is typically communicatedto a central processor or base-station by, if necessary, relayingit via intermediary nodes. The network of nodes may beself-organizing. Recent advances in micro-electromechanicalsystems (MEMS), wireless communications and transducertechnology have made large networks of this type commer-cially viable. WSNs now represent a flexible and scalable tech-nology that can be deployed in many applications includingmilitary [2], environmental monitoring [3], habitat monitoring[4], agriculture [5] [6], animal husbandry [7] [8], home [9],office [10], and others [11].

This paper reports some preliminary work on the applicationof antenna and base-station diversity in WSNs for animalhusbandry in the dairy and beef industries.

In mammal monitoring and tracking applications a singleantenna is traditionally attached to a collar worn aroundthe animal’s neck. As the animal moves the line-of-sight(LOS) path between collar-antenna and base-station (or collar-antenna and another node in the WSN) might be obscured,either by fixed obstacles or by other animals. The latter is themore likely in the context of animals with a herding instinct onopen grassland areas such as in the dairy and beef industries.

The incorporation of two antennas at the sensor node(spatial diversity) combined with two widely separated base-stations (base-station diversity) must clearly increase the prob-ability of LOS conditions. It is the degree to which signalfading statistics are improved that is the focus of this work.

II. RELATED WORK

One of the early projects related to habitat monitoring [4]was conducted at Great Duck Island; a 237-acre island located15 km south of Mount Deset Island, Maine. 43 sensor nodeswere deployed to monitor the microclimate in and aroundnesting burrows used by the Leachs Storm Petrel. The systemwas operated for something over four months and more than1.1 million readings were collected. Each node incorporated amicro-controller, a low-power radio transmitter, memory and abattery. The node sensors measured light intensity, temperatureand humidity. The sensor outputs were periodically sampledand the samples relayed to a base-station computer whichacted as an Internet gateway, thus allowing real-time dataaccess.

ZebraNet [12] used a WSN for tracking and monitoringzebras in Kenya. GPS-enabled sensor nodes were incorporatedin collars and geographical location data was passed to abase-station using, where necessary, other network nodes asrepeaters. The system hardware comprised a 16-bit TexasInstruments micro-controller, 4 Mbit of off-chip flash memory,a 900 MHz radio transceiver, and a low-power GPS chip.

’Wired Pigs’ [13] used a wireless sensor network to monitorthe movement of pigs.

Several trials using WSNs for cattle monitoring have beenreported [7] [8]. One trial [14] [15] equipped 45 animalswith collars incorporating two batteries, a GPS antenna, atransceiver and a radio antenna. The antennas were subject todamage, however, due to animals rubbing them on scratchingposts and biting them [15].

None of the applications described above incorporated anyform of diversity.

III. SYSTEM ARCHITECTURE AND METHODOLOGY

The transceiver used for this experiment was the MICAz[16] shown in Figure 1. It operates in the ISM band between2.4 GHz and 2.48 GHz. The transmitted power was set to-10dBm.

The transceiver is mounted on a PCB, Figure 2, whichalso includes an RF switch that connects the MICAz RFinput/output to two microstrip antennas.

Each antenna is an inset-fed microstrip patch with a ceramicelement attached to the top of the radiating surface. The

1-4244-2424-5/08/$20.00 ©2008 IEEE ICCS 2008 27

Fig. 1. The MICAz transceiver.

Fig. 2. PCB carrying the MICAz module.

radiation pattern of the antenna is shown in Figure 3 at threefrequencies, for two orthogonal linear polarisations in threeorthogonal planes.

Directivity angleVertical polarization

Horizontal polarization

0°

270°

90°

180°

2.4GHz2.442GHz2.485GHz

180°Vertical polarization

270° 90°

–5

–15

5(dBi)

–25

0°


180°Horizontal polarization

270° 90°

–5

–15

5(dBi)

–25

0°

0°

270°

90°

180°



270° 90°

–5

–15

5(dBi)

–25

0°



270° 90°

–5

–15

5(dBi)

–25

0°

0°

270°

90°

180°



270° 90°

–5

–15

5(dBi)

–25

0°



270° 90°

–5

–15

5(dBi)

–25

0°

Fig. 3. Radiation pattern of the patch antenna.

