research article measurement-based los/nlos channel

Research ArticleMeasurement-Based LoSNLoS Channel Modeling forHot-Spot Urban Scenarios in UMTS Networks

Jiajing Chen1 Xuefeng Yin1 Li Tian1 Nan Zhang1 Yongyu He1

Xiang Cheng2 Weiming Duan3 and Silvia Ruiz Boqueacute4

1 College of Electronics and Information Engineering Tongji University Shanghai China2 School of Electronics Engineering and Computer Science Peking University Beijing China3Huawei Technology Company Shanghai China4Department of Signal Theory and Communications Polytechnic University of Catalonia Castelldefels Spain

Correspondence should be addressed to Xuefeng Yin yinxuefengtongjieducn

Received 5 May 2014 Revised 14 July 2014 Accepted 29 July 2014 Published 17 August 2014

Academic Editor Jose F Paris

Copyright copy 2014 Jiajing Chen et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

A measurement campaign is introduced for modeling radio channels with either line-of-sight (LoS) or non-line-of-sight (NLoS)connection between user equipment (UE) and NodeB (NB) in an operating universal mobile telecommunications system A space-alternating generalized expectation-maximization (SAGE) algorithm is applied to estimate the delays and the complex attenuationsof multipath components from the obtained channel impulse responses Based on a novel LoS detection method of multipathparameter estimates channels are classified into LoS and NLoS categories Deterministic models which are named ldquochannel mapsrdquoand fading statisticalmodels have been constructed for LoS andNLoS respectively In addition statistics of new parameters such asthe distance between the NB and the UE in LoSNLoS scenarios the life-distance of LoS channel the LoS existence probability perlocation and per NB the power variation at LoS to NLoS transition and vice versa and the transition duration are extractedThesemodels are applicable for designing and performance evaluation of transmission techniques or systems used by distinguishing theLoS and NLoS channels

1 Introduction

Distinctive characteristics of signals coming from a line-of-sight (LoS) or non-line-of-sight (NLoS) path in a wirelessenvironment raised great concerns in mobile radio systemdesign The probability of whether the LoS path exists in thechannel or not depends on various factors for example ter-rain features building density and locations of users Statisti-cal models have been employed by the 3GPP spatial channelmodel (SCM) [1] and the WINNER II SCM-enhanced [2] todescribe the LoS existence probability Recently characteriza-tion of LoSNLoS channel becomes highly demanded in someapplications such as massive multiple-input-multiple-output(MIMO) mobile back-hauling user-specific beamform-ing and other adaptive techniques used in future het-erogeneous networks [3ndash6] Newly introduced multipathaccess schememdashnonorthogonal multiple access (NOMA)mdashfor future radio access also relies on the distinction of LoS

andNLoS channels [7ndash9] Additionally some ad-hoc wirelesscommunication scenarios and the implementation of somenew technologies also prefer the LoS channel for exam-ple dedicated short range communication (DSRC) systemfor device-to-device vehicle-to-vehicle and machine-to-machine communications [10ndash15] Furthermore the LoSexistence has been considered in localization applicationssuch as the distance estimation between the base station (BS)and the mobile station in [16] Modeling realistic charac-teristics of LoSNLoS channels based on measurements areimportant for designing and optimizing these techniques orsystems

A problem arising when characterizing the LoSNLoSchannels based onmeasurements is how to categorize a chan-nel as LoS orNLoS Detectionmethods addressed in the liter-ature can be split into two groups that is the range-measur-ement-based and channel-impulse-response- (CIR-) basedmethods The former method such as those introduced in

Hindawi Publishing CorporationInternational Journal of Antennas and PropagationVolume 2014 Article ID 454976 12 pageshttpdxdoiorg1011552014454976

2 International Journal of Antennas and Propagation

Dataacquisition

USRPGNUradio

GPS Diskstorage

Data processing

Filtering

Synchronization

Analysis and modeling

Scrb detection

CIR extraction

SAGE estimation

LoSNLoS detection

Models of fading

Models ofchannel parameters

Figure 1 A diagram of channel measurement data analysis and modeling procedure adopted for LoSNLoS channel characterization

[16ndash18] was adopted for LoS detection in the early stage Thelatter approach is mainly based on the shape or statisticalproperties of the CIRs for example the confidence matrix[19] the delay spread [19] the power delay profile (PDP) [20]and the narrowband 119870-factor [21 22] In our contributionthe LoS detection is performed based on the power differenceamong estimated multipath components which has beenproven to be effective through measurement campaigns

Some LoSNLoS propagation models have been reportedin the literature According to measurement environmentsthese models characterize channels in either indoor or out-door scenarios In indoor cases the distribution of the chan-nel gains for the LoSNLoS scenarios is presented [23 24]Coherence bandwidth and delay spread in LoSNLoS scenar-ios have been studied at 5GHz [25 26] and 60GHz [27] In[28 29] statistical models characterizing the shadowing inLoS and NLoS scenarios were extracted from indoor mea-surements In outdoor scenarios path loss models have beenattracting more attention [30ndash34] Furthermore statistics ofthe LoS coverage rate in urban areas for broadband wirelessaccess systems and the LoS existence probability versuselevations of departure of downlink channels have beeninvestigated in [35 36] respectively

Experiment-based stochastic channel modeling relies onmeasurement campaigns to be conducted extensively in theenvironments of interest Many conventional models havebeen extracted based on dedicated channel sounding activ-ities However due to practical concerns such as power con-straints poor mobility of the sounding equipment andlimited number ofmeasurement routes the channel observa-tions do not possess sufficient randomness to characterize thechannels ergodically leading to limited applicability of theresultant models As an alternative using in-service publicwireless communication systems or networks to measurechannels in large areas can be adopted In [37] a public basestation was used as a transmitter and the received signalstrength was measured in the whole coverage of the basestationThe advantage of channelmeasurements in in-servicenetworks is that the channel characteristics observed areidentical with those experienced by the user equipment(UE) thus the channel statistics extracted are more realisticCompared with hundreds of MHz bandwidth usuallyadopted in the dedicated wideband channel measurementcampaigns the measurement based on the conventional 2Gpublic system usually utilizes hundreds of KHz bandwidthwhich facilitates the study of narrowband channel propertiessuch as the fading-related parameters [38] or performance-related parameters for example the out-of-channel interfer-ence [39] Along with fast developing commercial systems

such as LTE-Advanced wideband characteristics of channelscan be measured based on the in-service networks Channelmodeling based on these data began to attract more attentionrecently [36 40]

In this work a receiver was constructed to collect thedownlink signals at a hot-spot area of Shanghai in a com-mercial in-service UMTS network The impulse responses ofthe channels are calculated and a high-resolution estimationalgorithm derived based on the SAGE scheme is applied toextract multipath parameters After appropriately detectingthe LoS and NLoS channels by using the novel LoS detectionvarious aspects of LoSNLoS channels are investigated andthe correspondingmodels are established based on the exten-sive observations They include the deterministic channelmaps which are the geographic map overlaid with spotscolor-coded by channel characteristics and statistical modelscharacterizing the channel gain shadowing and multipathfading Furthermore the statistics of some newly definedchannel parameters are studiedwhich include (i) the distancebetween NBs and the user equipment in LoS and NLoS con-ditions (ii) the life-distance of LoS and NLoS channels (iii)the LoS existence probability per carrier for a UE (iv) the LoSexistence probability per carrier for a NB and (v) powervariation at LoS-NLoS transition and the transition durationThesemodels are useful for the design and performance eval-uation of communication techniques adaptive to the LoS andNLoS scenarios

The rest of this paper is organized as follows Section 2introduces the measurement equipment environments andthe specifications applied for data acquisition in an in-serviceUMTS network The main steps of processing the mea-surement data to obtain the estimates of channel impulseresponses are also described In Section 3 the LoSNLoSdetection method is introduced Section 4 elaborates detailsof the models established based on the LoSNLoS channelcharacteristics obtained Finally conclusive remarks areaddressed in Section 5

2 Measurement Equipment CampaignsSetup and Postprocessing

Figure 1 illustrates the diagram of the data acquisition pro-cessing and modeling procedure adopted for characterizingLoSNLoS channels based on the signals of a UMTS networkThe equipment used to acquire the signals consists of thefollowing components a Universal Software-Defined RadioPeripheral (USRP) device of type N210 [41] controlled by theGNU radio software in a computer and a global positioning

International Journal of Antennas and Propagation 3

Figure 2 A photograph of the Nanjing Road

system (GPS) module that reports the longitude and latitudeof themeasurement location to the computer through a serialport the storage disk an omnidirectional antenna workingin 2-3GHz UMTS frequency band and a lead-acid batteryAll this equipment was loaded in a trolley moving alongpredefined routes during the field measurements

