28 GHz Indoor and Outdoor Propagation Analysis at a Regional Airport
11
28 GHz Indoor and Outdoor Propagation Analysisat a Regional Airport
Kairui Du, Ozgur Ozdemir, Fatih Erden, and Ismail GuvencDepartment of Electrical and Computer Engineering, North Carolina State University, Raleigh, NCEmail: { kdu, oozdemir, ferden, iguvenc } @ncsu.edu Abstract —In the upcoming 5G communication, the millimeterwave (mmWave) technology will play an important role dueto its large bandwidth and high data rate. However, mmWavefrequencies have higher free-space path loss (FSPL) in line-of-sight (LOS) propagation compared to the currently used sub-6 GHz frequencies. What is more, in non-line-of-sight (NLOS)propagation, the attenuation of mmWave is larger comparedto the lower frequencies, which can seriously degrade theperformance. It is therefore necessary to investigate mmWavepropagation characteristics for a given deployment scenario tounderstand coverage and rate performance for that environment.In this paper, we focus on 28 GHz wideband mmWave signalpropagation characteristics at Johnston Regional Airport (JNX),a local airport near Raleigh, NC. To collect data, we use anNI PXI based channel sounder at 28 GHz for indoor, outdoor,and indoor-to-outdoor scenarios. Results on LOS propagation,reflection, penetration, signal coverage, and multi-path compo-nents (MPCs) show a lower indoor FSPL, a richer scattering,and a better coverage compared to outdoor. We also observehigh indoor-to-outdoor propagation losses.
Index Terms —28 GHz, LOS, mmWave, multipath components,NLOS, propagation channel measurement, signal coverage.
I. I
NTRODUCTION
With the development of modern telecommunication, theuse of wireless devices and applications that require higherdata rates have increased tremendously in the recent decades.The sub-6 GHz frequency band is getting more congested bythe rapid growth of users and it can not sustain the supportof high data rates due to its limited channel bandwidth. Thedifficulty of supporting the demand of next generation com-munications at lower frequency bands motivated researchersto explore millimeter-wave (mmWave) bands, which offer asubstantially large amount of available bandwidth comparedto sub-6 GHz frequencies. Owing to the millimeter-levelwavelengths, large arrays of antennas can be used in smallsmart devices to achieve higher data throughput.At sub-6 GHz frequencies, the wavelength is significantlylarger compared to the mmWave bands which allows the sig-nals to penetrate through obstacles in our surroundings such aswalls, windows, doors, and foliage. On the contrary, the narrowwavelength of mmWave frequencies introduces propagationchallenges, such as high free-space path-loss (FSPL) and highattenuation while penetrating through different materials. It ishence necessary to fully study the propagation characteristicsof mmWave signals in different environments. Fortunately,with the opening of mmWave spectrum by FCC [1], therehas been a surge of channel measurements that report results
This work was supported by NASA under the Award ID NNX17AJ94A.Authors would also like to thank the Johnston Regional Airport for allowingto carry out the experiments. at common mmWave frequency bands at 28, 39, 60, 73, and91 GHz, which all help to analyze mmWave propagation.Reflection and penetration loss at 28 GHz in New Yorkurban environment [2] showed that the outdoor building ma-terials are better reflectors than indoor materials, and thatthe penetration loss at larger distances are affected by thesurrounding environment apart from distance and obstruc-tions. Another study [3] reported that both outdoor non-line-of sight (NLOS) and line-of-sight (LOS) environments hadrich multipath components at 28 GHz using steerable beamantennas. In [4], authors focused on penetration loss of severaltypical building materials in three popular mmWave bands(28, 73, 91 GHz). As expected, higher penetration loss wasobserved as the frequency increases. Plywood and clear glasssuffered higher attenuation (in dB/cm) compared to ceilingtile, drywall, and cinder blocks at all three frequency bands.Another study [5] focused on indoor reflection, penetration,scattering and path loss properties at both mmWave (28 GHzand 