1

A low profile, low-RF band, small antenna forunderground, in-situ sensing and wireless

energy-efficient transmissionGunjan Pandey (gpan@iastate.edu), Ratnesh Kumar, Fellow, IEEE (rkumar@iastate.edu)

and Robert J. Weber, Fellow, IEEE (weber@iastate.edu)Department of Electrical and Computer Engineering, Iowa State University, Ames, IA, 50014 USA

Abstract—A key challenge to underground, in-situ soil sensingwith wireless interface is the antenna size. Smaller operatingfrequency supports lower path losses but enhances the wavelengthand hence the size of standard monpole (8.2 cm in height at915MHz) or rectangular microstrip patch antenna (11.5 cm × 9.3cm at 915 MHz), which is prohibitive for underground sensors.Lowering the frequency below 915MHz is not an option as itonly further enhances the antenna size. To circumvent the sizeproblem, a composite right-left handed (CRLH) microstrip patchantenna for wireless transmission at 915 MHz that doubles upas an underground, sensing element (external capacitor) hasbeen designed and fabricated. The combined antenna/sensingCRLH patch integrates with the previously implemented on-board, multi-frequency dielectric based impedance sensor for soilmoisture and nitrate determination and provides almost 93%reduction over the standard microstrip antenna size. As a proofof concept, the input impedance of the CRLH sensor, surroundedby the soil containing moisture and nitrate ions, is measured atmultiple frequencies in the lab setting. It is shown that the changein moisture and nitrate can be successfully detected using thesensor. The small profile of the proposed antenna (3 cm × 2 cm),that is almost 93% smaller, makes it ideal for compact packaging.

Index Terms—Microstrip, Metamaterial, Composite-Right-Left-Handed transmission line, Spectroscopy.

I. INTRODUCTION

Nitrate based fertilizers are one of the most common typeof fertilizers used for increasing the agricultural productivity.However, excessive use of nitrate based fertilizers can leadto severe environmental hazards. A deeper understanding ofagricultural N cycling process is needed so that precise controlsover N fertilizer inputs that are key to sustainable agricul-ture can be implemented. In a generic precision agricultureapproach (see Fig. 1), intra- and inter-field variabilities arecharacterized using a network of sensor nodes spread over alarge area. Each sensor node sends local information aboutthe properties of the soil surrounding it. All the informationcollected is sent to a central node which processes the infor-mation and takes necessary control measures towards irrigationand fertilization. The in-situ, buried sensors require an efficienttransceiver system that can provide enough power to overcome

This research was supported in part by the National Science Foundationunder the grants NSF-ECCS-0801763, NSF-ECCS-0926029, and NSFCCF-1331390.

the losses incurred during signal transmission in soil and alsohave an antenna that is small enough to maintain a compactsize of the sensor.

Fig. 1. Conceptual Layout of a Soil Sensor network

Our previous work on soil-sensing, that uses modifiedquarter wavelength monopole type electrodes as probes [8],[9], has proven that multi-frequency impedance measurementsof a soil mixture have the capability to provide informationabout the soil moisture together with the concentration ofdifferent ions like nitrates in soil. In [8] we presented aself-calibrating, multi-frequency dielectric sensor for combinedmoisture and soil ions sensing, while in [9] we showed howdielectric-mixing models can be reasoned to analyze the multi-frequency dielectric measurements to estimate the soil moistureand ion concentrations. We have also shown that the quarterwavelength monopole antenna can be used dually as a sensorprobe as well as a transmitting/receiving antenna. This wasachieved by separating the low-frequency sensing path fromhigh frequency transmission path using a diplexer.