The entire configuration was attached to the collar, whichwas oriented on the animal such that one antenna was locatedon the left side of the neck and one on the right side as shownin Figure 4.

20 m

12 m

A2A1

4 mCollar 1

Collar 2

Fig. 4. Schematic illustration of antenna arrangement on collar and trialarea.

The area where the system was set up and the measurementswere undertaken was an approximately 20m x 12m and wasenclosed by thick brick walls and a pitched metallic roof,Figure 5.

Fig. 5. Trial area.

The base-station comprises a transceiver (identical to thoseattached to the animal collars) interfaced to an MIB600 pro-gramming board. A block diagram is shown in Figure 6. Theantenna used at the base-station was close to omnidirectionalwith a gain of 6 dBi [17].

Fig. 6. Base-station configuration.

Two test animals were released into the test area along withseven others and the received power was recorded for 75 min-utes at both base-stations. The two antennas at each animal-mounted node were alternately connected to the transceiverfor 1 s using the RF switch. The complete switching cycletherefore has a period of 2 s. The sampling frequency for aparticular antenna was therefore 0.5 Hz.

The movement of animals was sufficiently slow such thateach 1 s block of contiguous data received from a givenantenna can be assumed to originate from a single location.

The resulting data was smoothed by calculating the movingaverage of 15 samples representing a time window of 30 s.

IV. SIGNAL STATISTICS

The raw data (0.5 Hz samples) received from base-station 1(BS1) and base-station 2 (BS2) are shown in the upper plotsof Figures 7(a), (b), (c) and (d) and Figures 8(a), (b), (c) and(d), respectively. Figures (a) and (b) correspond to the datatransmitted from antenna 1 (A1) and antenna 2 (A2) mountedon collar 1 (C1). Figures (c) and (d) correspond to the datatransmitted from A1 and A2 mounted on C2.

28

0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60

Number of Packets Received

RSS

I(dB

m)

Raw Data for A1 of C1 using BS1

0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60

Moving Average of Raw Data


RSS

I(dB

m)

(a)

0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60


RSS

I(dB

m)


0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60



RSS

I(dB

m)

(b)

0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60


RSS

I(dB

m)


0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60



RSS

I(dB

m)

(c)

0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60


RSS

I(dB

m)


0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60



RSS

I(dB

m)

(d)

Fig. 7. Data received by base-station 1.

0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60


RSS

I(dB

m)


0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60



RSS

I(dB

m)

(a)

0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60


RSS

I(dB

m)


0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60


Number of Packets ReceivedR

SSI(

dBm

)

(b)

0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60


RSS

I(dB

m)


0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60



RSS

I(dB

m)

(c)

0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60


RSS

I(dB

m)


0 500 1000 1500 2000 2500 3000 3500 4000 4500−100

−90

−80

−70

−60



RSS

I(dB

m)

(d)

Fig. 8. Data received by base-station 2.

29

The fluctuation of power is large (up to 20 dB). The cor-responding lower plots show the 15-sample moving average.The horizontal line in the figures represents the mean powerfor each measurement set.

The peak-to-peak variation of received power recorded foreach antenna on each collar to each base-station over the totalobservation time is in Figure 9.

BS0C1A1 BS0C1A2 BS0C2A1 BS0C2A2 BS1C1A1 BS1C1A2 BS1C2A1 BS1C2A20

5

10

15

20

25

30

35

Pow

er D

iffe

renc

e (d

B)

Fluctuation of the reveived power for each antenna

Fig. 9. Peak-to-peak fluctuation of the received power at each individualantenna. From left to right, BS1C1A1, BS1C1A2, BS1C2A1, BS1C2A2,BS2C1A1, BS2C1A2, BS2C2A1, BS2C2A2.

For diversity advantage to be realised the fluctuation ofreceived power in the two channels must be decorrelated.

