Feasibility of a Millimeter-Wave MIMO System for Short-Range Wireless Communications in an...

4296 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 61, NO. 8, AUGUST 2013

Feasibility of a Millimeter-Wave MIMO Systemfor Short-Range Wireless Communications in an

Underground Gold MineIsmail Ben Mabrouk, Member, IEEE, Julien Hautcoeur, Larbi Talbi, Senior Member, IEEE,

Mourad Nedil, Senior Member, IEEE, and Khelifa Hettak, Senior Member, IEEE

Abstract—The performance of multiple-input-multiple-output(MIMO) system operating at the 60 GHz band is investigatedbased on experimental data in a real underground mine gallery.However, the millimeter wave (mmW) channels face some chal-lenges such as high propagation loss. In order to overcome thisissue, a planar microstrip antenna array has been designed, andfabricated. Moreover, the effect of miners’ activity in the vicinityof the short-range wireless link is studied. Statistical parametersof the propagation channel, such as RMS delay spread, pathloss, K-factor, channel correlation and capacity are extractedand analyzed. Results suggest that miners presence substantiallyaffects both received power and time dispersion parameters andshould therefore be considered when developing undergroundmine wireless networks in the unlicensed 60-GHz band.

Index Terms—Millimeter wavelength, multipath, propagation,underground mine.

I. INTRODUCTION

F UTURE wireless communication systems dedicated tounderground mine applications are calling for increasing

data rates [1], [2]. Video surveillance through infrequent snap-shots in mine gallery is very important to ensure miners safety.In the same way, the use of vehicular radar systems for collisionavoidance, remote control applications are also of interest tothe mine operators so that machinery can operate in extremeconditions. Therefore, MIMO technology at mmW is a verypromising candidate [3].With a huge spectrum of 5–7 GHz allo-cated as an unlicensed band worldwide, the 60-GHz frequencyrange has become attractive for future indoor networking.This frequency band offers the opportunity of developing newindustrial applications in the field of short-range radar, e.g.,high-resolution target detection, classification, miners trackingand imaging [4].

Manuscript received August 14, 2012; revised March 28, 2013; acceptedApril 12, 2013. Date of publication April 23, 2013; date of current version July31, 2013.I. Ben Mabrouk and L. Talbi are with the University of Quebec in Outaouais,

Computer Science and Engineering, Gatineau, QC J8X 3X7, Canada.J. Hautcoeur is with the Universit du Qubec en Outaouais, Informatique et

Ingnierie, Pavillon Lucien-Brault, Gatineau, QC J8X 3X7, Canada.M. Nedil is with the University of Quebec At Abitibi-Temiscamingue

(UQAT), Underground Communication, Val d’OR J9P 1S2, Canada.K. Hettak is with the Communication Research Center, Ottawa, Ontario,

Canada.Color versions of one or more of the figures in this paper are available online

at http://ieeexplore.ieee.org.Digital Object Identifier 10.1109/TAP.2013.2259454

Additionally, this solution also offers several advantagescompared to lower frequencies, namely, (1) very high data ratescan be reached (on the order of several Gbps) because of thelarge available spectrum, (2) low interference with neighboringnetworks due to the oxygen resonance around 60 GHz, and(3) creation of small-size antennas of just few cm (low oncomplexity, power efficient, and light in weight compared withsimilar systems operating at microwaves) which, for example,facilitates the use of multiple antennas at user terminals.On the downside, the potential of 60 GHz communications

is to some degree hampered by channel characteristics that im-pose tremendous challenges for system design and operation.The wireless systems experience large propagation loss com-pared to 2.4 GHz or ultra-wide-band (UWB) [5], and this is dueto the small wavelength. Consequently, this can be compensatedusing directive antennas with high gain while maintaining smallantenna dimensions. When such antennas are used, however,antenna obstruction (e.g., by a human body) may easily cause asubstantial drop of received power, which may nullify the gainprovided by the antennas. This effect is typical for millimeterwaves [6]. Therefore, during our measurement campaign, thehuman activity effect on theMIMO channel performance is con-sidered and studied.Even though MIMO propagation at low frequency

(e.g. 2.4 GHz, 5.8 GHz) and single-input-single-output(SISO)-mmW research have been well documented in theliterature [7]–[10], only a few studies on the MIMO- mmWare available. However, to the best of our knowledge, nomeasurements of radio propagation in underground mineshave studied the MIMO-mmW systems, including the effect ofminers’ activity. In [11] MIMO-mmW channel measurementsin line of sight (LoS) and non-LoS are presented. The MIMOchannel capacity is calculated for various antenna array sizes.A study conducted in [12] considered beamforming system inmmWwireless personal area networks (WPAN) as a solution tocompensate the performance degradation at 60 GHz. Also, thespatial multiplexing at mmW carrier frequencies for short-rangeindoor application is investigated in [5], [13]. Moreover, theeffect of human body shadowing on mmW wireless links hasreceived some coverage in the literature. In [14], it is reportedthat human body shadowing can cause attenuations of greaterthan 20 dB on indoor 60-GHz device-to-device links. Overall,the research on MIMO-mmW is still in its infant stage.This paper reports the design and fabrication of a 2 2