73 GHz) and sub-terahertz (140 GHz) frequencies. Theauthors found out that the reflection coefficient increaseslinearly as the incident angle increases. Reflection loss islower at higher frequencies at a given incident angle whilethe penetration loss increases with frequency. The authorsin [6] presented directional and omnidirectional path lossmodels, temporal and spatial channel models, and outageprobabilities based on more than 15,000 measured power delayprofiles (PDPs) at 28, 38, 60, and 73 GHz mmWave bandsusing wideband sliding correlator channel sounder and hornantennas.In our recent 28 GHz channel measurements [7] at James B.Hunt Jr. Library of NC State University, path loss for the LOSscenarios was obtained to be very close to the free space pathloss model. Models for power angular-delay profile (PADP)and large-scale path loss for both LOS and NLOS scenarioswere developed based on the measurements over distancesranging from 10 m to 50 m. We also explored the use ofpassive reflectors to improve NLOS signal coverage at 28 GHzfor both indoor and outdoor scenarios [8], [9]. The squaremetallic sheet reflectors were proved to be a simple, effective,and affordable way of enhancing mmWave coverage. Withthe proper shape and dimension of the reflector, the receivedpower could be improved significantly, and it even approachesthe Friis free space LOS received power at the same traveldistance as the NLOS signal.From these previous studies, mmWave features a high pathloss and a high material attenuation. A survey [10] statedthe demand of developing methodologies that support highlydirectional mmWave links over longer distances at airportsdue to the high path loss. Although the terrestrial application a r X i v : . [ ee ss . SP ] J a n PPS10 MHz PPS 10 MHzTX Chassis RX ChassisIF IF
FPGA
DAC
BB->IF IF->RF RF->IF
FPGA FPGA
ADC
IF->BB
TX Rotating Gimbal RX Rotating Gimbal
Fig. 1: Channel sounder block diagram for the NI PXI platform used in the experiments [14]. of mmWave systems is advancing at a rapid pace, the useof mmWave communication systems in aviation systems orairports is still in its infancy, partially due to the lack ofcharacterization of mmWave wireless channels for the aviationfield and the airport environment, and hence measurementsin different airports are needed. A study [11] conductedchannel measurements at Boise Airport in both LOS andNLOS scenarios and presented a large scale fading channelmodel of 60 GHz. Another study [12] focused on LOSmmWave propagation measurement and modeling also tookplace in Boise Airport. To the best of authors’ knowledge,only one study [13] exists on airport mmWave propagationcharacterization at 28 GHz in an indoor scenario. This workfocused on the specular propagation paths, specular and diffusepower contributions, polarization, and the delay and angu-lar spreads at Helsinki Airport, Finland. Further results andanalysis are needed at different airports (covering commercialaviation, private aviation, etc.) to develop better understandingof mmWave propagation in airport environments.In this work, we performed mmWave channel soundingmeasurement for both indoor, outdoor and indoor-to-outdoorscenarios at 28 GHz in Johnston Regional Airport close toRaleigh, NC, based on NI’s PXI-based channel soundingplatform [15] (see Fig. 1), using 17 dBi horn antennas onboth gimbal-assisted transmitter (Tx) and receiver (Rx) toanalyze propagation characteristics. A higher outdoor FSPLwas observed comparing to indoor FSPL. Indoor propagationalso showed more multipath components (MPCs) and bettersignal coverage than the outdoor propagation in the airport en-vironment, while indoor-to-outdoor propagation was difficultat 28 GHz. II. P
ROPAGATION M ODELING
A. LOS Propagation
Path loss (PL) is the reduction in power density (attenuation)of an electromagnetic wave as it propagates through space. It isone of the key factors that impact the received signal strengthand given by
PL (dB) = P t − P r , where P t is the transmitpower, and P r is the received power. We consider the FSPLmodel, which yields: FSPL (dB) = 20 log( d )+20 log( f )+20 log (cid:18) πc (cid:19) − G T − G R , (1)where G T and G R are Tx and Rx gain in the LOS direction, d is the distance between Tx and Rx, and f is the centerfrequency of the signal. B. Reflection