The correlation coefficient ρX,Y between two random vari-ables X and Y with expected values μX and μY and standarddeviations σX and σY is defined as:

ρX,Y =cov(X,Y )

σXσY=

E((X − μX)(Y − μY ))σXσY

(1)

where E denotes expectation and cov denotes covariance [18].The correlation coefficients between the antennas mounted

on the same collar are presented in Table I. From the definitionof correlation (Equation 1), the LOS component (which isessentially constant) is excluded which explains the very lowvalues of correlation.

Base Station Collar Correlation Coefficientbetween A1 and A2

BS 1 Collar 1 0.2407Collar 2 0.0414

BS 2 Collar 1 -0.0733Collar 2 0.0158

TABLE ICORRELATION COEFFICIENT OF SIGNALS RECEIVED BY ANTENNAS ON A

SINGLE COLLAR.

The probability distribution of the received signal from aparticular mobile antenna to a particular base-station might beexpected to be close to Ricean as a result of multipath propa-gation with a strong LOS component. The Ricean distributionis given by:

p(r) =

{r

σ2 e− (r2+A2)2σ2 I0(Ar

σ2 ) for (A ≥ 0, r ≥ 0)0 for (r < 0)

(2)

where A denotes the peak amplitude of the dominant signaland I0 is the modified zero order Bessel function of the firstkind [19]. The K-factor of a Ricean distribution is the ratiobetween the (constant) component of signal power due to theLOS path and the (fluctuating) component of signal power dueto all other paths, i.e.:

K =A2

2σ2(3)

As the LOS component becomes smaller K-factor decreasesand the probability density function (pdf) becomes moreskewed. As K tends to zero the Ricean distribution approachesa Rayleigh distribution. As the LOS component becomeslarger, K increases and the distribution becomes less skewed.As K tends to infinity, the Ricean distribution approaches anormal distribution.

Figure 10 shows the pdfs of the power received at BS1.Figures 10(a) and (b) represent the pdfs of the data obtainedfrom A1 and A2, respectively, regarding C1. Similarly, Figures10(c) and (d) refer to A1 and A2 on C2.

A normal distribution of power in dBm (i.e. a log-normaldistribution of power in watts) appears to be the best fit to thedata. If fading is due predominantly to multipath propagationthis suggests the presence of a strong LOS component. Analternative interpretation would be that the log-normal fadingreflects cascaded independent shadowing processes.

Figure 11 shows the pdf of the power received at BS2. Ina similar way to Figure 10, Figures 11(a) and (b) representthe data transmitted by C1 (for A1 and A2 respectively) andFigures 11(c) and (d) the data transmitted from C2 (for A1and A2 respectively).

Superficially this distribution appears to be closer toRayleigh (in dBm) than normal. The mean signal level is sig-nificantly lower than that for BS1 (due to the larger distance),however, and is approaching the receiver sensitivity which is -94 dBm. Since no signal is recorded when the received powerfalls below -94 dBm the pdf is effectively truncated at thislevel. It seems likely, therefore, that the pdf of the underlyingsignal is normal (in dBm) even though the pdf of the recorded(truncated) signal is skewed.

The corresponding cumulative distribution functions (cdfs)are presented in Figures 12 and 13. The best-fit normal curvesalong with 95 % confidence intervals are superimposed.

The close fit of the normal distribution for the data loggedat BS1 is apparent. The fit is less good for the data obtainedfrom BS2.

Figures 14 and 15 show similar plots for received voltage.Figure 16 represents similar data to that presented in Figures

7 and 8 but of shorter time duration (approximately 40minutes). The advantage of the use of base-station diversity isespecially apparent in this data. Figures 16(a) and (b) representthe power received at BS1 and BS2, respectively, from thesignal transmitted from A1 on C1. Figures 16(c) and (d)represent the power received at BS1 and BS2, respectively,from the signal transmitted from A2 on C1.

30

−90 −85 −80 −75 −70 −65 −600

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

Data

Den

sity

BS1−C1A1Normal Fit

(a)

−90 −85 −80 −75 −700

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

Data

Den

sity


(b)

−90 −85 −80 −75 −70 −650

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

Data

Den

sity


(c)

−90 −85 −80 −75 −700

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

Data

Den

sity


(d)

Fig. 10. Pdfs of power (dBm) at BS1 for (a) A1 on C1 and (b) A2 on C1and (c) A1 on C2 and (d) A2 on C2. (Smooth curve represents the best fitnormal distribution.)