MIMO antenna based on microstrip patch arrays suitable for

0018-926X/$31.00 © 2013 IEEE

BEN MABROUK et al.: FEASIBILITY OF A MILLIMETER-WAVE MIMO SYSTEM FOR SHORT-RANGE WIRELESS COMMUNICATIONS 4297

Fig. 1. Underground gallery map.

short range mmW applications. Followed by experimental mea-surements to study the propagation characteristics of a MIMOsystem within an underground mine environment. Extensivedata were collected, and main channel parameters are extractedsuch as RMS delay spread, K-factor, channel correlation andcapacity. Two scenarios are considered: (1) The environmentis physically static, and (2) the human motion (miners moving)is added to the propagation environment.The remainder of this paper is organized as follows. In

Section II, the underground environment is described, theexperimental setup for the MIMO channel measurements andthe antennas characteristics are illustrated. Experimental resultsare studied and discussed in Section III. Finally, Section IVconcludes the paper.

II. EXPERIMENTAL SETUP

A. Description of the Underground Mining Environment

Underground mines are very challenging environments forradio communications. The radio signal behavior is very dif-ferent in such environments compared to that in other regularenvironments. It has extensive labyrinths, rough sidewalls’ sur-face that exhibits a difference of 25 cm between the maximumand minimum surface variations, and a non uniform topology.In fact, measurements are performed in a gold mine. This is veryimportant from the perspective of radio communications, espe-cially at the mmW frequencies, since electromagnetic character-istics, such as dielectricity and conductivity, are different fromother confined areas. The mine gallery is located at 40 m deepunderground level. The gallery stretches over a length of 75 mwith a width and height both of approximately 5 m. A map ofthe underground gallery is shown in Fig. 1. The humidity is stillhigh, drops of water falling from everywhere and big pools ofwater cover the ground. The temperature may vary from 6 Cto 15 C along the year.

Fig. 2. Photography of the fabricated MIMO antenna.

B. MIMO Antenna Setup

Microstrip patch antennas are selected for our study due totheir small compact size, light weight, and planar structure. Toincrease the gain of the antenna, planar arrays of 4 elementswere designed. The arrays are fed by microstrip line feed net-works with T-junctions. In fact, a single patch antenna radiatesa broad beam toward the upper hemisphere and has a small an-tenna gain, typically close to 5 dBi. To fulfil the specifications ofthe link budget, the antenna gain has to be increased by placingseveral patches into an array configuration. The array gain is in-creased approximately by 3 dBi as the number of elements isdoubled. However, as the number of elements increases, alsothe losses of the feed network become higher due to a largerfeed network. Feed-network losses become a critical issue withlarge antenna arrays at mmW frequency bands since the con-ductor loss of the microstrip line is in the order of 1.2 dB/cmat 60 GHz [15]. For the purpose of this work, the number ofradiating elements for each antenna set is chosen to be 4 to ob-tain an adequate measured antenna gain, of about 11 dBi. Thequarter-wave matched T-junction is used as a power splitter forthe patch antenna array due to its simple structure and ease ofimplementation withmicrostrip lines. To operate as aMIMO an-tenna, a set of the designed antenna array is printed on the samesubstrate distant by a wavelength (5 mm at 60 GHz) from thecenter to center. A photography of the fabricated 2 2 MIMOantenna is shown in Fig. 2.The performance of the MIMO antenna is evaluated through

S-parameters using an on-wafer probe station and the gain mea-surements in an anechoic chamber. A good input matching isobtained for each array within the 60 GHz band. The return losspresented in Fig. 3 is below the target 10 dB between 57 GHzand 64 GHz. Moreover, the scattering parameters corre-sponding to the mutual coupling is plotted in Fig. 4. At mmW,antenna elements are located closely to each other. Hence, theelectric field generated by one antenna alters the current dis-tribution on the other antennas. As a consequence, the radiationpattern and input impedance of each array element are disturbedbecause of the other elements. This effect is known as mutualcoupling.Theoretical work has shown that mutual coupling has a

significant effect on MIMO channel correlation. Typically,


Fig. 3. Simulated (HFSS) and measured .