For reflection measurement, we consider the reflection co-efficient Γ , which is the ratio of the amplitude of reflectedsignal to the amplitude of incident signal. Measured reflectioncoefficient is derived as follows [16]: | Γ | = d total d LOS (cid:115) P r reflected P r LOS , (2)where d LOS is the distance between Tx and Rx for LOSpropagation, and d total is the total sum of travel distance ofthe incident and reflected ray. P r reflected is the received powerof the reflected ray, and P r LOS is the LOS received power.Theoretical reflection coefficient considers two conditions,which are perpendicular reflection coefficient Γ perpendicular when E-field is perpendicular to the plane of incidence, andparallel reflection coefficient Γ parallel when E-field is parallelto the plane of incidence: | Γ perpendicular | = (cid:12)(cid:12)(cid:12)(cid:12) Z cos θ i − Z cos θ t Z cos θ i + Z cos θ t (cid:12)(cid:12)(cid:12)(cid:12) , (3) | Γ parallel | = (cid:12)(cid:12)(cid:12)(cid:12) Z cos θ t − Z cos θ i Z cos θ t + η cos θ i (cid:12)(cid:12)(cid:12)(cid:12) , (4)where Z and Z are the wave impedance of media 1 and 2, θ i is the incident angle, and θ t is the refracted angle. Assumingthat the media are non-magnetic (i.e., relative permeability µ r = 1), the wave impedance is determined solely by therefractive index η . We can further substitute wave impedanceto refractive index using (5) and get (6) and (7): Z i = Z η i , (5) | Γ perpendicular | = (cid:12)(cid:12)(cid:12)(cid:12) η cos θ i − η cos θ t η cos θ i + η cos θ t (cid:12)(cid:12)(cid:12)(cid:12) , (6) | Γ parallel | = (cid:12)(cid:12)(cid:12)(cid:12) η cos θ t − η cos θ i η cos θ t + η cos θ i (cid:12)(cid:12)(cid:12)(cid:12) , (7)where η and η are the refractive indices for media 1 and 2.Using Snell’s law we can get the relationship of angle versusrefractive index: sin θ i sin θ t = η η . (8)We can also obtain the relationship between refractive indexand frequency dependent relative permittivity (cid:15) r ( f ) using: η = (cid:112) (cid:15) r ( f ) µ r = (cid:112) (cid:15) r ( f ) . (9)Since η is the refraction index of air, which is equal to 1, we Fig. 2: Indoor measurement at Johnston Regional Airport. can substitute (8) and (9) into (6) and (7) to further simplifyit as follows [17]: | Γ perpendicular | = (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)(cid:12) cos θ i − (cid:113) (cid:15) r ( f ) − sin θ i cos θ i + (cid:113) (cid:15) r ( f ) − sin θ i (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)(cid:12) , (10) | Γ parallel | = (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)(cid:12) (cid:15) r ( f ) cos θ i − (cid:113) (cid:15) r ( f ) − sin θ i (cid:15) r ( f ) cos θ i + (cid:113) (cid:15) r ( f ) − sin θ i (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)(cid:12) . (11)Note that the equations above contain only two parameters:incident angle and relative permittivity (frequency and materialdependent). Due to our measurement setup, we only considerparallel condition for our analysis and take the value of (cid:15) r =3 based on our previous measurement of clear glass relativepermittivity at 28 GHz in an indoor environment at NC StateUniversity. C. Penetration
For penetration loss, we do not have specific model as ourtheoretical formula of penetration loss. Therefore, we willtransmit mmWave signal through wall and glass door andcompare its loss with the LOS condition. We can calculatethe measured penetration loss by:
Penetration Loss (dB) = P r LOS − P r penetration . (12)III. M EASUREMENT S ETUP
Our channel sounder hardware is based on NI’s mmWavesystem at 28 GHz, as shown in Fig. 1 [14]. It consisted of NIPXIe-1085 Tx and Rx chassis, 28 GHz Tx and Rx mmWaveradio heads (fixed on FLIR PTU-D48E gimbals) from NI, andFS725 Rubidium (Rb) clocks. A 10 MHz pulse per second(PPS) signals generated by a single clock [18] was connectedto PXIe 6674T modules at both Tx and Rx. The 10 MHzsignal was used to generate the required local oscillator (LO)signals, and the transmission at TX side and reception at theRX side were triggered by the same PPS signal.The sounder code was based on LabVIEW, and periodicallytransmitted a Zadoff-Chu (ZC) sequence of length 2048 to