Themeasurement campaign was performed in the pedes-trian zone along Nanjing Road Shanghai China This areais a typical hot-spot of the local UMTS network wherecomplex streets and densely distributed skyscrapers jointlycreate a city canyon environment Multiple macro- micro-or pico-cellular UMTS stations were deployed in the areaFigure 2 shows a photograph taken on the Nanjing Road dur-ing the measurements Measurements along the streets withdifferent styles of corners provide a large amount of channelobservations which are sufficient statistically for modelingthe realistic LoSNLoS channels including the channel evo-lution characteristics during the LoS-to-NLoS transitions andvice versa

During the measurements the receiver moves at a walk-ing speed of about 12ms Data acquisition was triggeredwhen the receiver departed from its previous location by 5meters and the duration of collecting the data at each locationis 1 second The carrier frequency was 21376GHz andthe receiving bandwidth was 20MHz The scrambling codeexisting in more than 5 consecutive frames is considered tobe a valid code It was found later that the data collected in thewhole campaign was from 78NBs each with a distinctivecombination of a scrambling code and a physical carrier

The raw data collected during the measurements waspost-processed in the following five steps

Step 1 (filtering) The measurement results show that threeUMTS carriers were used with central frequencies of21326GHz 21376GHz and 21426GHz respectively Theraw data received with 20MHz effective bandwidth at thecenter frequency of 21376GHz is first filtered to obtainthree baseband signals 119903

119888(119905) 119888 = 1 2 3 for these carriers

individually

Step 2 (synchronization) The beginnings of time-slots in thedownlink synchronization channels are detected by correlat-ing the baseband signals with the primary synchronizationcode (PSC) Multiple NBs are found by checking the domi-nant peaks in the PDPs of the obtained CIRs In addition the

estimates of the starting time 1199050of individual data frame in

the common pilot channels (CPICHs) and the index 119894 of esti-mating scrambling code group are obtained by solving themaximization problem

(0 ) = arg max

1199050 119894

100381610038161003816100381610038161003816100381610038161003816

int

119879

0

119903119888(119905) (119888119894

119904(119905 minus 1199050))lowast

119889119905

100381610038161003816100381610038161003816100381610038161003816

2

119888 = 1 2 3

(1)

where 119888119894119904(119905) 119894 isin [1 2 64] consists of 15 secondary syn-

chronization codes permuted according to the 119894th predefinedsequence 119879 = 10ms is the duration of a CPICH frame [42]and (sdot)lowast denotes the complex conjugate of given argument

Step 3 (scrambling code detection) The detection of scram-bling code is necessary in our case due to the fact that theinformation that how the scrambling codes are allocated inthe carriers is unknown The index 119895 of the scrambling codeused to modulate the CPICH is detected by

119895 = arg max119895isin119869119894

100381610038161003816100381610038161003816100381610038161003816

int

0+119879

0

119903119888(119905) 119904lowast

119895(119905 minus 0) 119889119905

100381610038161003816100381610038161003816100381610038161003816

2

119888 = 1 2 3 (2)

where the vector 119869119894= ( minus 1) sdot 8 + [1 2 8] contains the

indices of 8 scrambling codes in the th group and 119904119895(119905)

denotes the 119895th scrambling code Notice that for some chan-nels although the index of scrambling code group can bedetected in Step 2 none of the scrambling codes within thegroup exhibits significantly higher likelihood of being thecorrect code than the others This is probably due to the factthat the frequency offset between the NBs and USRP is non-negligible resulting in low SNR or signal to interference andnoise power ratio (SINR) in the CPICHs In such cases thedata acquired at this location will not be considered formodeling

Step 4 (CIR extraction) For the channels identified with the119895th scrambling code at carrier 119888 the impulse response ℎ

119888119896(120591)

of the channels observed in the 119896th CPICH frame is calcu-lated as

ℎ119888119896(120591) = int

0+119879

0

119903119888(119905 + 119896119879) 119904

lowast

119895(119905 minus 120591) 119889119905 (3)

Step 5 (parameter estimation) It can be shown that the chan-nel coherence time is beyond the duration of one frame evenin the worst case where the people in the environment aremoving at the largest speed of 3ms Thus the channelobservations obtained in one data frame are considered to becoherent and applicable for multipath extraction A SAGEalgorithm is derived and used to estimate the delays and com-plex attenuations ofmultiple specular propagation paths fromthe CIRs observed in individual data frames Readers mayrefer to [43] for the derivation of the SAGE algorithm andthe steps for implementation Based on a practical principlethat the paths estimated with power 20 dB lower than themaximum are negligible we found that 10 paths per CIR aresufficient for reconstructing the dominant portion of theCIRs in most cases Furthermore observations not reported


1 2 3 4 5times10minus6

0

05

1

15

2

25

3

35

times10minus8

Delay (s)

Pow

er

Original PDPReconstructed PDP

Figure 3An example of an original channel PDP and the PDPof thechannel reconstructed based on the multipath parameter estimatesThe powers of the estimated paths versus the path delays are alsoillustrated

here also show that the parameter estimates usually convergewithin 30 iterations Figure 3 depicts the comparison betweenthe original PDP for the channel observed in a CPICH frameand that calculated from the reconstructed signals that arecomputed based on the estimated 10 specular pathsThe stemplot for the powers of the estimated paths with the asteroidmarker is also illustrated in Figure 3 It is obvious that theoriginal PDP is well approximated by the reconstructed PDPindicating that the path components estimated by using theSAGE algorithm can represent the channel impulse responseaccurately

3 LoSNLoS Detection

The LoS detection method considered here is based on theestimated multipath parameters It is a common belief thatthe LoS path arrives at the UE with the shortest delayHowever due to the limited delay resolution of the commu-nication signals a LoS path may be combined with NLoSpaths which have slightly larger delays than the formerThusa multipath cluster can be identified in the vicinity of the trueLoS pathThis is particularly the case in urban environmentswhere the local clutter around the NB may generate a largeamount of diffraction or reflection paths with delays close tothe LoS path Considering these situations the LoS detectionmethod adopted here is designed for two cases In the firstcase where the first-arrival path has the largest power amongall estimated paths a clear decision that the channel is a LoSchannel is made without further analysis In the second casewhere the first-arrival path does not have the largest powerthe deviation 120588 between the powers of the strongest path andthis first-arrival path is calculated and used to determine the

0 5 10 15 20 25 300

005

01

015

02

025

03

035

Power deviation 120588 (dB)

120588th

LoSNLoS

Prob

abili

ty d

ensit

y fu

nctio

n

Figure 4 The pdfs of the deviation 120588 between the powers of thestrongest path and the first-arrival path in the case where the first-arrival path does not have the largest power among all estimatedpaths

channel scenarios based on a hypothesis testing A predefinedthreshold 120588th is applied in the testing where the channel isLoS for 120588 gt 120588th and NLoS for 120588 le 120588th respectively Thesuitable value of 120588th was identified through themeasurementscollected at a corner on Nanjing Road where the receivermoved from LoS to NLoS scenarios Figure 4 depicts theempirical probability density functions (pdfs) of 120588 observedin the true LoS and NLoS cases It can be observed that thetwo pdfs interset to overlap around 120588 = 9ndash12 dB Our cal-culation shows that by setting 120588th = 10 dB the accumulativeprobability for detection error in this case can be kept lessthan 01Thus practically 120588th = 10 dB is chosen for the hypo-thesis testing of a channel being LoS or NLoS in the environ-ment considered here

It is worth mentioning that the measurement environ-ments were time-variant with people walking around whichcan significantly influence the accuracy of LoS detectionbased on a signal observation To improve the detectionaccuracy it is necessary to select the decision in majoritywhen multiple decisions are available In the case consideredhere the decision with higher occurrence frequency among49 CPICH frames per location is selected In most locationsmore than 35 frames were observed to provide common LoSdetection result

By using the aforementioned detection method LoS andNLoS channels observed with valid scrambling codes areidentified for allmeasurement locations Figures 5(a) and 5(b)demonstrate an example of the LoS and NLoS channel mapsforNanjingRoadwhere the geometricmapor the photographof the environment is overlaidwith spotsmarking the channelcategories observed by the UE Notice that since there aremany NBs deployed along the Nanjing Road transmittingsignals with different scrambling codes at three carriersFigure 5 illustrates the LoS detection results at all measured


65 131 196 261 326 392 457 522 587 653

336598131163196228261

Locations of LoS scenarios

x (m)

y (m

)

(a) Locations where LoS channels are observed

65 131 196 261 326 392 457 522 587 653

336598131163196228261

x (m)

y (m

)

Locations of NLoS scenarios

(b) Locations where NLoS channels are observed

Figure 5 The locations where the channels are found to be LoS and NLoS channels along the Nanjing Road

locations for the channel with the largest power among allvalid channels identified It can be observed that more LoSchannels are found than the NLoS channels at the left endof the Nanjing Road This is reasonable as that area is anopen crosswith twomain roads intersect FurthermoreNLoSchannels are widely observed along the Nanjing Road a typ-ical street canyon environment where the diffraction and thereflections from skyscrapers are themajor propagationmech-anisms