Fig. 4. Simulated (HFSS) and measured .

in MIMO systems, independent and uncorrelated signalingbetween channels is required to improve channel capacity [16].In our case, the measured coupling level is found to be about20 dB within our band of interest.Measurements agree with simulations and small differences

at the center frequencies due to the deviations in the dimensionsof the fabricated antennas. Furthermore, the measured gainranges between 10.1 dBi and 10.8 dBi within the mmW band.Good agreement between the measured and simulated gainplots is shown in Fig. 5.

C. Measurement Campaigns

Our analysis is based on channel measurements in the mmWfrequency ranging from 57 to 64 GHz, i.e., a bandwidth of

using a vector network analyzer. The systemis mainly composed of two identical sets of patch antennaarrays connected to the input and output ports of the VNA,two PIN switches, one power amplifier for the transmittingsignal and one low noise amplifier for the receiving signal.

Fig. 5. Simulated and measured gain.

Fig. 6. MIMO measurement setup.

Both amplifiers have a gain of 30 dB each. All measurementequipments operate at the 60 GHz band. A schematic setup ofthe system is shown in Fig. 6.Also, the measurement parameters are listed in Table I. Every

channel is measured sequentially using the switches assuming aquasi-static channel, and the overall measurement system is cal-ibrated to remove frequency-dependent attenuation and phasedistortion. Due to humidity in underground mine galleries, thephase stability is measured prior to measurements.In fact, the measurements were performed in an underground

mine gallery. Mining machinery are usually placed close tothe wall for the excavation. Therefore, the transmitting andreceiving antennas are fixed in the middle of the gallery toavoid machinery obstruction which can strongly affect thereceived signal.


TABLE I60 GHz CHANNEL MEASUREMENT PARAMETERS

Two different measurement scenarios were conducted. Thefirst one corresponds to a LoS scenario; the transmitter (TX) andreceiver (RX) are finely adjusted to obtain the best line of sight(LoS) signal level. The gallery was empty with no miners’ ac-tivity, therefore, the channel can be considered as physically sta-tionary. The second measurement scenario is as follows. Onlytwo miners are in motion in this experiment, walking back andforth along a predetermined path with a proximity of 15 cm par-allel to that of the LoS as indicated in Fig. 7. However, anyminor movement along the Y-axis leads to a complete change ofthe results. Moreover, the miners stop walking during channelsnapshots at each receiver position. For all experiments, the TXremains fixed at almost one end of a long (75 m) gallery, whilethe RX was moving along a LoS route from 1 m up to 10 m farfrom the transmitter at steps of 1 m. Since at 60 GHz the wave-length is only 5 mm, in order to realize accurately the measuredpositions, an automated system was used to precisely positionthe receiver antenna along a linear track. The accuracy of thelinear track is 2.5 . In order to estimate the local mean power,the received signal power is averaged over a local area [17]. Thelocal mean has been extracted for each location separately byaveraging the measured power of six static measurements takenat six different positions of the receive antenna around the spec-ified location. Therefore, the total number of channels per snap-shot is and the total number of measured channel is

for each scenario. These six positions were sep-arated from each other approximately by one wavelength alongthe x axis and half the wavelength along the y axis over an areaof 25 mm, resulting in six different receiver antenna placementsas shown in Fig. 8, so as to obtain independent measurements.For each measurement position, the complex transfer func-

tion is

(1)

.and are the measured magnitude and phase re-

sponses at frequency on position . The subchannels arethen used to generate the MIMO transfer functions for eachposition.

Fig. 7. Scenarios of miners movement.

Fig. 8. Measurement procedure of the local mean power.

III. EXPERIMENTAL RESULTS

The underground mine propagation channel is complicatedand random, as different structures cause different propaga-tion phenomena, like reflection, refraction, diffraction, andscattering, which result in multipath propagation. The wirelesschannel can be described by its impulse response.For any fixed location between transmitter and receiver, the

overall average of the magnitude squared of the impulse re-sponse is referred to as the power delay profile (PDP) and isgiven by

(2)

The scattering parameter was measured over the mmWbandwidth and the Inverse Discrete Fourier Transformation(IDFT) was applied. In fact, the PDP is estimated by averagingmeasurements taken at the specified locations of the receiveantenna. Moreover, in order to examine the human effectpresence on the propagation channel, two PDPs are presentedcorresponding to the LoS and human effect presence scenariosat a separation distance of 4 m as shownin Fig. 9 and Fig. 10, respectively. The specular components,which appear as peaks or spikes in the PDP were more clearlyidentifiable in the presence of miners. Moreover, it is interestingto note the absence of the direct path. This can be explainedby the close proximity of the miners to the LoS path. Also an


Fig. 9. Measured PDP sample for LoS scenario.

