An Empirical Ultra Wideband Channel Model for Indoor Laboratory Environments

68 N. NOORI, R. KARIMZADEH-BAEE, A. ABOLGHASEMI, AN EMPIRICAL ULTRA WIDEBAND CHANNEL MODEL FOR INDOOR An Empirica Utra Wideband Channe Mode for Indoor Laboratory Environments Narges NOORI, Roghieh KARIMZADEH-BAEE, Ai ABOLGHASEMI Iran Teecommunication Research Center, PO Box 455-396, End of North Kaargar St., Tehran, Iran {nnoori, rbaee, ai_ab}@itrc.ac.ir Abstract. Channe measurement and modeing is an important issue when designing utra wideband (UWB) communication systems. In this paper, the resuts of some UWB time-domain propagation measurements performed in modern aboratory (Lab) environments are presented. The Labs are equipped with many eectronic and measurement devices which mae them different from other indoor ocations ie office and residentia environments. The measurements have been performed for both ine of sight (LOS) and non-los (NLOS) scenarios. The measurement resuts are used to investigate arge-scae channe characteristics and tempora dispersion parameters. The custering Saeh- Vaenzuea (S-V) channe impuse response (CIR) parameters are investigated based on the measurement data. The sma-scae ampitude fading statistics are aso studied in the environment. Then, an empirica mode is presented for UWB signa transmission in the Lab environment based on the obtained resuts. Keywords Utra wideband measurement, propagation channe, path oss, tempora dispersion parameters, custering channe mode, sma-scae ampitude fading.. Introduction The use of utra wideband (UWB) technoogy has recenty attracted the intense interest of wireess system designers due to its potentia in short range high data rate communications [], []. According to the modern definition, any signa with a bandwidth wider than 5 MHz or a fractiona bandwidth greater than. can be considered as an UWB signa. A possibe technique for impementing UWB is impuse radio, which is based on transmitting extremey short (in the order of nanoseconds) ow power puses [3]. In a wireess system, the transmitted signa interacts with the physica environment in a compex manner. Therefore, the abiity to accuratey predict propagation characteristics of UWB signas is crucia to system design. A reaistic channe mode can be hepfu for better understanding of the propagation mechanisms and effects. Consequenty, UWB channe modeing in different environments has been a subject of intense research. Due to the broad variation of the mutipath propagation channe, statistica approaches are the best way to mode channe properties. The resuts of some UWB time or frequency-domain channe measurements at home, office and hospita environments have been presented in [4]-[7] for different frequency bands and transmitter-receiver separation distances. In [8], a genera mode has been presented for frequency range of 3.-.6 GHz in a number of different environments (residentia, office, outdoor ine of sight (LOS), outdoor non-los (NLOS), farm, industria and body area networ) as we as for the frequency range beow GHz in office environment. This mode has been presented based on the measurement and simuation of the authors and other papers in the open iterature. But, some of the measurements have not covered the fu frequency range. As it has been mentioned, the ist of the measured environments presented in [8] is not compete and other environments can be important for specific appications. Effective design and impementation of an UWB networ in the aboratory (Lab) environments needs an accurate channe mode. Actuay, most of the Lab environments are equipped with many eectronic and measurement devices which are ocated on tabes. The most of these devices are made of metaic materias. Therefore, the UWB propagation mechanism in the Lab may be different from other indoor ocations ie office and residentia environments, where the most of the furniture are made from wood and textie products. In the present wor, the resuts of utra wideband time-domain measurements in the aboratory (Lab) environments are investigated to present an empirica UWB channe mode. These measurements have been performed for both LOS and NLOS scenarios. In the foowing sections, first, the UWB time-domain measurement setup, ocation and procedure are described in detai. Then, the measurement data is anayzed to extract path oss and shadowing fading characteristics and tempora dispersion parameters of the channe for both measurement scenarios. The custering Saeh-Vaenzuea (S-V) channe impuse response (CIR) parameters are investigated based on the measurement data at both LOS and NLOS scenarios. Then, the sma-scae ampitude fading statistics are extracted in the Lab environment for both measurement scenarios.