−90 −85 −80 −75 −700

0.02

0.04

0.06

0.08

0.1

Data

Den

sity

BS2C1A1

(a)

−90 −85 −80 −75 −700

0.05

0.1

0.15

Data

Den

sity

BS2C1A2

(b)

−90 −85 −80 −750

0.02

0.04

0.06

0.08

0.1

0.12

0.14

Data

Den

sity

BS2C2A1

(c)

−92 −90 −88 −86 −84 −82 −80 −78 −760

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

Data

Den

sity

BS2C2A2

(d)

Fig. 11. Pdfs of power (dBm) at BS2 for (a) A1 on C1 and (b) A2 on C1and (c) A1 on C2 and (d) A2 on C2. (Smooth curve represents the best fitnormal distribution.

The horizontal lines in Figure 16 represent the mean powerfor each measurement. The correlation coefficients betweenBS1 and BS2 signals are presented in Table II.

Base-station diversity clearly offers advantage. The consis-tently small negative correlation is interpreted as being dueto essentially zero short-term correlation due to the physicallyindependent multipath propagation structure experience by thebase-stations, and a longer term negative correlation due tothe changes in distance between collar and base-stations asthe animal moves.

−90 −85 −80 −75 −70 −65 −600

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty

BS1C1A1confidence boundsNormal Fit

(a)

−90 −85 −80 −75 −700

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty


(b)

−90 −85 −80 −75 −70 −650

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty


(c)

−90 −85 −80 −75 −700

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty


(d)

Fig. 12. Cdfs of power (dBm) at BS1 for (a) A1 on C1 and (b) A2 on C1and (c) A1 on C2 and (d) A2 on C2. (Smooth curve represents the best fitnormal distribution.

−90 −85 −80 −75 −700

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty


(a)

−90 −85 −80 −75 −700

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty


(b)

−90 −85 −80 −750

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty


(c)

−92 −90 −88 −86 −84 −82 −80 −78 −760

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty


(d)

Fig. 13. Cdfs of power (dBm) at BS2 for (a) A1 on C1 and (b) A2 on C1and (c) A1 on C2 and (d) A2 on C2. (Smooth curve represents the best fitnormal distribution.

Collar Antenna Correlation Coefficientbetween BS1 and BS2

C1 A1 -0.2646A2 -0.3062

C2 A1 -0.0212A2 -0.2217

TABLE IICORRELATION COEFFICIENT OF SIGNALS RECEIVED BY DIFFERENT

BASE-STATIONS.

31

0.5 1 1.5 2 2.5 3

x 10−5

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty

BS1C1A1confidence boundsLognormal Fit

(a)

2 4 6 8 10 12 14

x 10−6

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty

BS1C1A2confidence boundsLognormal

(b)

0.5 1 1.5 2

x 10−5

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty


(c)

2 4 6 8 10 12

x 10−6

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty


(d)

Fig. 14. CDFs of detected voltage at BS1 for (a) A1 on C1 and (b) A2 onC1 and (c) A1 on C2 and (d) A2 on C2. (Smooth curve represents the bestfit log-normal distribution.)

2 4 6 8 10 12 14 16

x 10−6

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty


(a)

2 4 6 8 10 12 14

x 10−6

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty


(b)

1 2 3 4 5 6

x 10−6

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty


(c)

1 1.5 2 2.5 3 3.5 4 4.5 5

x 10−6

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Data

Cum

ulat

ive

prob

abili

ty


(d)

Fig. 15. CDFs of detected voltage at BS2 for (a) A1 on C1 and (b) A2 onC1 and (c) A1 on C2 and (d) A2 on C2. (Smooth curve represents the bestfit log-normal distribution.)