Fig. 10. Measured PDP sample for human effect scenario.

attenuation of about 14 dB relative to that of LoS scenario is ob-served. Therefore, investigating the effect of miners movementand how this perturbs the multipath signals is very importantfor a successful implementation of the MIMO-mmW systemsin an underground mine gallery.

A. RMS Delay Spread

It is the most commonly used statistical parameter to describethe time domain dispersion of a radio channel. It roughly char-acterizes the multipath propagation and gives us an idea aboutthe flatness of the channel in frequency. It is the square root ofthe second central moment of the averaged power, and it is de-fined as

(3)

where is the mean excess delay, is the average power and isthe received power (in linear scale) at corresponding arrivaltime of the path [18]. The RMS delay spread has been com-puted for each receiver position using (3) under both scenarios.Fig. 11 shows the cumulative distribution function (CDF) ofthe RMS delay spread. A threshold level of 20 dB is chosen in

Fig. 11. Cumulative distribution function of RMS.

TABLE IISTATISTICS OF THE

order to suppress the noise effect on the statistics of multipatharrival times. Such threshold level is considered as a relevantchoice for reliable channel-parameter estimation. The min/maxand mean/standard deviation (Std) of the for differentscenarios have been computed from the time domain responsesand summarized in Table II.From Table II, it is shown that the RMS ranges from 3.08 ns

to 11.22 ns for LoS scenario and from 4.01 ns to 24.86 ns forhuman effect scenario. The latter exhibits higher mean delayspread of about 7.5 ns. This is expected because the moremasking obstacles the signal has to penetrate, the higher thetime dispersion is.

B. K-Factor

The K-factor, is a useful measure of the communication linkquality. Therefore, estimation of K is of practical importancein a variety of wireless scenarios, including channel characteri-zation, link budget calculations, and adaptive modulation [19].Moreover, recent advances in space-time coding have shownthat the capacity and performance of MIMO systems dependon the Ricean K-factor [20]. It is demonstrated in [21] that at afixed SNR level, higher K-factor means more spatial correlationand hence a decrease in channel capacity. The Ricean K-factoris measured for each distance between the transmitter and thereceiver. This parameter characterizes the relative strength ofthe direct path signal power to that of the reflected (scattered)signals. In our data processing, the K-factor is estimated frommeasurement data for each MIMO channel using the methodpresented in [22], averaged over its all corresponding SISO sub-channels. For a channel response H, the K factor is estimated as

(4)


Fig. 12. Cumulative distribution function of K factor.

TABLE IIISTATISTICS OF K-FACTOR

where denotes the expectation (mean) value and var corre-sponds to the variance of H. The CDF is shown in Fig. 12. LowerK-factor values by about 2.33 dB are obtained, when includingminers’ movement within the MIMO propagation channel. Thismeans that the multipath phenomenon is more important thanwhen having LoS. Moreover, larger standard deviation is ob-tained for the human effect scenario. This can be explained bythe random movement of the miners between the transmitterand the receiver. Therefore, the human presence in the vicinityof a MIMO mmW system has an impact on the channel per-formance. The statistical parameters of the K-factor associatedwith the curves in Fig. 12 are summarized in Table III.

C. Correlation Properties of the MIMO Channel

It is well known that the necessary and sufficient condition fora linear increase in MIMO channel capacity is the presence oforthogonal subchannels. The independence of the subchannelsfading statistics can be tested through correlation analysis.The correlation of the channel gains to two different receivers

from the same transmitter (receiver correlation) is calculatedfirst and then, the correlation of the channel gains from twodifferent transmitters to the same receiver (transmitter correla-tion) [23].The correlation coefficients between antenna elements at both

transmit and receive sides can be expressed,respectively, as

(5)

(6)

Fig. 13. Cumulative distribution function of (a) transmit correlation, and (b)receive correlation.

where denotes the correlation between and is givenby

(7)

where E[.] is the expected value operator. The denominator in(7) normalizes the random variables, and, therefore, all corre-lation coefficients are up-bounded in absolute value by unity.Fig. 13 shows the CDFs of the receive and transmit correlation.Results show that the MIMO channel in the presence of

miners is almost 20% less correlated than when the LoS exist.Moreover, the transmit correlation is found to be higher than thereceive correlation for both scenarios. This is due to the localscattering around the receiver set. The results are summarizedin Tables IV.