A limitation of the monopole electrodes is their size, whichat carrier frequency of 915 MHz must be 8.2 cm standing ver-tically on a horizontal ground plane (Note we chose a carrierfrequency of 915 MHz, as although a lower frequency willoffer a superior range, the size of the antenna would becomeeven larger). An antenna height of 8.2 cm is clearly not veryconvenient for underground applications as the in-situ natureof agricultural application calls for a small embedded antenna.Integration of our soil sensors with microstrip antenna, a flat

2structure, can make the sensors more compact and more usablefor in-situ operation. For wireless interfacing, planar microstripantennas exist and are widely used owing to their small size,low cost and ease of integration. Such antennas have also beenused to dual as sensing probes [13] in soil-moisture sensingapplications.

Microstrip patch antenna/probe combination has also beeninvestigated as part of our own earlier research [7]. The inputimpedance of the microstrip patch was shown to vary withsurrounding nitrate and moisture concentrations and was usedto detect the changes in moisture and nitrate concentrations insoil. While flattened in vertical dimension, regular microstrippatch antennas still suffer the size issue since they are not smallenough in size in the other two dimensions (e.g. 11.5 cm x 9.3cm at 915MHz carrier frequency). The 11.5 cm x 9.3 cm sizeof a flattened regular patch antenna size is an improvementover a monopole, but the size is still much larger compared tothe rest of the circuit.

This work presents a metamaterial inspired small flat an-tenna that provides a practical solution for an undergroundapplication. The main contributions of this work are:

1) A new metamaterial inspired CRLH antenna that reducesthe antenna size by about 93% of the original patchantenna.

2) Application of the CRLH antenna as the sensing elementby using a diplexer that allows the use of the CRLHpatch as a probe at low frequencies and as transmit-ting/receiving antenna at higher frequencies.

3) Mapping the measured input impedance of the CRLHpatch embedded in its surroundings to its complex per-mittivity.

4) Improvement in the accuracy of the sensor by accountingfor the influence of the parasitic capacitances in ourmeasurements.

The new antenna design along with an inbuilt self-calibrating mechanism makes our sensor suitable for under-ground application such as soil nitrate management or for ahand held device such as in food safety or microbial detectionapplications. A multi-power mode transceiver system has beendesigned to support the implementation of an energy efficientmedium-access-control (MAC) protocol. Rest of the paper isorganized as follows: Section II provides an overview of ourmulti-frequency impedance measurement system. Section IIIpresents a discussion on design and fabrication of CRLHpatch. Section IV presents the experimental validation of realand imaginary parts of impedance measurements over thefrequencies of 1-40MHz. Section V concludes the paper.

II. OVERVIEW: MULTI-FREQUENCY IMPEDANCEMEASUREMENT

We have recently designed and tested a dielectric measure-ment based soil impedance sensor that can sense at multi-frequencies (hence accurate & reliable), is self-calibrating(hence robust), possesses wireless interface (hence can belocated in-situ), and is also energy-efficient [8]. The sensor

architecture, consisting of probe and antenna, directional cou-plers, phase locked loop (PLL), amplitude and phase detector,switches/diplexer, microprocessor & transceiver, is shown inFig. 2.

Fig. 2. Dielectric sensor architecture.

Upon startup, the microprocessor programs the I2C inter-face of the programmable PLL to generate a signal of knownfrequency. The frequency of the probing signal is chosen inthe range of 1-30 MHz, and is chosen so that a significantvariation in real and imaginary part of the soil impedance canbe observed. While the lower limit of 1 MHz on frequencyis put by the architecture of the sensor, the upper limit of30 MHz was obtained experimentally as above this value thesoil reactance becomes close to zero. A slight increase in thisvalue can provide more data points to analyze but beyondthat no useful information on soil ionic concentration canbe extracted from it. The probing signal is sent through thetransmission line to the SP6T switch, which is programmedby the microprocessor to select among a set of known loadsplus the unknown soil-sample load. The incident and reflectedsignals to and from the load are captured using the directionalcouplers and are passed on to a detector which calculates theamplitude and phase of each signal and passes this informationto the microprocessor for further processing and transmissionvia antenna. These values are received by the microprocessorthrough an in-built 12-bit ADC. In the calibration mode,when the loads are of known values, these values are used tocalculate the calibration parameters that correlate the reflectioncoefficients (ratio of reflected to incident) measured at thecouplers to those at the load through a 3-parameter bilineartransform. In the measurement mode, when the load is thesoil-sample, these calibration parameters are used to find outthe reflection coefficient for an unknown load from its valuemeasured at the directional coupler, through the same bilineartransform whose parameters were determined in the calibrationmode. The reflection coefficient value is then used to determinethe unknown load impedance that contains the informationabout the soil contents (moisture and nutrients).