Fig. 3: Outdoor measurement at Johnston Regional Airport. sound the channel. It was then filtered by the root-raised-cosine (RRC) filter, and the generated samples were uploadedto PXIe-7902 FPGA. These samples were sent to PXIe-3610digital-to-analog (DAC) converter with a sampling rate of f s = 3 . GS/s. The PXIe-3620 module up-converted thebase-band signal to IF, and the 28 GHz mmWave radio headup-converted the IF signal to RF with 2 GHz bandwidth and10 dBm transmit power. At the Rx, 28 GHz mmWave radiohead down-converted the RF signal to IF, which was thendown-converted to base-band at the PXIe-3620. The PXIe-3630 analog-to-digital converter (ADC) sampled the base-bandanalog signal with the sampling rate of f s = 3 . GS/s.The channel sounder provides /f s = 0 . ns delayresolution in the time domain. Therefore, any multipath com-ponent with a delay higher than . ns can be resolved,which represent a path-length difference of . m. Thecorrelation and averaging operations were performed in PXIe-7976R FPGA operation, and the complex CIR samples and thepower-delay-profile (PDP) were sent to the PXIe-8880 hostPC for further processing and saving to local disk. Calibrationand equalization was then performed to eliminate the channeldistortion due to the non-idealities of the hardware themselves.After that, directional horn antennas were connected to themmWave radio heads at the Tx and the Rx sides with 17 dBigains, and 26 degree and 24 degree half power beam-widthsin the elevation and azimuth planes, respectively. For all themeasurements, both Tx and Rx scanned an azimuth degree of − . to − . with a resolution of degree at eachTx-Rx position. A. Indoor Measurement Setup
As shown in Fig. 2, for the indoor measurements, Tx andRx were placed inside the terminal hall at the same heightof 1.5 m and were aligned to each other. Tx was fixed atone position, and the received PDP of Rx was measured atseparation of 5, 10, and 15 m from Tx.
B. Outdoor Measurement Setup
In the outdoor case shown in Fig. 3, the Tx and the Rx wereplaced outside the airport terminal. The Tx was fixed, and theRx was aligned to the Tx with a separation of 5.3, 9.6, and13.9 m from the Tx, as shown in Fig. 4.
Tx Rx1
Measure at different Tx azimuth angles-167.98 degree to 167.98 degree3 different Tx-Rx separation, d=5.3m, 9.6m, 13.9m correspondingMeasure at different Rx azimuth angles-167.98 degree to 167.98 degree
Rx2 Rx3
Concrete wall, thickness=0.6 m Thin glass wallTx 0 degree Rx 0 degree
Fig. 4: Outdoor measurement scenario (looking from above). TxRx1 =6.3m, d . m Rx2 Rx3
Concrete wall, thickness=0.6 m Measured at different Tx azimuth angles:-167.98 degree to 167.98 degree . m Thin glass wallMeasure at different Rx azimuth angles:-167.98 degree to 167.98 degreeTx 0 degreeRx 0 degree
Fig. 5: Indoor-to-outdoor measurement scenario (looking from above).
C. Indoor-to-Outdoor Measurement Setup
In the indoor-to-outdoor measurement scenario, Tx wasfixed inside the terminal hall, and Rx was placed outside at aseparation of 13.4 m. The Rx would later move along the wallat distance of 6.3 m and 10.7 m away from its first position,which is shown in Fig. 5. The direct LOS propagation wasblocked by the terminal wall and clear glass door, of whichbetween Tx and Rx1 was a concrete wall, between Tx andRx2 and Tx and Rx3 was a glass wall.IV. R
ESULTS AND ANALYSIS
A. LOS Propagation
We measured the LOS path loss in both indoor and outdoorscenarios. Each measured point was taken when Tx and Rxwere aligned to each other at a certain Tx-Rx separation. Asshown in Fig. 6, the theoretical FSPL calculated from (1) isfurther compared with the measured path-loss. From Fig. 6,indoor path loss matches the theoretical FSPL. However out-door path loss is approximately 2-3 dB higher than indoor pathloss. This might be a consequence of a more complicated andunstable outdoor environment (wind, temperature, humidity,environment noise, etc.) compared to indoor.
Distance (m) P a t h l o ss ( d B ) Theorectical modelIndoor measurementsOutdoor measurements
Fig. 6: FSPL at 28 GHz. Blue line is the theoretical FSPL, and red asterisk isthe measured indoor FSPL at 5, 10, and 15 m. Purple circle is the measuredoutdoor FSPL at 5.3, 9.6, and 13.9 m.