RADIOENGINEERING, VOL. 8, NO., APRIL 9 69. UWB Time-Domain Measurements. Measurement Setup A diagram of the time-domain measurement setup is shown in Fig.. At the transmitter side, a puse generator was used as an UWB signa source. The puse generator produces an utra-fast puse with 5 ps fa time. If an impuse forming networ is attached to the puse head output, the width of the output waveform of puse generator wi be ess than 5 ps. The resuted impuse wave form is shown in Fig.. The generator was connected to the transmitting antenna through a ow oss wideband cabe. The output signa of the receiving antenna was ampified by a ow noise ampifier with a gain of 8 db and 3 db bandwidth of GHz. A digita samping oscioscope was used at the receiver side which samped the received signa at a rate of sampe per.5 ps. This oscioscope has a bandwidth up to 7 GHz. The puse generator and digita samping oscioscope were synchronized through a reference coc signa at a frequency of Hz. Measurements were performed by a pair of -8 GHz douberidged waveguide horn antennas. These transmitting and receiving antennas were both paced on moving carts at a height of 35 cm above ground. It shoud be noted that since the transmitting and receiving antennas are incuded as parts of the UWB channe (as shown in Fig. ), the UWB time domain measurements and presented channe mode are antenna dependent. Ampitude (V) Fig.. A diagram of the used time-domain measurement setup..5 -.5 - -.5 - -.5 3 4 5 6 7 8 9 Time (ps) Fig.. Output waveform of puse generator.. Measurement Location The time-domain measurement campaign was conducted for both LOS and NLOS scenarios at a basement foor with a pan shown in Fig. 3. A the main rooms of this foor are modern Labs. The buiding was are made of bric with metaic stud. The partitions are auminum frame structured with fabric, wood and gass surface. The foor of the rooms is covered with ties. The doors are made of wood and have metaic frames. The furniture inside each room consists of many different eectronic and measurement devices, metaic and wooden cupboards and cabinets, tabes made of wood, mid bac wor chairs, computers, etc..3 Measurement Procedure To perform the measurements, five different transmitter ocations were considered. The receiver points were chosen at those ocations where the received signa coud be ceary detected. The measurements were coected at each receiver ocation by moving the receiver antenna over a square grid of 9 points spaced 5 cm apart as shown in Fig. 3. In order to cance out the noise, measurements were averaged at each measurement point. A system caibration was made to compensate any imperfection of the system components. Then, any dc offset that had not been taen into account by the caibration was removed. Fig. 3. Pan of the measurement environment with different Tx and Rx ocations. 3. Measurement Resut Anaysis 3. Path Loss and Large-Scae Anaysis Large-scae fading can be categorized by path oss and shadowing. To investigate UWB arge-scae fading behavior, the distance dependency of the path oss which

7 N. NOORI, R. KARIMZADEH-BAEE, A. ABOLGHASEMI, AN EMPIRICAL ULTRA WIDEBAND CHANNEL MODEL FOR INDOOR can describe attenuation of the median power as a function of distance, is introduced as [3]: d PL( d) = PL( d ) + γ og + d d d χ () where PL(d) represents the path oss at a distance d, PL(d ) is the path oss at reference distance d, γ is the path oss exponent and χ is the shadowing fading parameter. This parameter shows that path oss observed at any given point wi deviate from its average vaue. The vaues of PL(d ) and γ can be obtained through a east square inear fit to the measured data. The reference distance is set to.5 m and 3.4 m for LOS and NLOS path oss mode, respectivey where these vaues are the minimum observation distances in each scenario. Thus, the path oss modes are vaid for distances above these reference vaues. The extracted parameters are PL(d )= PL(.5 m)=8.7 db and γ =.8 for LOS scenario and PL(d )= PL(3.4 m)=39.6 db and γ =3.45 for NLOS scenario. The scatter pots of the path oss versus the transmitter-receiver separation distance are shown in Fig. 4 for both scenarios. 5 the shadowing fading parameter is found by cacuating deviation from the obtained fit. Based on our measurements, the vaues of σ χ are.6 db and 5.7 db for LOS and NLOS scenarios, respectivey. A comparison of the theoretica cumuative distribution function () of a zero-mean Gaussian random variabe with the empirica of the shadowing fading parameter is shown in Fig. 5 for both LOS and NLOS scenarios..9.8.7.6.5.4.3.. Empirica (LOS) Theoretica -4-4 6 8 χ (db) 45.9 Path Loss (db) 4 35.8.7.6.5 3 5 7 65 6 LOS Data Path Loss Exponent=.8 Transmitter-Receiver Separation Distance (m).4.3.. Empirica (NLOS) Theoretica -5 - -5 5 5 χ (db) Fig. 5. A comparison of the theoretica of a zero-mean Gaussian random variabe with empirica of the shadowing fading parameter. Path Loss (db) 55 5 45 4 35 3 NLOS Data Path Loss Exponent=3.45 Transmitter-Receiver Separation Distance (m) Fig. 4. Scatter pots of the path oss versus the transmitterreceiver separation distance. The shadowing fading parameter is a zero-mean Gaussian distributed random variabe in db with standard deviation σ χ which is aso in db. The standard deviation of 3. Tempora Dispersion Resuts In a wireess system, the signa arriving at a receiver point is in genera a summation of severa mutipath components (MPCs). The power deay profie (PDP) provides an indication of dispersion of the transmitted power over various paths. The channe tempora dispersion characteristics show tempora distribution of the power reative to the first arriving component. These characteristics are usuay quantified in terms of the mean excess deay and RMS deay spread. To obtain these parameters, the power deay profie is normaized and a signas beow a specific threshod x db reative to the maximum are forced to zero [9]. This is done to remove the noise that varies from one measurement setup to another.