The resistive versus reactive soil impedance measurementsover 1-30 MHz by our sensor are shown in Figs. 3 and4. The accuracy of our in-situ sensor is confirmed againstthe measurements from a lab equipment, a network analyzer,HP8714ES (plots also shown in the same figures). A more than

390% accuracy over the range of 3-30 MHz in soil reactancewas observed.

Fig. 3. Measured soil resistance.

Fig. 4. Measured soil reactance.

III. CRLH PATCH ANTENNA/SENSOR-PROBE

The propagation constant for a signal traveling in soil isgiven by:

γ =√j2πfµ(2πf(ε” + jε)) = α+ jβ (1)

where f is the transmission frequency of 915 MHz, α is theattenuation factor while β is the phase constant. On solvingthis equation for α we get:

α = 2πf

√µε

2(

√1 + (

ε′′

ε′)2 − 1), (2)

where ε′ and ε′′ are the real and imaginary parts of soilpermittivity. It can be observed that the attenuation factor αincreases linearly with frequency (and so loss exponentiallywith frequency). Thus, increasing the frequency, f , increasesthe losses in the transmission signal and decreases the antennarange. On the other hand, lowering the frequency (to counterpath losses) increases the width, W as well as the length, Lof a patch antenna as can be seen from the equations (3) and(4) in which f appears in the denominator [1].

W =c

2f√

εr+12

, (3)

L =c

2f+ 2∆L. (4)

Fig. 5. Dimensions of the CRLH patch.

Here, εr is the relative permittivity of the substrate and ∆L isthe length correction factor given by:

∆L = 0.412hεeff + 0.3Wh + 0.264

εeff − 0.258Wh + 0.8. (5)

In (5), h is the height of the dielectric substrate and εeffis the effective permittivity due to multiple media (substratedielectric and air) involved, and is given by:

εeff =εr + 1

2+εr − 1

2[1 + 12

h

W]−

12 . (6)

Since√εr + 1 appears in denominator of (3), it seems that the

size of the antenna can be reduced by using a very high per-mittivity substrate. But a problem is that the antenna efficiencyalso goes down with increasing substrate permittivity.

For the transmission frequency of 915 MHz, the length andwidth were calculated to be 11 cm and 8.8 cm respectively fora relative substrate permittivity of 3.55 and substrate height of0.813 mm. The ground plane size was decided based on theanalysis presented in [5] as: (L+6h)×(W +6h) = 11.5 cm ×9.3 cm. Since the sensor circuitry can be designed to fit into arelatively smaller size (7 cm × 5 cm for our on-board design),the antenna size is the limiting factor in overall sensor size.To address this technological challenge, we look beyond thestandard materials, towards the so called metamaterials.

Metamaterials are specially constructed designs which offerelectric and magnetic properties opposite of materials foundin nature such as negative permittivity and permeability. Thiscreates a possibility of engineering a small-sized metamaterialsmatching network between the antenna and the surroundingmedium so that the energy stored in the near field is radiatedaway. Size improvements of factor greater than 10 comparedto standard antennas have been observed in [2], [15], [14].Authors in [2] discuss a composite right-left handed (CRLH)transmission line based antenna, with potentially improved effi-ciency for small patch antennas. Another such implementationhas been reported in [3]. In this paper, we present anotherdesign based on a standard CRLH structure [2] that can alsobe dueled as an underground sensing element (see Fig. 5 andits fabrication in Fig. 8).