B. Reflection
The first-order reflection from the terminal glass wall andglass door at 3 different distances (which represent 3 different
Incident angle (degree) R e f l e c t i on c oe ff i c i en t m agn i t ude Fig. 7: Reflection coefficient of theoretical derivation and measurement result.Blue line indicates the theoretical value, and the dashed line with red asteriskis the measured coefficient at incident angle of . , . , and . degree. For theoretical plot, the antenna radiation pattern is neglected. incident angles) from the outdoor measurement results weretaken for further analysis. They are also compared with thetheoretical reflection coefficient, as shown in Fig. 7, of parallelcondition (vertical polarized wave for our horn antennas)calculated via (11) at a relative permittivity of 3 for clear glassat 28 GHz.As seen in Fig. 7, the measured results do not align withthe theoretical calculation under the given assumptions. Wehave neglected the specific radiation pattern of the Tx/Rxantennas and the associated antenna gains for the Tx/Rx signalthat correspond to the reflection point, and further work isneeded to accurately model this artifact. The type of glassfor the wall and door are not available to us and the rangingof relative permitivity for different types of glass could beranging from 2-10 [19]. More thorough study needs to beconducted for different types of glass at different mmWavebands. Notably, the measurement environment (the double-layer glass door) might also introduce some unresolvableMPCs due to scattering and higher-order reflections whichmay lead to a lower observed reflection loss. C. Penetration
A comparison of concrete wall and glass wall penetrationloss with the reference measurement of concrete blocks andglass board [4] are shown in Table I. Path loss withoutblockage is the calculated FSPL of the LOS MPC (at thesame travel distance as the blocked case) from (1). Path losswith blockage is the measured indoor-to-outdoor path lossof the blocked LOS ray. Penetration loss is calculated from(12) based on the difference of path loss with and withoutblockage. The attenuation factor in dB/cm is the averagedpenetration loss over thickness. The results show that the largethickness of concrete wall and the high attenuation factorof glass led to high penetration loss for both walls. Theconcrete attenuation factor matched the reference result. Theslight larger attenuation is due to the outdoor propagationof the penetrated rays: since the reference measurement wastaken indoor, a higher attenuation compared to the referenceresult was expected. For the attenuation result of glass in thismeasurement, the incident ray is not perpendicular to the glasswall and the effective thickness should be higher than the
TABLE IP
ENETRATION L OSS M EASUREMENT AT
28 GH Z . Parameter Concrete Glass
Path loss without blockage (dB) 50.90 52.69Path loss with blockage (dB) 119.76 97.00Penetration loss (dB) 68.86 47.31Thickness (cm) 60 10Measured attenuation factor (dB/cm) 1.15 4.73Reference attenuation factor (dB/cm) [4] 1.06 4.39 measured thickness, which makes it higher than the referenceresult.
D. Signal Coverage and MPCs
The signal coverage and MPC distribution results for indoormeasurement are shown in Fig. 8. Each colored grid representsthe received power or the number of MPCs at a givencombination of Tx and Rx azimuth angle. As the Tx and theRx separation got larger, the number of MPCs increases and itsdistribution varies a lot. This is because in the indoor scenario,as the distance increases, more reflective objects (e.g. chair,desk, etc.) in the TX and the RX surroundings are introducedto the channel, which results in an increase in the numberof MPCs. Also, there is a dominant LOS propagation (for thestrongest MPC) with azimuth angle of arrival (AoA) and angleof departure (AoD) both at 0 degree, together with severalstrong NLOS MPCs whose received power get close to theLOS MPC’s received power.For the outdoor measurement, the number of MPCs in Fig. 9is more stationary compared to indoor environment whenthe Tx