4 LoSNLoS Channel Models

In this section various aspects of the LoS and NLoS channelsare analyzed and the corresponding stochastic models areestablished based on 664 LoS and 1181 NLoS channel obser-vationsThese characteristics include the conventional fadingstatistics and the distributions of some novel parameters Fornotational convenience we use ℎLoS119898(120591) and ℎNLoS119899(120591) todenote the CIR of the119898th observation of LoS channel and the119899th observation of NLoS channel respectively The corre-sponding narrowband channel coefficients ℎLoS119898 and ℎNLoS119899are calculated as ℎLoS119898 = int ℎLoS119898(120591)119889120591 and ℎNLoS119899 =

int ℎNLoS119899(120591)119889120591 respectively The channel gains 119875LoS119898 and119875NLoS119899 represented in dB are calculated as 119875LoS119898 =

20 log10|ℎLoS119898| and 119875NLoS119899 = 20 log

10|ℎNLoS119899| respectively

41 Channel Gain Shadowing and Multipath Fading Thechannel gain 119875LoS119898 can be written as 119875

119903= 119875119905minus 119875119901+ 119875119904+ 119875119898

where 119875119905 119875119901 119875119904 and 119875

119898represent the transmission power

path loss shadowing and multipath fading denoted in dBrespectively Figure 6 depicts the empirical cumulative dis-tribution functions (cdfs) of the channel gains 119875LoS119898 and119875NLoS119899 respectively It can be observed that the channel gainsfor LoS are concentrated in the region more to the right ofthe abscissa than to the NLoS scenario Different kinds ofdistributions including the 119898-Nakagami Rice Weibull nor-mal and 120572-120583 are used to fit with the empirical results TheKolmogorov-Smirnov test (K-S test) has been conducted forevaluating the consistency of the analytical pdf and the empir-ical graphs The results show that both cdfs are well fittedby normal distributions with appropriately chosen parame-ters reported in Table 1 It can be observed from the table thatthe average of 119875LoS119898 is 612 dB higher than that in the NLoS

minus150 minus140 minus130 minus120 minus110 minus100 minus90 minus80 minus700

01

02

03

04

05

06

07

08

09

1

Channel gain Pr (dB)

Cum

ulat

ive d

istrib

utio

n fu

nctio

n

Empirical data for LoS Empirical data for NLoSNormal cdf fitted to LoS Normal cdf fitted to NLoS

Figure 6 The empirical and fitted cdfs of the channel gains in theLoS and NLoS scenarios

Table 1 Parameters of analytical pdfs fitted to the empirical data

LoS scenarios NLoS scenariosChannel gain(normal) 120583 = minus9068 120590 = 827 120583 = minus9680 120590 = 1292

Shadowing (120572-120583) 120572 = 09184 120583 = 45918 120572 = 16327 120583 = 09184Multipath (120572-120583) 120572 = 24490 120583 = 21429 120572 = 25510 120583 = 10204

scenario and the standard deviation of the channel gain in theLoS scenario is 465 dB less than that in the NLoS scenarioThese statistical differences indicate that the existence of theLoS path in the channel can not only bring 6 dB gain inaverage for the received signal but also reduce the variationof the channel gain significantly

To calculate the gain due to the shadowing 119875119904which

attributes to the large structures around the receiver it isnecessary to subtract the path loss 119875

119901caused by free-space

propagation and the terrain features In our case since theexact locations of the NBs are unknown the path loss maynot be calculated analytically by using existing models As


05 1 15 2 25 3 35 4 45 5 55

01

02

03

04

05

06

07

08

09

1

Cum

ulat

ive d

istrib

utio

n fu

nctio

n

Channel gain due to shadowing Ps (linear)

Empirical data for LoS Empirical data for NLoS120572-120583 cdf fitted to LoS 120572-120583 cdf fitted to NLoS

Figure 7The empirical and fitted cdfs of shadowing in the LoS andNLoS scenarios

an alternative the empirical path loss is computed in thefollowing manner For any location say 119860 channel gainsobserved in the surrounding locations within 50-m radius areaveraged to obtain the estimate of the path losses 119875

119901LoS and119875119901NLoS at 119860 for LoS and NLoS scenarios respectively The

shadowing of a LoS channel 119875119904LoS at the location 119860 is then

calculated by taking the mean of 119875LoS119898 minus 119875119901LoS where 119875LoS119898is the channel gain observed in a LoS scenario at any locationless than 10-m deviation from 119860 Similar calculations areapplied to calculate the shadowing of a NLoS channel 119875

119904NLoSFigure 7 depicts the empirical cdfs of the shadowing in

the LoS and NLoS scenarios respectively The cdfs of 120572-120583 distributions which were found to fit the best with theempirical graphs among multiple candidates are also illus-trated in Figure 7 The parameters of the pdfs of the adopteddistributions are reported in Table 1 It can be observed fromFigure 7 that the mean of the gain due to shadowing both inLoS and NLoS scenarios are around 1 in linear scale whichindicates that the path loss has been correctly removed in thecalculationThe spread of the shadowing is larger in theNLoSchannels than that in the LoS channels This observationshows that more complex environments exist around thereceiver in the NLoS scenarios which create larger variationof the shadowing than in the LoS scenarios

Multipath fading in a channel can be calculated by sub-tracting the path loss and shadowing from the over channelgain Figure 8 depicts the empirical cdfs of channel gainattributed to the multipath fading in the LoS and NLoS sce-narios respectively The 120572-120583 distributions which were shownby the K-S test to fit the best to the empirical graphs are alsodepicted in Figure 8 It can be observed that the multipathfading magnitudes in the linear scaling are distributed withinthe interval of [0 3] with mean values equal to 1 for both sce-narios Furthermore the cdf is more concentrated in the LoSthan in the NLoS scenarios indicating that the variation ofthemultipath fading ismore severe inNLoS scenarios Table 1


05 1 15 2 25 3 35

01

02

03

04

05

06

07

08

09

1

Channel gain due to multipath fading Pm (linear)

Cum

ulat

ive d

istrib

utio

n fu

nctio

n

Figure 8 The empirical and fitted cdfs of multipath fading in theLoS and NLoS scenarios

reports the parameters of the 120572-120583 cdfs that fit well with theempirical data

42 Distance between the NB and the UE As the locations ofthe NBs are unknown in our case it is difficult to establish thepath loss models for both LoS and NLoS scenarios Howeverthe distributions of the path loss which can be calculatedbased on amethod that will be introduced later are applicableto deducing the statistics of the distances between NBs andUE in either LoS or NLoS cases Such information is consid-ered to be useful for channel simulations and designing thetransmission power for NBs and UE with respect to the LoSand NLoS scenarios

Here119875119905can be considered to be deterministic119875

119901depends

on the direct distance 119889 between the NB and the UE and 119875119904

119875119898are uncorrelated random variables In the case considered

here the unknown 119889 can be viewed as a random variable andconsequently119875

119901 as a deterministic function of 119889 is a random

variable as wellThe pdf of 119875119903is the convolution of the pdfs of

minus119875119901 119875119904 and 119875

119898 Under the condition that the empirical pdfs

of 119875119904and 119875119898are known we can calculate the pdf of 119875

119901by the

deconvolution operation Figure 9 depicts the estimatedempirical pdf of the119875

119901in LoS andNLoS scenarios It is worth

mentioning that since 119875119905is unknown the absolute value of

the abscissa of Figure 9 may be incorrect It can be observedthat the mean path loss is 7 dB less in LoS than in NLoSscenarios and the spread of the path loss is larger in NLoSthan in LoS scenarios

Since the path loss is a deterministic function of thedistance 119889 between the NB and the UE we may deduce thedistribution of 119889 from the estimated pdf of 119875

119901 For simplicity

the following path loss model [44 Section 26] is used

119875119901= minus10120574 log

10(119889

1198890

) + 20 log10(

120582

41205871198890

) (4)


40 50 60 70 80 90 100 110 120 130 140 1500

001

002

003

004

005

006

007

008

Path loss (dB)

Estim

ated

pro

babi

lity

dens

ity fu

nctio

n

LoSNLoS

Figure 9The empirical pdfs of estimated path loss in LoS andNLoSscenarios

where the path loss exponent 120574 should be selected within[27 35] and the reference distance 119889

0is between 10 and

100m for urban micro-cellular environments In our case120574 = 3 and 119889

0= 50 are selected which can be changed if more

accurate path lossmodel is available for the environment con-sidered By assuming that the transmit power 119875

119905= 0 dB the

pdf of 119889 is obtained from the empirical pdf of 119875119901by replacing

119875119901with (4) Figure 10 illustrates the resultant pdfs of direct

distance from the NB to UE in LoS and NLoS scenariosrespectively It can be observed from Figure 10 that the LoSusers are maximum 25 km from the NBThemost of the LoSusers are located at 200m from the NB The NLoS users arewidely spread within the range from 0 to 4 km and theirmajority is around 300m from the NBs The nonzero proba-bility observed for large range up to 4 km indicates that thismay be due to the existence of macro-cellular NBs Theseresults clearly demonstrate that the LoS users are locatedcloser to the NB than the NLoS users In other words theNBs with small coverage areas tend to serve more LoS usersthan theNBs with larger coverage ranges in theNanjing Roadenvironments