Resonance in CRLH type antenna can be understood byconsidering the unit-cell structure in a small patch resonating

4

Fig. 6. Unit cell structure for a right handed transmission line.

structure. For a regular transmission line, the well-known dis-tributed parameters structure is depicted by a series inductancefollowed by a shunt capacitance as shown in Fig. 6. Suchstructure supports only the right-handed wave propagationwhich means that the phase shift observed in an incidentsignal along the length of structure is positive [6]. In a CRLHstructure, this phase shift can be either positive or negative dueto the effect of apparent negative permittivity and permeabilityin the structure. One type of structure that can achieve thisapparent negative permittivity/permeability is shown in Fig. 5,with its distributed parameter model depicted in Fig. 7, whichcontains additional series capacitances and shunt inductances.For such a structure, the series impedance is given by:

Zseries = j(2πfLs −1

2πfCs). (7)

Similarly, shunt reactance is given by:

Yshunt = j(2πfCsh −1

2πfLsh). (8)

It has been shown in [2] that better efficiency for a CRLH

Fig. 7. Unit cell structure for a CRLH transmission line.

antenna is achieved when series and shunt parts of the structure

resonant at same frequency which is given by:

ωresonant = 2πfresonant =1√LsCs

=1√

LshCsh(9)

In our antenna, series capacitance is introduced by the interdigitized finger capacitor while shunt inductance was realizedusing a meander shaped microstrip stub (see Fig. 5 and its fab-rication in Fig. 8). The CRLH antenna structure was simulatedusing ADS (Advanced Design System, Agilent Technologies)software and resonance was observed for both series and shuntstructures at the desired frequency of interest (915 MHz).

Fig. 8. Fabricated CRLH patch antenna.

The antenna is fabricated on a pc board (see Fig. 8) withrelative substrate permittivity 3.55, thickness 0.813 mm and di-electric loss tangent of 0.002 (Rogers R4003C laminate). Fromsimulations it was observed that although 5% improvement inefficiency can be achieved by reducing dielectric constant to2, the size requirement goes up by 60%. The thickness waschosen to fit the antenna structure in a compact volume in orderto keep a small size for the overall sensor. Hence, the chosenvalues give a good balance between efficiency, availability andsize for a patch structure. The metal on the top has a thicknessof 0.0355 mm and a conductivity of 5.8e7 S/m (copper). Thedimensions of the antenna are shown in Fig. 5 while Fig. 8shows the actual fabrication. Resonant frequency of 915 MHzis demonstrated in the reflectivity plot of Fig. 9. The antennadimensions are 3 cm × 2 cm which is 93.3% smaller in areathan the standard patch antenna of size 11.5 cm × 9.3 cm).

TABLE ISIGNAL ATTENUATION FOR A SENSOR BURIED 1M BELOW GROUND

Real relative Imaginary relative Attenuation in soil Range in air above soilpermittivity permittivity (dB) (m)

5 1 -31.25 40175 2 -35.75 23925 3 -40.03 14605 4 -44.05 9195 5 -47.80 597

The built in transceiver system in our sensor can transmitat a maximum power of 25 dBm while the receiver has asensitivity of -110 dBm, meaning a 135 dB path-loss can betolerated. The range of our antenna can be calculated usingFrii’s equation for path loss:

Pr(dB) = Pt(dB) +Gt(dB) +Gr(dB) + 20log10(c

4πfr)

+20log10(e−αr)(10)

5where Pr and Pt are received and transmitted powers, Gtand Gr are transmitting antenna gain and receiving antennagain, and r is the distance between transmitting and receivingantenna. In our setting, Pr = −110 dBm, Pt = 25 dBm,Gr = Gt = 0 dBm, f = 915 MHz, and α is determinedusing Equation 2 by plugging ε′ and ε′′ values from Table I,which also lists the calculated ranges r, assuming that antennais buried 0.25 m below the soil surface.