and the Rx separation increases because only certainobjects (e.g. wall, ground) could generate MPCs. The receivedpower in the LOS MPC with the azimuth AoA and AoDof 0 degree dominates all the other MPCs. Fewer reflectionsare observed compared to indoor scenario but some reflectedpower captured by the receiver is still close to the receivedpower of the LOS MPC.For indoor to outdoor measurement result shown in Fig. 10,the number of MPCs together with the received power overallhave a sharp decrease when compared to indoor propagationand outdoor propagation due to higher penetration loss. Onlysmall portion of the transmitted power is observed in theblocked LOS direction (around the color grid with the highestpower). The peak received power suffers more than 90 dBpath loss and is approximately 50 dB lower than the peak re-ceived power in indoor and outdoor LOS propagation withoutblockage. Propagation from indoor to outdoor is quite difficultand requires both AoA and AoD aligned to the blocked LOSdirection for the received signal to be detectable.V. C
ONCLUSION
In this work, we conducted measurements at JohnstonRegional Airport to analyze the propagation of 28 GHzmmWave signals. We compared the measurement results inindoor, outdoor, and indoor-to-outdoor scenarios with the thetheoretical propagation characteristics. The study showed thatmmWave outdoor propagation had a higher FSPL comparedto indoor scenario. Indoor environment had rich scatteringsand a wider signal coverage while received power in the out-door airport environment was only dominated by a few rays.Moreover, mmWave had high FSPL and penetration loss, and -150 -100 -50 0 50 100 150
Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) -110-100-90-80-70-60-50-40-30 R e c e i v ed po w e r ( d B m ) (a) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) -110-100-90-80-70-60-50-40-30 R e c e i v ed po w e r ( d B m ) (b) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) -110-100-90-80-70-60-50-40-30 R e c e i v ed po w e r ( d B m ) (c) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) N u m be r o f M P C s (d) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) N u m be r o f M P C s (e) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) N u m be r o f M P C s (f)Fig. 8: Indoor received power at different Tx and Rx azimuth angle for a Tx-Rx separation of (a) 5 m, (b) 10 m, (c) 15 m. Number of MPCs at different Txand Rx azimuth angle for a Tx-Rx separation of (d) 5 m, (e) 10 m, (f) 15 m. The angular resolution at the Tx and the Rx is degrees. hence it would be challenging to accomplish indoor-to-outdoorcommunication: the indoor-to-outdoor propagation thereforemay need to be highly directional to recover the penetrationloss through directionality gain. For both indoor and outdoorpropagation, there were still a considerable number of reflectedMPCs that had comparable received powers to the LOS MPC’sreceived power, which may allow a feasible way to achievemmWave NLOS communications via the reflective objects inthe channel environment.R Proc. IEEE Int. Conf.Commun. (ICC) , Budapest, Hungary, June 2013, pp. 5163–5167.[3] Y. Azar, G. N. Wong, K. Wang, R. Mayzus, J. K. Schulz, H. Zhao,F. Gutierrez, D. Hwang, and T. S. Rappaport, “28 GHz propagationmeasurements for outdoor cellular communications using steerable beamantennas in New York city,” in
Proc. IEEE Int. Conf. Commun. (ICC) ,Budapest, Hungary, June 2013, pp. 5143–5147.[4] N. Hosseini, M. Khatun, C. Guo, K. Du, O. Ozdemir, D. W. Matolak,I. Guvenc, and H. Mehrpouyan, “Attenuation of several common build-ing materials in millimeter-wave frequency bands: 28, 73 and 91 GHz,” arXiv preprint arXiv:2004.12568 , 2020.[5] Y. Xing, O. Kanhere, S. Ju, and T. S. Rappaport, “Indoor wirelesschannel properties at millimeter wave and sub-terahertz frequencies,”in
Proc. IEEE Global Commun. Conf. (GLOBECOM) , Waikoloa, HI,Dec. 2019, pp. 1–6.[6] T. S. Rappaport, G. R. MacCartney, M. K. Samimi, and S. Sun, “Wide-band millimeter-wave propagation measurements and channel models forfuture wireless communication system design,”