43 Life-Distance of LoS Channel To implement for exam-ple the user-specific beamforming technique a UE is usuallyexpected to experience LoS scenarios in a long distanceThusit is important to investigate the statistics of the so-called ldquolife-distance of the LoSNLoS channelrdquo 119889

119888within which a UE

experiences the LoS channel constantly Based on the locationinformation of the measured spots the measurement routesare accurately reproduced In the case where the LoS or NLoSchannel appears in a certain number of consecutive locations119889119888can be calculated as the displacement from the first to the

last location Figure 11 illustrates an example of the fragmentswhere LoS orNLoS channels were constantly observed for theNBwith the scrambling code number 188 In this example themaximum of the life-distance of LoS channel is calculated as

0 500 1000 1500 2000 2500 3000 3500 40000

001

002

003

004

005

006

007

008

Estim

ated

pro

babi

lity

dens

ity fu

nctio

n

Distance between NB and UE (m)

LoSNLoS

Figure 10 The empirical pdfs of the distance between the NB andthe UE in LoS and NLoS scenarios

38 76 115 153 191 229 268 306

38

96

154

x (m)

y(m

)

Fragments of LoS channelsFragments of NLoS channels

Figure 11 An example of calculating life-distance of LoS and NLoSchannels for a NB with the scrambling code number 188

25 meters which was found along the spacious street Noticethat due to the limited GPS resolution and the fixed distanceof 5m between adjacent measured spots the minimum life-distance of LoS or NLoS channel is lower-bounded by 5m

Figure 12 depicts the cdf of the life-distance of LoS andNLoS channels calculated based on the observations of totally148 fragments for LoS and 298 fragments for the NLoS chan-nels It can be observed from Figure 12 that the longest LoSand NLoS life-distance can be up to 281m and 160mrespectively Furthermore nearly 50 of the LoS and NLoSlife-distances observed are 5 meters and 20 of the LoS andNLoS life-distances are within the range of 5 to 10metersThemajority of the life-distances observed are below 10 meterswhich indicates that the LoS and NLoS scenarios transitfrequently when a UEwalks on the city canyon environmentsas in the case of Nanjing Road

44 LoS Existence Probability for a UE and for a NB Thepro-bability of a UE experiencing the LoS channel at an arbitrarylocation in the network is of interest to know Considering


08 1 12 14 16 18 2 22 24 2604

05

06

07

08

09

1

LoSNLoS

282m 160m

5m

10m

Life-distance in LoS and NLoS scenarios in (log10 m)

Empi

rical

cum

ulat

ive d

istrib

utio

n fu

nctio

n

Figure 12 The empirical cdfs of the LoS and NLoS life-distance

that the UE may receive signals from more than one NB theldquoLoS existence probability for a UErdquo 119875LoSUE is defined as thepercentage of the NBs with the fact that the UE can set up aLoS channel among all valid NBs In the measurement datathe case of receiving two scrambling codes per carrier is themajority The 119875LoSUE for the two-scrambling-code cases isinvestigated based on available observations Figure 13depicts the empirical percentage of 119875LoSUE for three carriersIt can be observed that the distributions of 119875LoSUE are similarfor all three carriers which is probability due to the fact thatevery NB in Nanjing Road makes use of three carriers Fur-thermore the case of no LoS path existed to the two NBs hasthe probability larger than 04 The opposite situation that allNBs have LoS connections to UE has the probability around015 It is easy to deduce that the case where theUE has at leastone LoS connection to the surrounding NBs can happen withprobability 06 These results may be used to provide proba-bilistic assumptions for designing the cooperative multipoint(CoMP) techniques when the coexisting multilinks involveboth LoS and NLoS channels

For individual NBs it is important to know the distri-bution of a specific percentage of UE that are categorized asLoS users among all UE served by the NBThe ldquoLoS existenceprobability per NBrdquo 119875LoSNB is defined as the ratio of numberof LoS UE to the number of all UE in the coverage of a NBFigure 14 illustrates the empirical percentages of LoS exis-tence probability per NB per carrier calculated based onthe observations in total 78 NBs It can be observed fromFigure 14 that for three carriers the situation of all UE in aNB experiencing LoS channel is zero the situation in which20 of the UE in a NB are LoS users occurs with 04probability and the case with 80 of the UE being LoSusers happens with probability close to 02 This resultindicates that on the Nanjing Road the existing NBs can becategorized into two groups one with the most users beingLoS users and the other with the most users being NLoS

0 05 10

005

01

015

02

025

03

035

04

045

05

LoS existence probability per location

The l

ocat

ions

()

Carrier 1Carrier 2Carrier 3

Figure 13The empirical percentage of LoS existence probability perlocation


02 04 06 08 10

005

01

015

02

025

03

035

04

045

LoS existence probability per NB

The N

Bs (

)

Figure 14The empirical percentage of LoS existence probability perNB

users This categorizing scheme probability is due to the factthat the most NBs deployed on Nanjing Road are eithermicro- or pico-cellular NBs Depending on the type of theNB distinctive distribution of ldquoLoS existence probability perNBrdquo may be caused

In addition an interesting phenomenon observed fromFigure 14 is that for the carriers 2 and 3 higher 119875LoSNB isobserved with larger probability than in the case of carrier 1This is probably due to the reason that the carriers 2 and 3might be more widely used in the pico-cellular NBs than the


minus4 minus2 0 2 4 6 8 100

005

01

015

02

025

Prob

abili

ty d

istrib

utio

n fu

nctio

n

Empirical dataLaplacian pdf fitted

Power difference for LoS-to-NLoS transition (dB)

Figure 15 The empirical pdf and fitted Laplacian pdf for thepower difference of the channels before and after the LoS-to-NLoStransition

carrier 1 Since more LoS users exist in the pico-cellular NBsthan other types of NBs [45] the carriers 2 and 3 exhibithigher probability than the carrier 1 for larger 119875LoSNB

45 Channel Characterization of LoS-NLoSTransition Powerdiversity techniques used in the NOMA systems mitigatethe multipath fading by adapting the transmission powerfor example allocating low power for UE in LoS scenariosand higher power for UE in NLoS scenarios Designingsuch techniques requires accurate modeling of the powerdifference between the LoS and NLoS channels particularlyduring the transition process To design the transmissiontechniques used in the LoS and NLoS scenarios specificallyadaptation operations such as switching the techniques ormodifying operational parameters are necessary when thechannel changes from the LoS scenario to the NLoS scenarioor vice versa Thus it is necessary to characterize the channelvariations during the period of LoS-to-NLoS orNLoS-to-LoStransition Measurements were conducted around multiplecrosses along Nanjing Road to acquire the downlink data for1 minute while the receiver was moving The transitionsbetween LoS and NLoS channels are identified and theinterested parameters are analyzed

451 Power Difference at LoS-NLoS Transition Figure 15demonstrates that the empirical pdf of the power differencesΔ119875LoSminusNLoS = 119875LoS minus 119875NLoS at LoS-to-NLoS transition pointsThe fitted analytical graph is also illustrated It can beobserved from Figure 15 that the majority of the distributionof Δ119875LoSminusNLoS is located above 0 dB in the abscissa TheLaplace distribution with its parameters 120583 = 142 dB and 119887 =222 fits well with the empirical data Furthermore the obser-vation that Δ119875LoSminusNLoS is less than 0 dB indicates that forsome transitions the NLoS scenarios have higher received

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0

10

20

0

10

20

0

10

5

15

Threshold

Threshold

Threshold

Threshold

Transition duration T (s)




Type (i) transition

Type (ii) transition

Type (iii) transition

Type (iv) transition

10

5

15

20

ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)

Figure 16 Examples of the four types of transition behavior

power than the LoS scenarios This phenomenon may be dueto the fact that for some locations such as the corners ofstreets when a UE turns its moving direction significantlythe LoS-to-NLoS transition occurs and the constellation ofmultipath propagation may include new NLoS paths whichaddmore power to the channel than the power lost due to theabsence of the LoS pathThis is different with the obstructioncases where the overall propagation constellation is retainedwith only the LoS path absent due to blockage

452 LoS-to-NLoS Transition Duration The LoS-to-NLoStransition duration 119879 is defined as the time spent on theperiod from the starting of transition to the moment thatthe NLoS channel begins to stabilize Here the channel isconsidered stabilized if the LoS or NLoS status is maintainedfor no less than 10 frames The observations show that allLoS-to-NLoS transitions can be categorized into four typesdepending on the times of back-and-forth between LoS andNLoS scenarios Figure 16 illustrates examples for the fourtransition types Type (i) is referred to as the case where a LoSchannel changes to a NLoS channel which stabilizes imme-diately For such cases 119879 is regarded to be 1 frame Type (ii)


001 003 004 0050

01

02

03

04

05

06

07

08

09

Transition duration (s)