The real part of soil permittivity is dominated by salinewater concentration, whereas the imaginary part of the soilpermittivity is governed by the concentration of free ions.According to [4], maximum value of nitrate concentration ina typical clay-loam field is 200 mg−1L−1 which results in asoil permittivity of approximately 5 + j5 at 20% by volumemoisture content for frequencies above 14 MHz. Hence, wehave used this value as the worst case scenario for makingrange calculations; the other range values are calculated atthe same moisture level but lower nitrate concentrations sothe soil permittivity ranges from 5+j to 5+5j, yielding rangevalues between 4017 m and 597 m. It can be seen thateven for large values of nitrate concentration in soil (whichcause the imaginary part of relative permittivity to rise) whilekeeping the same moisture level, the antenna range remains597 m. Thus, the proposed antenna can be effectively used tocommunicate with above the air satellite/base station locatedin the field while the sensors are buried 0.25 m below theground.

IV. TEST RESULTS

A. Comparison with Network Analyzer

The soil impedance was measured by our on-board sensor[8] and for comparison of its accuracy also by a NetworkAnalyzer (HP8714ES). The CRLH patch sensing element,along with its surrounding medium, namely soil, presents itselfas a load impedance to the sensor. The measurement datarecoded by the on-board sensor is transmitted to a receiverwhich first calculates the calibration parameters and usingthose, calculates the unknown load of the surrounding soil.The imaginary part of impedance measured using the on-boardsensor showed a better than 85% match with those measuredusing the network analyzer in the range 1-40 MHz while realpart showed accuracy better than 70% for frequencies above15 MHz. (see Figs. 10 and 11). The lack of accuracy in realpart could be due to non-uniform distribution of ions in thevicinity of the sensor during two sets of measurement. Also

Fig. 9. Return loss for the fabricated antenna.

Fig. 10. Comparison of measured input reactance of patch buried in soilmeasured with on-board sensor and network analyzer.

Fig. 11. Comparison of measured input resistance of patch buried in soilmeasured with on-board sensor and network analyzer.

the higher range of 1-40 MHz as compared to the previous1-30 MHz provides a larger dataset so that a more informativeanalysis on soil ionic concentration can be carried out.

B. Admittance variation with varying moisture and nitrateconditions

Figs. 12 and 13 show the variation in patch admittancewith changing values of sodium nitrate solution. A 100 milimolar sodium nitrate solution was added in steps of 4% byvolume increments to the soil that had sensor with the patch,acting as a probe (as well as antenna, at another frequency),buried into it. It was observed that the measured conductance(reciprocal of the real-part of impedance) of the patch in-creased as the concentration of sodium nitrate was increased

6

Fig. 12. Conductance variation with varying sodium nitrate solution concen-tration.

Fig. 13. Susceptance variation with varying sodium nitrate solution concen-tration.

in soil, whereas there was a much smaller variation in thesusceptance (reciprocal of the imaginary-part of impedance)value. This demonstrates that the accurate measurement of soilimpedance (equivalently, admittance) using microstrip patchsensing probe at multiple frequencies has the potential tosuccessfully detect changes in ionic concentration in soil. Thedielectric mixing models [12] that determine the permittivityof a mixture as a function of the composition and content ofthe mixture, together with the dielectric relaxation models [11]that determine the permittivity as a function of the frequencycan be employed to estimate the concentrations of moistureversus nitrates versus air in the soil from the measurements,as is the case in [10].

V. CONCLUSION AND FURTHER WORK

An on-board self-calibrating multi-frequency dielectric sen-sor with small sized planar patch for sensing as well as wirelessinterfacing was designed, fabricated and validated against anetwork analyzer. The sensor was shown to accurately measurethe soil impedance at multiple frequencies over 1-40 MHz,

with less that 15% error in reactance when compared toa benchtop network analyzer (HP8714ES). The impedancesmeasured by the sensor is useful in estimating the contentsof individual ions and moisture in soil. This work improvesupon our previous work on underground soil moisture andnitrate sensing [9], [10] by reducing by almost 93% antennadimensions thus allowing the design of a compact overallsensor size, making it suitable for field-deployment and hand-held applications.