IEEE Trans. Commun. ,vol. 63, no. 9, pp. 3029–3056, Sept. 2015.[7] F. Erden, O. Ozdemir, and I. Guvenc, “28 GHz mmWave channelmeasurements and modeling in a library environment,” in
Proc. IEEERadio Wireless Symp. (RWS) , San Antonio, TX, Jan. 2020, pp. 52–55. [8] W. Khawaja, O. Ozdemir, Y. Yapici, I. Guvenc, and Y. Kakishima,“Coverage enhancement for mmWave communications using passivereflectors,” in
Proc. Global Symp. Millimeter Waves (GSMM) , Boulder,CO, May 2018, pp. 1–6.[9] W. Khawaja, O. Ozdemir, F. Erden, I. Guvenc, M. Ezuma, and Y. Kak-ishima, “Effect of passive reflectors for enhancing coverage of 28 GHzmmWave systems in an outdoor setting,” in
Proc. IEEE Radio WirelessSymp. (RWS) , Orlando, FL, Jan. 2019, pp. 1–4.[10] M. Khatun, H. Mehrpouyan, D. Matolak, and I. Guvenc, “Millimeterwave systems for airports and short-range aviation communications: Asurvey of the current channel models at mmWave frequencies,” in
Proc.IEEE/AIAA Digital Avionics Syst. Conf. (DASC) , St. Petersburg, FL,Sept. 2017, pp. 1–8.[11] M. Khatun, H. Mehrpouyan, and D. Matolak, “60-GHz millimeter-wavepathloss measurements in Boise Airport,” in
Proc. IEEE Global Conf.Signal Inf. Process. (GlobalSIP) , Anaheim, CA, Nov. 2018, pp. 1276–1280.[12] M. Khatun, C. Guo, L. Moro, D. Matolak, and H. Mehrpouyan,“Millimeter-wave path loss at 73 GHz in indoor and outdoor airportenvironments,” in
Proc. IEEE Veh. Technol. Conf. (VTC) , Honolulu, HI,Sept. 2019, pp. 1–5.[13] J. Vehmas, J. Jarvelainen, S. L. H. Nguyen, R. Naderpour, andK. Haneda, “Millimeter-wave channel characterization at Helsinki Air-port in the 15, 28, and 60 GHz bands,” in
Proc. IEEE Veh. Technol.Conf. (VTC) , Montreal, QC, Sept. 2016, pp. 1–5.[14] F. Erden, O. Ozdemir, W. Khawaja, and I. Guvenc, “Correction ofchannel sounding clock drift and antenna rotation effects for mmWaveangular profile measurements,”
IEEE Open J. Ant. Propag.
IEEE Trans. Antennas Propag. ,vol. 44, no. 3, pp. 341–351, Mar. 1996.[17] J. Lee, J. Choi, J. Lee, and S. Kim, “Permittivity effect of buildingmaterials on 28 GHz mmWave channel using 3D ray tracing simulation,”in
Proc. IEEE Global Commun. Conf. (GLOBECOM) -150 -100 -50 0 50 100 150
Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) -110-100-90-80-70-60-50-40-30 R e c e i v ed po w e r ( d B m ) (a) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) -110-100-90-80-70-60-50-40-30 R e c e i v ed po w e r ( d B m ) (b) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) -110-100-90-80-70-60-50-40-30 R e c e i v ed po w e r ( d B m ) (c) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) N u m be r o f M P C s (d) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) N u m be r o f M P C s (e) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) N u m be r o f M P C s (f)Fig. 9: Outdoor received power at different Tx and Rx azimuth angle for a Tx-Rx separation of (a) 5.3 m, (b) 9.6 m, (c) 13.9 m. Number of MPCs at differentTx and Rx azimuth angle for a Tx-Rx separation of (d) 5.3 m, (e) 9.6 m, (f) 13.9 m. The angular resolution at the Tx and the Rx is degrees. -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) -110-100-90-80-70-60-50-40-30 R e c e i v ed po w e r ( d B m ) (a) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) -110-100-90-80-70-60-50-40-30 R e c e i v ed po w e r ( d B m ) (b) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) -110-100-90-80-70-60-50-40-30 R e c e i v ed po w e r ( d B m ) (c) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) N u m be r o f M P C s (d) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) N u m be r o f M P C s (e) -150 -100 -50 0 50 100 150 Tx azimuth angle (degree) -150-100-50050100150 R x a z i m u t h ang l e ( deg r ee ) N u m be r o f M P C s (f)Fig. 10: Indoor-to-outdoor received power at different Tx and Rx azimuth angle for (a) Rx position 1, (b) Rx position 2, (c) Rx position 3. Number of MPCsat different Tx and Rx azimuth angle for a Tx-Rx separation of (d) Rx position 1, (e) Rx position 2, (f) Rx position 3. The angular resolution at the Tx andthe Rx is20