Empi

rical

occ

urre

nce f

requ

ency

Figure 17 The empirical occurrence frequency of transition dura-tion from LoS points to NLoS points or NLoS points to LoS points

denotes the situation that after the transition starts theNLoSchannel changes back to the LoS scenario once and thentransits again to the NLoS scenario For this case 119879 is con-sidered to be 3 frames Type (iii) is similar with type (ii) butthe difference is that during the transition theNLoS scenariolasts for two frames before transiting back to the LoS Type(iv) is similar with type (iii) except for the fact that the LoSchannel remains for two frames before transiting to the NLoSscenario in the second time For the latter two types 119879 isconsidered to be 4 and 5 frames respectively Figure 17 depictsthe empirical probability of 119879 for the four types consideredIt can be observed that the type I transition occurs with thehighest probability of 09 among all types Only a smallamount of transitions undergoes the fluctuations in types (ii)to (iv) which cost a longer time for channel being stabilizedThese results show that for the walking speed in a citycanyon environment most of the LoS-to-NLoS transitionsare accomplished rapidly in one frame that is 10ms Longertransitions lasting for more than one frame can be neglectedwhen designing systems for their low occurrence probabili-ties

5 Conclusions

A channel measurement campaign has been conductedaiming at characterizing LoSNLoS channels by using thedownlink signals in a commercial UMTS network for apedestrian zone in Nanjing Road Shanghai Techniques weredeveloped for extracting the channel impulse responses andmultipath parameters were estimated by using the SAGEalgorithm Channels observed were categorized into LoSand NLoS by using a novel multipath-based LoS detectionmethod A map overlaid by the channel properties observedat multiple locations demonstrated that LoS channels mostlyexist in open areas while NLoS channels are dominant in thecity canyons with densely distributed skyscrapers Statisticalmodels for the fading coefficientswere first establishedwhich

have shown that the existence of the LoS path can bring6 dB gain on average and reduce the gain variation by 5 dBAdditionlly the statistics of some newly defined parameterswere also investigated which include the distributions of theLoS and NLoS life-distance the LoS existence probability peruser equipment and per NB power difference at the LoS-to-NLoS transition and the transition duration Importantfindings were discovered for example the LoS and NLoSlife-distances are usually less than 10m the LoS-to-NLoStransition can be accomplished less than 10ms and soforth These characteristics are important for the design andpreference evaluation of the techniques or systems adaptiveto LoS and NLoS scenarios

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors wish to acknowledge Dr Junhe Zhou TongjiUniversity for his valuable comments on the paperThis workis jointly supported by Huawei Technology Company in theresearch Project ldquoAnalysis and Verification of Channel Fin-gerprinting Characteristics in In-Service UMTS Networksrdquo(YBWL2010KJ010) Huawei innovation research ProjectldquoHigh Dimensional MIMO Channel Estimation Modelingand Measurement Combined with Antenna Statusrdquo the Sci-ence and Technology Commission of Shanghai MunicipalityldquoSystem Design and Demo-Construction for CooperativeNetworks of High-Efficiency 4G Wireless CommunicationsinUrbanHot-Spot Environmentsrdquo (13510711000) and theKeyProgram of National Natural Science Foundation of China(Grant no 61331009)

References

[1] Universal Mobile Telecommunications System (UMTS) SpacialChannel Model for Multiple Input Multiple Output (MIMO)Simulations (3GPP TR 25996 Version 800 Release 8) ETSI Std

[2] WINNER II interim channel models IST-4-027756 WINNERD111 Std

[3] D Gesbert M Shafi D Shiu P J Smith and A Naguib ldquoFromtheory to practice an overview of MIMO space-time codedwireless systemsrdquo IEEE Journal on Selected Areas in Communi-cations vol 21 no 3 pp 281ndash302 2003

[4] M Coldrey H Koorapaty J Berg et al ldquoSmall-cell wirelessbackhauling a non-line-of-sight approach for point-to-pointmicrowave linksrdquo in Proceedings of the 76th IEEE VehicularTechnology Conference (VTC Fall rsquo12) September 2012

[5] S Mukherjee WiMAX Antennas PrimermdashA Guide to MIMOand Beamforming 2010

[6] B Furht and S A Ahson Long TermEvolution 3GPP LTERadioand Cellular Technology 2009

[7] Y Saito Y Kishiyama A Benjebbour T Nakamura A Li andK Higuchi ldquoNon-orthogonal multiple access (NOMA) forfuture radio accessrdquo in Proceedings of the IEEE 77th VehicularTechnology Conference (VTC rsquo13) 2013


[8] K Higuchi and Y Kishiyama ldquoNon-orthogonal access withrandom beamforming and intra-beam SIC for cellular mimodownlinkrdquo in Proceedings of the IEEE 78th Vehicular TechnologyConference (VTC rsquo13) pp 1ndash5 Las Vegas Nev USA September2013

[9] J Umehara Y Kishiyama and K Higuchi ldquoEnhancing userfairness in non-orthogonal access with successive interferencecancellation for cellular downlinkrdquo in Proceedings of the IEEE14th International Conference on Communication Systems (ICCSrsquo12) pp 324ndash328 Singapore November 2012

[10] D Jiang and L Delgrossi ldquoIEEE 80211p towards an interna-tional standard for wireless access in vehicular environmentsrdquoin Proceedings of the 67th IEEE Vehicular Technology Conference(VTC rsquo08) pp 2036ndash2040 May 2008

[11] X Cheng C Wang B Ai and H Aggoune ldquoEnvelope levelcrossing rate and average fade duration of nonisotropic vehicle-to-vehicle Ricean fading channelsrdquo IEEE Transactions on Intel-ligent Transportation Systems vol 15 no 1 pp 62ndash72 2014

[12] X ChengQ YaoMWenCWang L Song andB Jiao ldquoWide-band channel modeling and intercarrier interference cancel-lation for vehicle-to-vehicle communication systemsrdquo IEEEJournal on Selected Areas in Communications vol 31 no 9 pp434ndash448 2013

[13] V Nurmela T Jamsa P Kyosti V Hovinen and J MedboldquoChannel modelling for device-to-device scenariosrdquo IC 1004TD(13)08009 2013

[14] D W Matolak ldquoChannel modeling for vehicle-to-vehicle com-municationsrdquo IEEE Communications Magazine vol 46 no 5pp 76ndash83 2008

[15] Y Zhang R Yu M Nekovee Y Liu S Xie and S GjessingldquoCognitive machine-to-machine communications visions andpotentials for the smart gridrdquo IEEE Network vol 26 no 3 pp6ndash13 2012

[16] J Borras P Hatrack and N B Mandayam ldquoDecision theoreticframework for NLOS identificationrdquo in Proceedings of the 48thIEEE Vehicular Technology Conference (VTC rsquo98) pp 1583ndash1587May 1998

[17] M Wylie-Green and J Holtzman ldquoLocating mobile stationswith non-line-of-sight measurementsrdquo Advances in WirelessCommunications p 359 1998

[18] S Venkatraman J Caffery Jr and H- You ldquoLocation usingLOS range estimation inNLOS environmentsrdquo in Proceedings ofthe 55th IEEE Vehicular Technology Conference (VTC rsquo02) vol2 pp 856ndash860 May 2002

[19] J Schroeder S Galler K Kyamakya and K Jobmann ldquoNLOSdetection algorithms for ultra-wideband localizationrdquo in Pro-ceedings of the 4th Workshop on Positioning Navigation andCommunication (WPNC rsquo07) pp 159ndash166 March 2007

[20] A Maali H Mimoun G Baudoin and A Ouldali ldquoA new lowcomplexity NLOS identification approach based on UWBenergy detectionrdquo in Proceedings of the IEEE Radio andWirelessSymposium (RWS rsquo09) pp 675ndash678 San Diego Calif USAJanuary 2008

[21] A Lakhzouri E S Lohan R Hamila and M Rentors ldquoExten-ded Kalman filter channel estimation for line-of-sight detectionin WCDMA mobile positioningrdquo EURASIP Journal on AppliedSignal Processing vol 2003 no 13 pp 1268ndash1278 2003

[22] F Benedetto G Giunta A Toscano and L Vegni ldquoDynamicLOSNLOS statistical discrimination of wireless mobile chan-nelsrdquo in Proceedings of the 65th IEEE Vehicular TechnologyConference (VTC rsquo07) pp 3071ndash3075 2007

[23] T Svantesson and J Wallace ldquoStatistical characterization ofthe indoor mimo channel based on losnlos measurementsrdquo inProceedings of the 36th Asilomar Conference on Signals Systemsand Computers vol 2 pp 1354ndash1358 IEEE 2002

[24] K Yu M Bengtsson B Ottersten D McNamara P Karlssonand M Beach ldquoModeling of wide-band MIMO radio channelsbased on NLoS indoor measurementsrdquo IEEE Transactions onVehicular Technology vol 53 no 3 pp 655ndash665 2004