D. Path Loss

In this section, the path loss (PL) is presented, since it is arelevant parameter for characterizing the MIMO-mmW wire-less channel. The measured path loss is defined as the ratio ofthe transmitted power and a local average of the received power.The path loss for the undertaken gallery can be defined as

(8)

where is the expectation operator over all receiving antennaelements, transmitting antenna elements, frequencies, and snap-shots.Usually, path loss is modeled as a function of the distance

between transmitter and receiver as [18]

(9)

where is the mean path loss at the reference distance, which is often chosen as 1 m for indoor environment, 10( is the mean path loss referenced to , and is a


TABLE IVTRANSMIT AND RECEIVE CORRELATION FOR DIFFERENT SCENARIOS

Fig. 14. Average path loss.

zero mean Gaussian random variable expressed in dB. Param-eter is the standard deviation of variations . The meanpath loss at and the path loss exponent were determinedthrough the least square regression analysis and they are highlydependent onmeasurement environment and scenario. Based onMIMO 60 GHz measurements, path loss as a function of dis-tance for LoS and human effect scenarios are shown in Fig. 14.Path loss is found to be highly correlated with distance.In one hand, it can be shown that when using high gain an-

tennas at 60 GHz, the signals propagate better than free spaceunder LoS scenario. The value of the path loss exponent

has been determined to be less than the free space valueof 2.In the other hand, it is seen that the receiver power is at-

tenuated by approximately 14 dB due to the miners’ presence.Higher value of , is obtained. Indeed, the perturba-tion caused by the human body is very significant and affectsstrongly the transmitter-receiver link. The path loss exponentvalues that are obtained are quite similar to those found in theliterature at mmW [7], [17], [24]. Furthermore, is definedas the shadowing, which is the distance between the points ofthe local average power and their linear regression line. Dueto the random movement of the miners within the propagation

TABLE VPATH LOSS EXPONENT AND STANDARD DEVIATION

channel, a large signal power deviation from its mean is ob-served at some measured locations. This explains the highershadowing comparing to LoS scenario. Path loss exponentand are presented in Table V according to the measurementsscenario.

E. Capacity

Since the large spectral efficiencies associated with MIMOchannels are based on the premise that a rich scattering en-vironment provides independent transmission paths from eachtransmit antenna to each receive antenna [19]. Thus, MIMOchannel capacity depends heavily on the statistical propertiesand antenna element correlations of the channel.In fact, if the channel is completely unknown at the trans-

mitter, i.e., channel state information (CSI) is not available atthe transmitter, the mutual information capacity of a flat-fading

MIMO channel can be expressed by (10) given below,assuming transmitted power to be uniformly distributed amongthe transmitting antennas [25]

(10)where denotes the identity matrix of size , is the av-erage receive signal-to-noise ratio (SNR) and the upper scriptrepresents the hermitian conjugate of the normalized channel

matrix H. The H matrix is normalized such that at each instanceor each realization [26], [27]

(11)

where represents the Frobenius norm.However, this normalization removes any power variation of

the measurement path and thus the changes in the path loss withtime are not included. This normalization is used for scenariowhere the transmitted power compensates for the total receivedpower variation in order to keep the average SNR per receiverantenna fixed for each realization of the channel, irrespectiveof path loss. This method is useful to investigate the multipathrichness of the environment [26], [27]. Furthermore, for a fre-quency selective channel

(12)

where denotes the statistical mean over the channel band-width. The average channel capacity for SISO and 2 2MIMOsystem is calculated for the two scenarios, assuming a constantSNR of 10 dB, whatever the position of the receiver. The rela-tionship between the channel capacity C and based on(12) is shown in Fig. 15.


Fig. 15. Channel capacity for different scenarios.

TABLE VICHANNEL CAPACITY

Since the presence of miners in the propagation channel ex-hibits lower channel correlation, higher capacities are expected.It can be seen that the mean capacity of the MIMO channel in-cluding miners’ presence is 1.16 bit/s/Hz higher than when LoSoccurs. Also, it is interesting to note that better MIMO channelcapacities are obtained for the LoS scenario atand 2 m. This can be explained by the fact that at small dis-tances, the effect of multipath is negligible for the human ef-fect scenario and the channel is found to be more correlatedin these locations comparing to the LoS scenario. However, atlarge distances, stronger multipath are received leading to thelarger channel capacity. Hence, the achievable MIMO capacitygain depends on the multipath characteristics of the propagationchannel (the number of useful propagation paths). Furthermore,the improvement in channel capacity offered by MIMO overthe same channel with SISO link is noticeable. The throughputgain, i.e. the amount of improvement offered by MIMO overSISO under LoS, is less than 2 due to some degree of correlationbetween the subchannels. Moreover it is found that SISO linksunder LoS exhibit higher capacities than the one under humaneffect scenario by about 0.46 bit/s/Hz. Channel capacities fordifferent scenarios are summarized in Table VI.