Currently, we are working towards developing a model torelate the input impedance of this CRLH patch sensor to thesurrounding permittivity value. Such models can be appliedto data obtained in this work to determine the permittivity ofthe surrounding soil, and the permittivity values at multiplefrequencies can then be used to estimate ionic concentrations[9], [10].

REFERENCES

[1] Constantine A. Balanis. Antenna Theory: Analysis and Design, 3rdEdition. John Wiley and sons Inc., Arizona State University, 2005.

[2] Christophe Caloz, Tatsuo Itoh, and Andre Rennings. Crlh metamaterialleaky-wave and resonant antennas. Antennas and Propagation magazine,IEEE, 50(5):25–39, October 2008.

[3] Zhou Cheng, Wang Guang-ming, and Liang Jian-gang. Novel zeroth-order resonator antenna based on composite right/left handed (crlh)transmission line. In Cross Strait Quad-Regional Radio Science andWireless Technology Conference (CSQRWC), 2011, pages 379–381,2011.

[4] Giorgi Chighladze, Amy Kaleita, Stuart Birrell, and Sally Logsdon.Estimating soil solution nitrate concentration from dielectric spectrausing partial least squares analysis. Soil Science Society of AmericaJournal, 76(5):1536–1547, November 2011.

[5] G. Kumar and K.P. Ray. Broadband Microstrip Antennas. Artech HouseInc., 2003.

[6] Shaowei Liao, Jianhua Xu, Feng Wan, Wenxiang Wang, and YubingGong. Left-handed/right-handed transmission line subwavelength cavityresonators. IEEE ANTENNAS AND WIRELESS PROPAGATION LET-TERS, VOL. 8, 2009, 8(5):80–83, April 2009.

[7] Gunjan Pandey, Ratnesh Kumar, and Robert J. Weber. Design andimplementation of a self-calibrating, compact micro strip sensor for in-situ dielectric spectroscopy and data transmission. In IEEE Sensors 2013,Baltimore, MD, 2013.

[8] Gunjan Pandey, Ratnesh Kumar, and Robert J. Weber. A multifrequency,self-calibrating, in-situ soil sensor with energy-efficient wireless inter-face. In Defense, Sensing and Security Conference, SPIE, Baltimore,MD, 2013.

[9] Gunjan Pandey, Ratnesh Kumar, and Robert J. Weber. Real timedetection of soil moisture and nitrates using on-board in-situ impedancespectroscopy. In IEEE System, Man and Cybernetics Society 2013,Manchester, UK, 2013.

[10] Gunjan Pandey, Ratnesh Kumar, and Robert J. Weber. Real timedetection of soil moisture and nitrates using on-board in-situ impedancespectroscopy. In Defense, Sensing and Security Conference, SPIE,Baltimore, MD, 2013.

[11] J. Carlos Santamarina, K. A. Klein, and Moheb A. Fam. Soils andWaves: Particulate Materials Behavior, Characterization and ProcessMonitoring. J. Wiley and Sons, J. Carlos Santamarina, Georgia Instituteof Technology, 2001.

[12] A H Sihvola and J A Kong. Effective permittivity of dielectric mixtures.IEEE Transactions on Geoscience and Remote Sensing, 26(4):420–429,July 1988.

[13] Pichitpong Soontornpipit, Cynthia M. Furse, You Chung Chung, andBryan M. Lin. Optimization of a buried microstrip antenna for simultane-ous communication and sensing of soil moisture. IEEE TRANSACTIONSON ANTENNAS AND PROPAGATION, 54(3):797–800, March 2006.

[14] Richard W. Ziolkowski and Aycan Erentok. Metamaterial-based effi-cient electrically small antennas. IEEE Transactions on Antennas andPropagation, 54(7):2113–2130, July 2006.

7[15] Richard W. Ziolkowski, Peng Jin, and Chia-Ching Lin. Metamaterial-inspired engineering of antennas. Proceedings of the IEEE, 99(10):1720–1731, October 2011.