[25] C Bergljung and P Karlsson ldquoPropagation characteristics forindoor broadband radio access networks in the 5GHz bandrdquo inProceedings of the 9th IEEE International Symposium on Per-sonal Indoor and Mobile Radio Communications (PIMRC rsquo98)vol 2 pp 612ndash616 September 1998

[26] K Yu M Bengtsson B Ottersten D McNamara P Karlssonand M Beach ldquoSecond order statistics of NLOS indoor MIMOchannels based on 52 GHzmeasurementsrdquo inProceedings of theIEEE Global Telecommunicatins Conference (GLOBECOM rsquo01)pp 156ndash160 November 2001

[27] H Yang P F M Smulders and M H A J Herben ldquoFrequencyselectivity of 60GHz LOS and NLOS indoor radio channelsrdquo inProceedings of the 63rd IEEE Vehicular Technology Conference(VTC rsquo06) vol 6 pp 2727ndash2731 July 2006

[28] Z Lin X Peng K B Png and F Chin ldquoKronecker modellingfor correlated shadowing in UWB MIMO channelsrdquo in Pro-ceedings of the IEEE Wireless Communications and NetworkingConference (WCNC rsquo07) pp 1585ndash1589 March 2007

[29] J C Liberti and T S Rappaport ldquoStatistics of shadowing inindoor radio channels at 900 and 1900 MHzrdquo in Proceedings ofthe IEEE Military Communications Conference (MILCOM rsquo92)pp 1066ndash1070 1992

[30] J E Berg R Bownds and F Lotse ldquoPath loss and fadingmodelsfor microcells at 900 MHzrdquo in Proceedings of the 42th IEEEVehicular Technology Conference (VTC rsquo92) pp 666ndash671 1992

[31] S Y Tan and H S Tan ldquoPropagation model for microcellularcommunications applied to path loss measurements in OttawaCity streetsrdquo IEEETransactions onVehicular Technology vol 44no 2 pp 313ndash317 1995

[32] V Erceg L J Greenstein S Y Tjandra et al ldquoEmpirically basedpath loss model for wireless channels in suburban environ-mentsrdquo IEEE Journal on Selected Areas in Communications vol17 no 7 pp 1205ndash1211 1999

[33] E Green and M Hata ldquoMicrocellular propagation measure-ments in an urban environmentrdquo in Proceedings of the IEEEPersonal Indoor and Mobile Radio Communications (PIMRCrsquo91) pp 324ndash328 1991

[34] Y Oda K Tsunekawa and M Hata ldquoAdvanced LOS path-lossmodel inmicrocellularmobile communicationsrdquo IEEETransac-tions on Vehicular Technology vol 49 no 6 pp 2121ndash2125 2000

[35] G Bauer andR Jakoby ldquoAnalysis and estimation of line-of-sightcoverage in urban areas for fixed broadband wireless accesssystemsrdquo in Proceedings of the 10th European Conference onWireless Technology (ECWT rsquo07) pp 229ndash232 October 2007

[36] J Yu X Yin J Chen et al ldquoChannelmaps and stochasticmodelsin elevation based on measurements in operating networksrdquoin Proceedings of the IEEE Wireless Communications and SignalProcessing (WCSP rsquo13) pp 1ndash6 2013

[37] M Ylitervo and M Vayrynen ldquoMethods for making activechannel measurements in a personal base station environmentrdquoUS Patent 5 726 981 1998

[38] K Pentikousis M Palola M Jurvansuu and P Perala ldquoActivegoodput measurements from a public 3GUMTS networkrdquoIEEE Communications Letters vol 9 no 9 pp 802ndash804 2005


[39] J Potman FW Hoeksema and C H Slump ldquoAdjacent channelinterference in UMTS networks proriscrdquo in Proceedings of the17th Annual Workshop on Circuits Systems and Signal Process-ing Veldhoven The Netherlands 2006

[40] X Yin N Zhang Z Zhong et al ldquoSystem-level channel mod-eling based on active measurements from a public 3GUMTSnetworkrdquo in Proceedings of the 78th IEEE Vehicular TechnologyConference (VTC rsquo13) pp 1ndash5 2013

[41] ldquoUSRP n210 datasheetrdquo Tech Rep httpswwwettuscompro-ductdetailsUN210-KIT

[42] Technical Specification Group Radio Access Network PhysicalChannels and Mapping of Transport Channels onto PhysicalChannels (FDD)(Release 9) 3rd Generation Partnership Project3GPP TS 25211 V920 (2010-09) Std

[43] B H Fleury M Tschudin R Heddergott D Dahlhaus andK I Pedersen ldquoChannel parameter estimation in mobile radioenvironments using the SAGE algorithmrdquo IEEE Journal onSelected Areas in Communications vol 17 no 3 pp 434ndash4501999

[44] A GoldsmithWireless Communications CambridgeUniversityPress 2005

[45] L Thiele M Peter and V Jungnickel ldquoStatistics of the Riceank-factor at 52 GHz in an urban macro-cell scenariordquo in Pro-ceedings of the 17th IEEE International Symposium on PersonalIndoor andMobile Radio Communications (PIMRC rsquo06) pp 1ndash5September 2006

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Active and Passive Electronic Components

Control Scienceand Engineering

Journal of



RotatingMachinery


Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics


Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation



DistributedSensor Networks



Dataacquisition

USRPGNUradio

GPS Diskstorage

Data processing

Filtering

Synchronization

Analysis and modeling

Scrb detection

CIR extraction

SAGE estimation

LoSNLoS detection

Models of fading

Models ofchannel parameters

Figure 1 A diagram of channel measurement data analysis and modeling procedure adopted for LoSNLoS channel characterization

[16ndash18] was adopted for LoS detection in the early stage Thelatter approach is mainly based on the shape or statisticalproperties of the CIRs for example the confidence matrix[19] the delay spread [19] the power delay profile (PDP) [20]and the narrowband 119870-factor [21 22] In our contributionthe LoS detection is performed based on the power differenceamong estimated multipath components which has beenproven to be effective through measurement campaigns

Some LoSNLoS propagation models have been reportedin the literature According to measurement environmentsthese models characterize channels in either indoor or out-door scenarios In indoor cases the distribution of the chan-nel gains for the LoSNLoS scenarios is presented [23 24]Coherence bandwidth and delay spread in LoSNLoS scenar-ios have been studied at 5GHz [25 26] and 60GHz [27] In[28 29] statistical models characterizing the shadowing inLoS and NLoS scenarios were extracted from indoor mea-surements In outdoor scenarios path loss models have beenattracting more attention [30ndash34] Furthermore statistics ofthe LoS coverage rate in urban areas for broadband wirelessaccess systems and the LoS existence probability versuselevations of departure of downlink channels have beeninvestigated in [35 36] respectively

Experiment-based stochastic channel modeling relies onmeasurement campaigns to be conducted extensively in theenvironments of interest Many conventional models havebeen extracted based on dedicated channel sounding activ-ities However due to practical concerns such as power con-straints poor mobility of the sounding equipment andlimited number ofmeasurement routes the channel observa-tions do not possess sufficient randomness to characterize thechannels ergodically leading to limited applicability of theresultant models As an alternative using in-service publicwireless communication systems or networks to measurechannels in large areas can be adopted In [37] a public basestation was used as a transmitter and the received signalstrength was measured in the whole coverage of the basestationThe advantage of channelmeasurements in in-servicenetworks is that the channel characteristics observed areidentical with those experienced by the user equipment(UE) thus the channel statistics extracted are more realisticCompared with hundreds of MHz bandwidth usuallyadopted in the dedicated wideband channel measurementcampaigns the measurement based on the conventional 2Gpublic system usually utilizes hundreds of KHz bandwidthwhich facilitates the study of narrowband channel propertiessuch as the fading-related parameters [38] or performance-related parameters for example the out-of-channel interfer-ence [39] Along with fast developing commercial systems

such as LTE-Advanced wideband characteristics of channelscan be measured based on the in-service networks Channelmodeling based on these data began to attract more attentionrecently [36 40]

In this work a receiver was constructed to collect thedownlink signals at a hot-spot area of Shanghai in a com-mercial in-service UMTS network The impulse responses ofthe channels are calculated and a high-resolution estimationalgorithm derived based on the SAGE scheme is applied toextract multipath parameters After appropriately detectingthe LoS and NLoS channels by using the novel LoS detectionvarious aspects of LoSNLoS channels are investigated andthe correspondingmodels are established based on the exten-sive observations They include the deterministic channelmaps which are the geographic map overlaid with spotscolor-coded by channel characteristics and statistical modelscharacterizing the channel gain shadowing and multipathfading Furthermore the statistics of some newly definedchannel parameters are studiedwhich include (i) the distancebetween NBs and the user equipment in LoS and NLoS con-ditions (ii) the life-distance of LoS and NLoS channels (iii)the LoS existence probability per carrier for a UE (iv) the LoSexistence probability per carrier for a NB and (v) powervariation at LoS-NLoS transition and the transition durationThesemodels are useful for the design and performance eval-uation of communication techniques adaptive to the LoS andNLoS scenarios