V. DIVERSITY GAIN

The advantage offered by selection diversity has been evalu-ated for each collar. Selection diversity describes the selectionof the strongest signal from the diversity channels at each timeinstant which corresponds, in this case, to the selection of thestrongest of the signals arising from the two antennas on agiven collar.

Figure 17 shows the results demonstrating a clear diversitybenefit (gain).

Figure 18 shows the cdfs of the expected diversity gain for

0 200 400 600 800 1000 1200 1400 1600 1800 2000−100

−90

−80

−70

−60

−50


RSS

I(dB

m)

Base Station 1 − Collar 1 − Antenna 1

0 200 400 600 800 1000 1200 1400 1600 1800 2000−100

−90

−80

−70

−60

−50Moving Average of Raw Data


RSS

I(dB

m)

(a)

0 200 400 600 800 1000 1200 1400 1600 1800 2000−100

−90

−80

−70

−60

−50


RSS

I(dB

m)


0 200 400 600 800 1000 1200 1400 1600 1800 2000−100

−90

−80

−70

−60



RSS

I(dB

m)

(b)

0 200 400 600 800 1000 1200 1400 1600 1800 2000−100

−90

−80

−70

−60

−50


RSS

I(dB

m)


0 200 400 600 800 1000 1200 1400 1600 1800 2000−100

−90

−80

−70

−60



RSS

I(dB

m)

(c)

0 200 400 600 800 1000 1200 1400 1600 1800 2000−100

−90

−80

−70

−60

−50


RSS

I(dB

m)


0 200 400 600 800 1000 1200 1400 1600 1800 2000−100

−90

−80

−70

−60



RSS

I(dB

m)

(d)

Fig. 16. Data received from A1 on C1 at (a) BS1 and (b) BS2, and datareceived from A2 on C1 at (c) BS1 and (d) BS2.

32

−90 −85 −80 −75 −70 −65 −600

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Signal Level (dBm)

Cum

ulat

ive

prob

abili

tySelection and Mean Diversity Gain for C1 of BS1

A1A2Selection DiversityMean Diversity

(a)

−90 −85 −80 −75 −70 −650

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Signal Level (dBm)

Cum

ulat

ive

prob

abili

ty

Selection and Mean Diversity Gain for C2 of BS1


(b)

−90 −85 −80 −75 −700

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Signal Level (dBm)

Cum

ulat

ive

prob

abili

ty

Selection and Mean Diversity Gain for C1 of BS2


(c)−92 −90 −88 −86 −84 −82 −80 −78 −76 −74

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Signal Level (dBm)

Cum

ulat

ive

prob

abili

tySelection and Mean Diversity Gain for C2 of BS2


(d)

Fig. 17. Selection and mean diversity cdfs of the two antennas on eachcollar.

(a) C1 to BS1, (b) C2 to BS1, (c) C1 to BS2 and (d) C2 toBS2. The median diversity gain averaged over all four cdfs is2.17 dB.

0 1 2 3 4 50

0.2

0.4

0.6

0.8

1

Diversity Gain (dB)

Cum

ulat

ive

prob

abili

ty

Expected Diversity Gain for C1 of BS1

Diversity Gain

(a)

0 1 2 3 4 50

0.2

0.4

0.6

0.8

1

Diversity Gain (dB)

Cum

ulat

ive

prob

abili

ty


Diversity Gain

(b)

0 1 2 3 4 50

0.2

0.4

0.6

0.8

1

Diversity Gain (dB)

Cum

ulat

ive

prob

abili

ty


Diversity Gain

(c)

0 0.5 1 1.5 2 2.5 3 3.5 4 4.50

0.2

0.4

0.6

0.8

1

Diversity Gain (dB)

Cum

ulat

ive

prob

abili

ty


Diversity Gain

(d)

Fig. 18. Diversity gain of the two antennas on each collar.

VI. CONCLUSIONS AND FUTURE WORK

An experiment to investigate the practicality of applyingantenna and base-station diversity to the wireless monitoringof farm animals has been described. Some preliminary resultsrelating to the statistical distributions of received signals

and antenna/base-station signal correlations have been sum-marised.

The advantage offered by selection diversity has been eval-uated.