F. Discussion

Same scenarios and analysis are already performed at 2.45GHz and for UWB frequency bands. It is important to find outthe suitable frequency band for MIMO systems operating in un-derground mine environment. This will ultimately help in de-signing appropriate wireless communication devices for suchconfined environment.One of the main advantages of using the 60 GHz radio over

lower frequency bands is related to compact component sizes atmmW frequencies. Moreover, based on the findings of the ex-perimental results, it is shown that low frequency signals werecapable of covering long distance communication in an under-ground mine gallery, whereas mmW signals were generally lim-ited to short range communication (up to 10 m), due to sig-nificant higher attenuation. In addition, signals at 2.4 GHz andUWB are less sensitive to human shadowing and machinery ob-struction than at the mmW band.Furthermore, it is observed that RMS delay spreads for

60 GHz propagation was lower than 2.4 GHz and UWB prop-agation. This might be due to the first Fresnel zone at 60 GHzwhich is significantly smaller than at lower frequency. There-fore, fewer objects in the environment are capable of perturbingthe LoS signal. In addition to the large bandwidth for the mmWtechnology, the low RMS leads to the possibility of providingvery high data rate communications. 60-GHz systems wouldbe ideal for a femtocellular network [28].

IV. CONCLUSION

This paper addresses the statistical characterization of un-derground mine radio channels in the 60 GHz frequency band.In order to overwhelm the propagation losses at the mmW,a 2 2 planar microstrip antenna array with high gain isfabricated and used for the measurement campaign. In addi-tion, the effect of miners’ presence in the close vicinity of thepropagation channel is studied. It is demonstrated that humanbody shadowing greatly affects the received signal strengtheven when considering a multipath environment where sev-eral rays contribute to the received power. Moreover, pathloss measurements are reported in LoS and in the presence ofminers. It is shown that signals propagate better than in a freespace under LoS. However, the path loss exponent is about2.4 for the human effect case. Furthermore, it is found thatwhen miners move between the two terminals the K factor de-creases and the channel become less correlated. Therefore, thechannel capacity is improved accordingly if assuming a fixedSNR level of 10 dB. In addition, the improvement of MIMOchannel capacity over SISO is illustrated. The throughput gainis almost doubled. Our analysis thus shows that MIMO is aneffective strategy for boosting the performance of mmW com-munication systems. The presented results are useful for thedesign of MIMO-mmW radio system dedicated for the miningindustry.

REFERENCES

[1] I. B. Mabrouk, L. Talbi, M. Nedil, and K. Hettak, “Effect of miningmachinery on MIMO-UWB radiowave propagation within an under-ground gallery,” IEEE Trans. Antennas Propag., to be published.


[2] I. B. Mabrouk, L. Talbi, and M. Nedil, “Performance evaluation of aMIMO system in underground mine gallery,” IEEE Antennas WirelessPropag. Lett., vol. 11, pp. 830–833, 2012.

[3] A. Arvanitis, G. Anagnostou, N. Moraitis, and P. Constantinou, “Ca-pacity study of a multiple element antenna configuration in an indoorwireless channel at 60 GHz,” in Proc. IEEE 65th Vehicular Tech. Conf.,Apr. 2007, pp. 609–613.

[4] A. P. Garcia Ariza, R. Müller, F. Wollenschläger, A. Schulz, M.Elkhouly, Y. Sun, S. Glisic, U. Trautwein, R. Stephan, J. Müller, R. S.Thomä, and M. A. Hein, “60 GHz ultrawideband polarimetric MIMOsensing for wireless multi-gigabit and radar,” IEEE Trans. AntennasPropag., to be published.

[5] K.-C. Huang and Z.Wang, Millimeter Wave Communication System.Hoboken, NJ, USA: Wiley, Apr. 2011.

[6] S. K. Yong, P. Xia, and A. V. Garcia, “60 GHz Technology for GbpsWLAN and WPAN,” in From Theory to Practice, 1st ed. Hoboken,NJ, USA: Wiley, 2011.

[7] S. Geng, J. Kivinen, X. Zhao, and P. Vainikainen, “Millimeter-wavepropagation channel characterization for short-range wireless commu-nications,” IEEE Trans. Veh. Technol., vol. 58, no. 1, pp. 3–13, Jan.2009.

[8] N. Moraitis and P. Constantinou, “Measurements and characterizationof wideband indoor radio channel at 60 GHz,” IEEE Trans. WirelessCommun., vol. 5, no. 4, pp. 880–889, Apr. 2006.

[9] H. Yang,M.H. A. J. Herben, and P. F.M. Smulders, “Impact of antennapattern and reflect environment on 60 GHz indoor radio channel char-acteristics,” IEEEAntennasWireless Propag. Lett., vol. 4, pp. 300–303,2005.

[10] J. M. Molina-Garcia-Pardo, J.-V. Rodríguez, and L. Juan-Llacer,“Polarized indoor MIMO channel measurements at 2.45 GHz,” IEEETrans. Antennas Propag., vol. 56, no. 12, pp. 3818–3828, Dec. 2008.