The rest of this paper is organized as follows Section 2introduces the measurement equipment environments andthe specifications applied for data acquisition in an in-serviceUMTS network The main steps of processing the mea-surement data to obtain the estimates of channel impulseresponses are also described In Section 3 the LoSNLoSdetection method is introduced Section 4 elaborates detailsof the models established based on the LoSNLoS channelcharacteristics obtained Finally conclusive remarks areaddressed in Section 5

2 Measurement Equipment CampaignsSetup and Postprocessing

Figure 1 illustrates the diagram of the data acquisition pro-cessing and modeling procedure adopted for characterizingLoSNLoS channels based on the signals of a UMTS networkThe equipment used to acquire the signals consists of thefollowing components a Universal Software-Defined RadioPeripheral (USRP) device of type N210 [41] controlled by theGNU radio software in a computer and a global positioning









individually




(0 ) = arg max

1199050 119894

100381610038161003816100381610038161003816100381610038161003816

int

119879

0

119903119888(119905) (119888119894

119904(119905 minus 1199050))lowast

119889119905

100381610038161003816100381610038161003816100381610038161003816

2

119888 = 1 2 3

(1)




119895 = arg max119895isin119869119894

100381610038161003816100381610038161003816100381610038161003816

int

0+119879

0

119903119888(119905) 119904lowast

119895(119905 minus 0) 119889119905

100381610038161003816100381610038161003816100381610038161003816

2

119888 = 1 2 3 (2)





119888119896(120591)


ℎ119888119896(120591) = int

0+119879

0

119903119888(119905 + 119896119879) 119904

lowast

119895(119905 minus 120591) 119889119905 (3)




0

05

1

15

2

25

3

35

times10minus8

Delay (s)

Pow

er




3 LoSNLoS Detection


0 5 10 15 20 25 300

005

01

015

02

025

03

035


120588th

LoSNLoS

Prob

abili

ty d

ensit

y fu

nctio

n






65 131 196 261 326 392 457 522 587 653

336598131163196228261


x (m)

y (m

)


65 131 196 261 326 392 457 522 587 653

336598131163196228261

x (m)

y (m

)











119903= 119875119905minus 119875119901+ 119875119904+ 119875119898

where 119875119905 119875119901 119875119904 and 119875




01

02

03

04

05

06

07

08

09

1


Cum

ulat

ive d

istrib

utio

n fu

nctio

n












05 1 15 2 25 3 35 4 45 5 55

01

02

03

04

05

06

07

08

09

1

Cum

ulat

ive d

istrib

utio

n fu

nctio

n












05 1 15 2 25 3 35

01

02

03

04

05

06

07

08

09

1


Cum

ulat

ive d

istrib

utio

n fu

nctio

n





119901depends






minus119875119901 119875119904 and 119875



119901by the








119875119901= minus10120574 log

10(119889

1198890

) + 20 log10(

120582

41205871198890

) (4)


40 50 60 70 80 90 100 110 120 130 140 1500

001

002

003

004

005

006

007

008

Path loss (dB)

Estim

ated

pro

babi

lity

dens

ity fu

nctio

n

LoSNLoS



0is between 10 and




119905= 0 dB the








0 500 1000 1500 2000 2500 3000 3500 40000

001

002

003

004

005

006

007

008

Estim

ated

pro

babi

lity

dens

ity fu

nctio

n


LoSNLoS


38 76 115 153 191 229 268 306

38

96

154

x (m)

y(m

)







08 1 12 14 16 18 2 22 24 2604

05

06

07

08

09

1

LoSNLoS

282m 160m

5m

10m


Empi

rical

cum

ulat

ive d

istrib

utio

n fu

nctio

n




0 05 10

005

01

015

02

025

03

035

04

045

05


The l

ocat

ions

()




02 04 06 08 10

005

01

015

02

025

03

035

04

045


The N

Bs (

)





minus4 minus2 0 2 4 6 8 100

005

01

015

02

025

Prob

abili

ty d

istrib

utio

n fu

nctio

n







0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0

10

20

0

10

20

0

10

5

15

Threshold

Threshold

Threshold

Threshold





Type (i) transition




10

5

15

20

ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)





001 003 004 0050

01

02

03

04

05

06

07

08

09


Empi

rical

occ

urre

nce f

requ

ency



5 Conclusions





Acknowledgments


References


















































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation

















individually




(0 ) = arg max

1199050 119894

100381610038161003816100381610038161003816100381610038161003816

int

119879

0

119903119888(119905) (119888119894

119904(119905 minus 1199050))lowast

119889119905

100381610038161003816100381610038161003816100381610038161003816

2

119888 = 1 2 3

(1)




119895 = arg max119895isin119869119894

100381610038161003816100381610038161003816100381610038161003816

int

0+119879

0

119903119888(119905) 119904lowast

119895(119905 minus 0) 119889119905

100381610038161003816100381610038161003816100381610038161003816

2

119888 = 1 2 3 (2)





119888119896(120591)


ℎ119888119896(120591) = int

0+119879

0

119903119888(119905 + 119896119879) 119904

lowast

119895(119905 minus 120591) 119889119905 (3)




0

05

1

15

2

25

3

35

times10minus8

Delay (s)

Pow

er




3 LoSNLoS Detection


0 5 10 15 20 25 300

005

01

015

02

025

03

035


120588th

LoSNLoS

Prob

abili

ty d

ensit

y fu

nctio

n






65 131 196 261 326 392 457 522 587 653

336598131163196228261


x (m)

y (m

)


65 131 196 261 326 392 457 522 587 653

336598131163196228261

x (m)

y (m

)











119903= 119875119905minus 119875119901+ 119875119904+ 119875119898

where 119875119905 119875119901 119875119904 and 119875




01

02

03

04

05

06

07

08

09

1


Cum

ulat

ive d

istrib

utio

n fu

nctio

n












05 1 15 2 25 3 35 4 45 5 55

01

02

03

04

05

06

07

08

09

1

Cum

ulat

ive d

istrib

utio

n fu

nctio

n












05 1 15 2 25 3 35

01

02

03

04

05

06

07

08

09

1


Cum

ulat

ive d

istrib

utio

n fu

nctio

n





119901depends






minus119875119901 119875119904 and 119875



119901by the








119875119901= minus10120574 log

10(119889

1198890

) + 20 log10(

120582

41205871198890

) (4)


40 50 60 70 80 90 100 110 120 130 140 1500

001

002

003

004

005

006

007

008

Path loss (dB)

Estim

ated

pro

babi

lity

dens

ity fu

nctio

n

LoSNLoS



0is between 10 and




119905= 0 dB the








0 500 1000 1500 2000 2500 3000 3500 40000

001

002

003

004

005

006

007

008

Estim

ated

pro

babi

lity

dens

ity fu

nctio

n


LoSNLoS


38 76 115 153 191 229 268 306

38

96

154

x (m)

y(m

)







08 1 12 14 16 18 2 22 24 2604

05

06

07

08

09

1

LoSNLoS

282m 160m

5m

10m


Empi

rical

cum

ulat

ive d

istrib

utio

n fu

nctio

n




0 05 10

005

01

015

02

025

03

035

04

045

05


The l

ocat

ions

()




02 04 06 08 10

005

01

015

02

025

03

035

04

045


The N

Bs (

)





minus4 minus2 0 2 4 6 8 100

005

01

015

02

025

Prob

abili

ty d

istrib

utio

n fu

nctio

n







0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0

10

20

0

10

20

0

10

5

15

Threshold

Threshold

Threshold

Threshold





Type (i) transition




10

5

15

20

ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)





001 003 004 0050

01

02

03

04

05

06

07

08

09


Empi

rical

occ

urre

nce f

requ

ency



5 Conclusions





Acknowledgments


References


















































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation











0

05

1

15

2

25

3

35

times10minus8

Delay (s)

Pow

er




3 LoSNLoS Detection


0 5 10 15 20 25 300

005

01

015

02

025

03

035


120588th

LoSNLoS

Prob

abili

ty d

ensit

y fu

nctio

n






65 131 196 261 326 392 457 522 587 653

336598131163196228261


x (m)

y (m

)


65 131 196 261 326 392 457 522 587 653

336598131163196228261

x (m)

y (m

)











119903= 119875119905minus 119875119901+ 119875119904+ 119875119898

where 119875119905 119875119901 119875119904 and 119875




01

02

03

04

05

06

07

08

09

1


Cum

ulat

ive d

istrib

utio

n fu

nctio

n












05 1 15 2 25 3 35 4 45 5 55

01

02

03

04

05

06

07

08

09

1

Cum

ulat

ive d

istrib

utio

n fu

nctio

n












05 1 15 2 25 3 35

01

02

03

04

05

06

07

08

09

1


Cum

ulat

ive d

istrib

utio

n fu

nctio

n





119901depends






minus119875119901 119875119904 and 119875



119901by the








119875119901= minus10120574 log

10(119889

1198890

) + 20 log10(

120582

41205871198890

) (4)