REFERENCES

[1] I.F. Akyildiz, Weilian Su, Y. Sankarasubramaniam, and E. Cayirci.Wireless sensor networks: A survey. 38(4):393–422, March 2002.

[2] The Defense Advanced Research Projects Agency(DARPA). Self-healing minefield,http://www.darpa.mil/sto/smallunitops/shm/index.htm.

[3] J. K. Hart K. Martinez and R. Ong. Environmental sensor networks.37(8), 2004.

[4] A. Mainwaring, J. Polastre, R. Szewczyk, D. Culler, and J. Anderson.Wireless sensor networks for habitat monitoring. In Proceedings ofthe ACM International Workshop on Wireless Sensor Networks andApplications (WSNA’02), Atlanda, USA, 2002.

[5] H.G Goh, M.L. Sim, and H.T. Ewe. An overview of rice field monitoringusing wireless sensor networks and mobile internet application. In Pro-ceedings of the 8th International Conference on Electronics, Informationand Communication (ICEIC’06), volume 1, pages 114–117, Ulaanbaatar,Mongolia, June 2006.

[6] J. Burrell, T. Brooke, and R. Beckwith. Vineyard computing: sensornetworks in agricultural production. 3(1):38–45, 2004.

[7] Wireless mobile ad-hoc sensor networks for very large scale cattlemonitoring, Berlin, Germany, August 2006.

[8] K. Mayer, K. Taylor, and K. Ellis. Cattle health monitoring usingwireless sensor networks. In The 2nd IASTED International Conferenceon Communication and Computer Networks, November 2004.

[9] E. Callaway, P. Gorday, L. Hester, J.A. Gutierrez, M. Naeve, B. Heile,and V. Bahl. Home networking with ieee 802.15.4: a developing standardfor low-rate wireless personal area networks. 40(8):70–77, Aug. 2002.

[10] Seng-Yong Lau, Ting-Hao Chang, Shu-Yu Hu, Hsing-Jung Huang,Lung-de Shyu, Chui-Ming Chiu, and Polly Huang. Sensor networksfor everyday use: The bl-live experience. In Proc. IEEE InternationalConference on Sensor Networks, Ubiquitous, and Trustworthy Comput-ing, volume 1, pages 336–343, 2006.

[11] Kay Soon Low, W.N.N. Win, and Meng Joo Er. Wireless sensornetworks for industrial environments. In Computational Intelligence forModelling, Control and Automation, 2005 and International Conferenceon Intelligent Agents, Web Technologies and Internet Commerce, Inter-national Conference on, volume 2, pages 271–276, 28-30 Nov. 2005.

[12] P. Zhang, C.M. Sadler, S.A. Lyon, and M. Martonosi. Hardwaredesign experiences on zebranet. In Proceedings of SenSys, Baltimore,Maryland, USA, 2004.

[13] I. McCauley, B. Matthews, L. Nugent, A. Mather, and J. Simons. Wiredpigs: Ad-hoc wireless sensor networks in studies of animal welfare.In Proc. EmNetS-II Embedded Networked Sensors The Second IEEEWorkshop on, pages 29–36, 30–31 May 2005.

[14] Y. Guo, P. Corke, G. Poulton, T. Wark, G. Bishop-Hurley, and D. Swain.Animal behaviour understanding using wireless sensor networks. InLocal Computer Networks, Proceedings 2006 31st IEEE Conference on,pages 607–614, Nov. 2006.

[15] P. Corke and P. Sikka. Results from the farm.http://eecs.harvard.edu/emnets/papers/corkeEmnets06.pdf (Last visitedon 24/02/08).

[16] http://www.xbow.com, (Last visited on 18th May 2008).[17] http://www.hyperlinktech.com, (Last visited on 18th May 2008).[18] http://en.wikipedia.org/wiki/correlation, (Last visited on 7th July 2008).[19] Theodore S.Rappaport. Wireless Communications Principles and Prac-

tices. Prentice Hall PTR, second edition, 2002.

33

wireless sensor network for animal monitoring using both antenna and base-station diversity

Documents