[11] S. Ranvier, J. Kivinen, and P. Vainikainen, “Millimeter-wave MIMOradio channel sounder.,” IEEE Trans. Instrum. Meas., vol. 56, no. 3,pp. 1018–1024, Mar. 2007.

[12] H.-H. Lee and Y.-C. Ko, “Low complexity codebook-based beam-forming forMIMO-OFDM systems in millimeter-waveWPAN,” IEEETrans. Wireless Commun., vol. 10, no. 11, pp. 3607–3612, Nov. 2011.

[13] E. Torkildson, U. Madhow, and M. Rodwell, “Indoor millimeter waveMIMO: Feasibility and performance,” IEEE Trans. Wireless Commun.,vol. 10, no. 12, pp. 4150–4160, Dec. 2011.

[14] S. Collonge, G. Zaharia, and G. El Zein, “Influence of the humanactivity on wide-band characteristics of the 60 GHz indoor radiochannel,” IEEE Trans.Wireless Commun., vol. 3, no. 3, pp. 2396–2406,Nov. 2004.

[15] A. E. I. Lamminen, J. Saily, andA. R. Vimpari, “60GHz patch antennasand arrays on LTCC with embedded-cavity substrates,” IEEE Trans.Antennas Propag., vol. 56, no. 9, pp. 2865–2874, Sep. 2008.

[16] J. W. Wallace and M. A. Jensen, “Mutual coupling in MIMO wirelesssystems: A rigorous network theory analysis,” IEEE Trans. WirelessCommun., vol. 3, no. 4, pp. 1317–1325, Jul. 2004.

[17] N. Moraitis and P. Constantinou, “Indoor channel measurements andcharacterization at 60 GHz for wireless local area network applica-tions,” IEEE Trans. Antennas Propag., vol. 52, no. 12, pp. 3180–3189,2004.

[18] T. S. Rappaport, Wireless Communications, Principles and Practice,2nd ed. Upper Saddle River, NJ, USA: Prentice Hall, 2002.

[19] R.-T. Juang, H.-P. Lin, and M.-J. Tseng, “Adaptive transmissionscheme based on Rician channel k-factor for EGPRS systems,” pre-sented at the Autonomic and Autonomous Systems Conf. Tahiti, Oct.2005.

[20] V. Tarokh, N. Seshadri, and A. R. Calderbank, “Space-time codes forhigh data rate wireless communication: Performance criterion and codeconstruction,” IEEE Trans. Inf. Theory, vol. 44, no. 3, pp. 744–765,Mar. 1998.

[21] I. Khan and P. S. Hall, “Experimental evaluation of MIMO capacityand correlation for narrowband body-centric wireless channels,” IEEETrans. Antennas Propag., vol. 58, no. 1, pp. 195–202, Jan. 2010.

[22] C. Tepedelenlioglu, A. Abdi, and G. Giannakis, “The Ricean kfactor: Estimation and performance analysis,” IEEE Trans. WirelessCommun., vol. 2, no. 4, pp. 799–810, Apr. 2003.

[23] W. Q. Malik, “Spatial correlation in ultrawideband channels,” IEEETrans. Wireless Commun., vol. 7, no. 2, pp. 604–610, Feb. 2008.

[24] P. F. Smulders, “Statistical characterization of 60-GHz indoor radiochannels,” IEEE Trans. Antennas Propag., vol. 57, no. 10, pp.2820–2829, Oct. 2009.

[25] A. J. Paulraj, R. Nabar, and D. Gore, Introduction to Space-TimeWireless Communications. Cambridge, U.K.: Cambridge Univ.Press, 2003.

[26] T. Svantesson and J. Wallace, “On signal strength and multipath rich-ness in multi-input multi-output systems,” in Proc. IEEE Int. Conf.Commun., May 2003, vol. 4, pp. 2683–2687.

[27] H. Carrasco, R. Feick, and H. Hristov, “Experimental evaluation ofindoorMIMO channel capacity for compact arrays of planar inverted-Fantennas,”Microw. Opt. Technol. Lett., vol. 49, no. 7, pp. 1754–1756,Jul. 2007.

[28] C. R. Anderson and T. S. Rappaport, “In-building wideband parti-tion loss measurements at 2.5 and 60 GHz,” IEEE Trans. WirelessCommun,, vol. 3, no. 3, pp. 922–928, Mar. 2004.