40 50 60 70 80 90 100 110 120 130 140 1500

001

002

003

004

005

006

007

008

Path loss (dB)

Estim

ated

pro

babi

lity

dens

ity fu

nctio

n

LoSNLoS



0is between 10 and




119905= 0 dB the








0 500 1000 1500 2000 2500 3000 3500 40000

001

002

003

004

005

006

007

008

Estim

ated

pro

babi

lity

dens

ity fu

nctio

n


LoSNLoS


38 76 115 153 191 229 268 306

38

96

154

x (m)

y(m

)







08 1 12 14 16 18 2 22 24 2604

05

06

07

08

09

1

LoSNLoS

282m 160m

5m

10m


Empi

rical

cum

ulat

ive d

istrib

utio

n fu

nctio

n




0 05 10

005

01

015

02

025

03

035

04

045

05


The l

ocat

ions

()




02 04 06 08 10

005

01

015

02

025

03

035

04

045


The N

Bs (

)





minus4 minus2 0 2 4 6 8 100

005

01

015

02

025

Prob

abili

ty d

istrib

utio

n fu

nctio

n







0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0

10

20

0

10

20

0

10

5

15

Threshold

Threshold

Threshold

Threshold





Type (i) transition




10

5

15

20

ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)





001 003 004 0050

01

02

03

04

05

06

07

08

09


Empi

rical

occ

urre

nce f

requ

ency



5 Conclusions





Acknowledgments


References


















































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










65 131 196 261 326 392 457 522 587 653

336598131163196228261


x (m)

y (m

)


65 131 196 261 326 392 457 522 587 653

336598131163196228261

x (m)

y (m

)











119903= 119875119905minus 119875119901+ 119875119904+ 119875119898

where 119875119905 119875119901 119875119904 and 119875




01

02

03

04

05

06

07

08

09

1


Cum

ulat

ive d

istrib

utio

n fu

nctio

n












05 1 15 2 25 3 35 4 45 5 55

01

02

03

04

05

06

07

08

09

1

Cum

ulat

ive d

istrib

utio

n fu

nctio

n












05 1 15 2 25 3 35

01

02

03

04

05

06

07

08

09

1


Cum

ulat

ive d

istrib

utio

n fu

nctio

n





119901depends






minus119875119901 119875119904 and 119875



119901by the








119875119901= minus10120574 log

10(119889

1198890

) + 20 log10(

120582

41205871198890

) (4)


40 50 60 70 80 90 100 110 120 130 140 1500

001

002

003

004

005

006

007

008

Path loss (dB)

Estim

ated

pro

babi

lity

dens

ity fu

nctio

n

LoSNLoS



0is between 10 and




119905= 0 dB the








0 500 1000 1500 2000 2500 3000 3500 40000

001

002

003

004

005

006

007

008

Estim

ated

pro

babi

lity

dens

ity fu

nctio

n


LoSNLoS


38 76 115 153 191 229 268 306

38

96

154

x (m)

y(m

)







08 1 12 14 16 18 2 22 24 2604

05

06

07

08

09

1

LoSNLoS

282m 160m

5m

10m


Empi

rical

cum

ulat

ive d

istrib

utio

n fu

nctio

n




0 05 10

005

01

015

02

025

03

035

04

045

05


The l

ocat

ions

()




02 04 06 08 10

005

01

015

02

025

03

035

04

045


The N

Bs (

)





minus4 minus2 0 2 4 6 8 100

005

01

015

02

025

Prob

abili

ty d

istrib

utio

n fu

nctio

n







0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0

10

20

0

10

20

0

10

5

15

Threshold

Threshold

Threshold

Threshold





Type (i) transition




10

5

15

20

ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)





001 003 004 0050

01

02

03

04

05

06

07

08

09


Empi

rical

occ

urre

nce f

requ

ency



5 Conclusions





Acknowledgments


References


















































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










05 1 15 2 25 3 35 4 45 5 55

01

02

03

04

05

06

07

08

09

1

Cum

ulat

ive d

istrib

utio

n fu

nctio

n












05 1 15 2 25 3 35

01

02

03

04

05

06

07

08

09

1


Cum

ulat

ive d

istrib

utio

n fu

nctio

n





119901depends






minus119875119901 119875119904 and 119875



119901by the








119875119901= minus10120574 log

10(119889

1198890

) + 20 log10(

120582

41205871198890

) (4)


40 50 60 70 80 90 100 110 120 130 140 1500

001

002

003

004

005

006

007

008

Path loss (dB)

Estim

ated

pro

babi

lity

dens

ity fu

nctio

n

LoSNLoS



0is between 10 and




119905= 0 dB the








0 500 1000 1500 2000 2500 3000 3500 40000

001

002

003

004

005

006

007

008

Estim

ated

pro

babi

lity

dens

ity fu

nctio

n


LoSNLoS


38 76 115 153 191 229 268 306

38

96

154

x (m)

y(m

)







08 1 12 14 16 18 2 22 24 2604

05

06

07

08

09

1

LoSNLoS

282m 160m

5m

10m


Empi

rical

cum

ulat

ive d

istrib

utio

n fu

nctio

n




0 05 10

005

01

015

02

025

03

035

04

045

05


The l

ocat

ions

()




02 04 06 08 10

005

01

015

02

025

03

035

04

045


The N

Bs (

)





minus4 minus2 0 2 4 6 8 100

005

01

015

02

025

Prob

abili

ty d

istrib

utio

n fu

nctio

n







0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0

10

20

0

10

20

0

10

5

15

Threshold

Threshold

Threshold

Threshold





Type (i) transition




10

5

15

20

ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)





001 003 004 0050

01

02

03

04

05

06

07

08

09


Empi

rical

occ

urre

nce f

requ

ency



5 Conclusions





Acknowledgments


References


















































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










40 50 60 70 80 90 100 110 120 130 140 1500

001

002

003

004

005

006

007

008

Path loss (dB)

Estim

ated

pro

babi

lity

dens

ity fu

nctio

n

LoSNLoS



0is between 10 and




119905= 0 dB the








0 500 1000 1500 2000 2500 3000 3500 40000

001

002

003

004

005

006

007

008

Estim

ated

pro

babi

lity

dens

ity fu

nctio

n


LoSNLoS


38 76 115 153 191 229 268 306

38

96

154

x (m)

y(m

)







08 1 12 14 16 18 2 22 24 2604

05

06

07

08

09

1

LoSNLoS

282m 160m

5m

10m


Empi

rical

cum

ulat

ive d

istrib

utio

n fu

nctio

n




0 05 10

005

01

015

02

025

03

035

04

045

05


The l

ocat

ions

()




02 04 06 08 10

005

01

015

02

025

03

035

04

045


The N

Bs (

)





minus4 minus2 0 2 4 6 8 100

005

01

015

02

025

Prob

abili

ty d

istrib

utio

n fu

nctio

n







0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0

10

20

0

10

20

0

10

5

15

Threshold

Threshold

Threshold

Threshold





Type (i) transition




10

5

15

20

ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)





001 003 004 0050

01

02

03

04

05

06

07

08

09


Empi

rical

occ

urre

nce f

requ

ency



5 Conclusions





Acknowledgments


References


















































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










08 1 12 14 16 18 2 22 24 2604

05

06

07

08

09

1

LoSNLoS

282m 160m

5m

10m


Empi

rical

cum

ulat

ive d

istrib

utio

n fu

nctio

n




0 05 10

005

01

015

02

025

03

035

04

045

05


The l

ocat

ions

()




02 04 06 08 10

005

01

015

02

025

03

035

04

045


The N

Bs (

)





minus4 minus2 0 2 4 6 8 100

005

01

015

02

025

Prob

abili

ty d

istrib

utio

n fu

nctio

n







0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0

10

20

0

10

20

0

10

5

15

Threshold

Threshold

Threshold

Threshold





Type (i) transition




10

5

15

20

ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)





001 003 004 0050

01

02

03

04

05

06

07

08

09


Empi

rical

occ

urre

nce f

requ

ency



5 Conclusions





Acknowledgments


References


















































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










minus4 minus2 0 2 4 6 8 100

005

01

015

02

025

Prob

abili

ty d

istrib

utio

n fu

nctio

n







0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0 001 002 003 004 005 006 007

0

10

20

0

10

20

0

10

5

15

Threshold

Threshold

Threshold

Threshold





Type (i) transition




10

5

15

20

ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)ΔP

LoSminus

NLo

S(d

B)





001 003 004 0050

01

02

03

04

05

06

07

08

09


Empi

rical

occ

urre

nce f

requ

ency



5 Conclusions





Acknowledgments


References


















































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










001 003 004 0050

01

02

03

04

05

06

07

08

09


Empi

rical

occ

urre

nce f

requ

ency



5 Conclusions





Acknowledgments


References


















































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation



















































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation









research article measurement-based los/nlos channel

Documents