Ismail Ben Mabrouk (S’11–M’12) was born inHammam Lif, Tunisia, in 1984. He received theB.A.Sc. and M.A.Sc. degrees from the University ofLille, Lille, France, in 2006 and 2007, respectively,and the Ph.D. degree from the University of Quebecin Outaouais, Ottawa region, Canada, in 2012, all inelectrical engineering.Currently, he is a postdoctoral fellow at the Infor-

mation Engineering, Computer Science, and Mathe-matics Department, University of L’aquila, Italy. Hisresearch activities have been centered on propagation

studies for multiple-input-multiple-output systems, wireless body area network(WBAN), measurement campaigns in underground mine environments, and an-tenna design at millimeterwave.

Julien Hautcoeur was born in Nantes, France, in1983. He received the degree in Electronics Sys-tems Engineering and Industrial Informatics, EcolePolytechnique, University of Nantes, followed bythe M.S. degree in Radio Communication Systemsand Electronics in 2007 and the Ph.D. degree insignal processing and telecommunications from theInstitute of Electronics and Telecommunications ofRennes 1, Rennes, France, in 2011.Since 2011, he has been doing Postdoctoral

training at the University of Quebec in Outaouais.His current research field is optically transparent antenna systems for telecom-munications (transparent and conductive materials, thin films, meshed antenna).

Larbi Talbi (S’95–M’97–SM’05) received the M.S.and Ph.D. degrees in electrical engineering fromLaval University, Quebec City, QC, Canada, in 1989and 1994, respectively.He completed a Post-Doctoral Fellowship with

the INRS-Telecommunications, Montreal, QC,within the Personal Communications ResearchGroup (1994–1995), where he led projects supportedby Bell-Canada. From 1995 to 1998, he was anAssistant Professor at the Electronics EngineeringDepartment of Riyadh College of Technology in

Saudi Arabia. During 1998–1999, he was an Invited Professor at the Electricaland Computer Engineering Department, University Laval, QC. Since 1999,he has been a Professor with the Department of Computer Science andEngineering of University of Quebec, Outaouais/Ottawa region, Canada, wherehe is the Ph.D. program Chair in Sciences and Information Technologies.His research activities include experimental characterization and modeling ofUHF/EHF indoor radio propagation channels and design of antennas and mi-crowave circuits for wireless communication systems. Currently, he is activelyinvolved in major projects related to the deployment of wireless technologies inunderground mines, mainly, experimental characterization of the undergroundmine channels using MIMO antennas at 60 GHz, design of microwave andRF components using SIW technique and metamaterials, antenna array forwireless applications. He has authored or coauthored more than 170 journaland conference papers. He is a member of the Ordre des ingénieurs du Québec.


MouradNedil (M’08–SM’12) received the Dipl.Ing.degree from the University of Algiers (USTHB),Algiers, Algeria, in 1996, the D.E.A. (M.S.) degreefrom the University of Marne la Vallée, Marne laVallée, France, in 2000, and the Ph.D. degree inelectrical engineering from the Institut National dela Recherche Scientifique (INRS-EMT), Universitéde Québec, Montréal. QC, Canada, in April 2006.He completed a Postdoctoral Fellowship at

INRS-EMT, within the RF Communications Sys-tems Group, from 2006 to 2008. In 2008, he joined

the Underground Communications Laboratory (LRTCS), University of Quebecat abitibi-Témiscamingue, Val d’OR, QC, Canada, as an Assistant Professor.His research interests include antennas, microelectromechanical systems(MIMO) radio-wave propagation, and microwave devices.

Khelifa Hettak (M’06–SM’06) received the Dipl.-Ing. in telecommunications from the University ofAlgiers, Algeria, in 1990, and his M.A.Sc. and Ph.D.,in signal processing and telecommunications, fromUniversity of Rennes 1, France, in 1992 and 1996,respectively.In January 1997, he has been with the Personal

Communications Staff of INRS-Télécommunica-tions, where is was involved. He joined the electricalengineering department of Laval University inOctober 1998 as an associate researcher, where he

was involved in RF aspects of Smart antennas. Since August 1999, he has beenwith Terrestrial Wireless Systems Branch at Communications Research Centre(CRC), Ottawa, Canada, as Research Scientist. He was involved in developingMMICs at 60 GHz, low temperature cofired ceramic (LTCC) packaging, RFMEMS switches, and GaN robust Tx/Rx modules. He is actively involved inmicrowave/millimeter-wave systems and related front-end analog electroniccircuits, phased arrays, and satellite communication systems. He also is activein planar antenna design include wide scan-angle antennas at 60 GHz forwireless applications. He recently started an effort in CMOS/SiGe RFIC designfor the 60-GHz region, with particular emphasis on developing miniaturephase shifters using CMOS/SiGe technology for phased-array applications,oscillators, and switches for millimeter-wave communication systems.

Feasibility of a Millimeter-Wave MIMO System for Short-Range Wireless Communications in an...

Documents

Transcript of Feasibility of a Millimeter-Wave MIMO System for Short-Range Wireless Communications in an...