RADIOENGINEERING, VOL. 8, NO., APRIL 9 7 The mean excess deay, m, is defined as the first moment of the PDP [] P( ) m =. ( ) P( ) The RMS deay spread is the square root of the second centra moment of the power deay profie [] where RMS = m ( m ) ( 3) P( ) m =. ( 4) P( ) We use a db threshod to cacuate the deay spread parameters. The resuted mean excess deay is varied between.45 ns and 7.64 ns in LOS conditions and between.75 ns and 5.73 ns in NLOS conditions. The of the RMS deay spread is potted in Fig. 6 for both measurement scenarios. This figure shows that the most probabe vaues of RMS deay spread are around.7 ns and 4.7 ns for LOS and NLOS scenarios, respectivey. The obtained parameter vaues indicate that the channe suffers from higher tempora dispersion in NLOS conditions. Probabiity RMS Deay Spread <Abscissa.9.8.7.6.5.4.3.. LOS Data NLOS Data 4 6 8 4 RMS Deay Spread (ns) Fig. 6. Cumuative distribution functions for the RMS deay spread ( db). Another important parameter which can be used to evauate tempora dispersion characteristics of the channe is the ratio of mean excess deay to RMS deay spread. The ower vaues of this ratio indicate high concentration of the power at sma vaues of the mean excess deay; whie, the higher vaues of this ratio indicate high concentration of the power at arger vaues of the mean excess deay [5, 6, 9]. The reation between the mean excess deay and RMS deay spread for both LOS and NLOS scenarios is shown in Fig. 7. The best east square inear fit to the measured data gives the vaues of.4 and.67 for this ratio in LOS and NLOS conditions, respectivey. This means that in LOS conditions, the power is more concentrated at the received components with ower time deay vaues, whie in NLOS conditions, the power is more concentrated at the received components with higher time deay vaues. Mean Excess Deay (ns) Mean Excess Deay (ns) 8 6 4 8 6 4 LOS Data Sope=.4 4 6 8 4 RMS Deay Spread (ns) 3 5 5 5 NLOS Data Sope=.67 4 6 8 RMS Deay Spread (ns) Fig. 7. Scatter pots for the mean excess deay versus RMS deay spread. 3.3 Custering Channe Mode As it is reported in most of the UWB measurement campaigns, the arriving MPCs tend to form custers in the tempora domain []-[3]. An UWB channe impuse response (CIR) which account for the custering phenomenon of MPCs has been proposed in [3] based on the conventiona Saeh-Vaenzuea (S-V) channe mode [4], where the mode parameters have been derived using measurement data coected in the frequency band of 3- GHz in various types of high-rise apartment under different propagation scenarios. The custering CIR of the UWB channe can be expressed by S-V mode as foows [3]: L K h( t) = a, δ ( t T, ) (5) = = where δ(.) is the Dirac deta function, L is the number of custers, K is the number of MPCs within the th custer, a, is the mutipath gain coefficient of the th component in the th custer, T is the deay of the th custer which is defined as the time of arriva of the first arriving MPC within the th custer and, is the deay of the th MPC reative to the th custer arriva time, T. From (5), {L,T } and {K,,, a, } are cassified as inter-custer and intra-

7 N. NOORI, R. KARIMZADEH-BAEE, A. ABOLGHASEMI, AN EMPIRICAL ULTRA WIDEBAND CHANNEL MODEL FOR INDOOR custer parameters, respectivey [3]. The number of custers, L, is modeed by a Poisson distribution as proposed in [5]. The presence of some objects in the environment under consideration can increase the number of custers. It was found that the number of MPCs per custer, K, can be modeed by exponentia distribution. It shoud be noted that the number of MPCs per custer (thus, the number of custers) is dependent on the resoution of the parameter estimation technique, the type of the transmitting and receiving antennas, the transmitter-receiver separation distance, the physica ayout of the environment and the dynamic range of the measurement system. More custers are observed in a heaviy cuttered environment. Based on the S-V channe mode, the custer inter-arriva times and the ray intra-arriva times are described by two independent exponentia probabiity density functions as foows [4]: ( T T ) Λ Λ( T T ) p = exp[ ] > (6) p(, ( ), ) = λ exp[ λ(, ( ), )] > (7) where Λ is the mean custer arriva rate and λ is the mean ray arriva rate. The average power of both custers and rays within the custers are assumed to decay exponentiay: a T Γ, γ, = a,. e. e (8) where a, is the expected vaue of the power of the first arriving MPC, Γ is the decay exponent of the custers and γ is the decay exponent of the rays within the custers. From anaysis of the recorded measurements in the Lab environment, the average number of custers, L, is obtained equa to.3 and.9 for LOS and NLOS scenarios, respectivey. The resuted cumuative distribution functions (s) of the number of MPCs per custer, K, are shown in Fig. 8 for both scenarios. As can be seen in this figure, the s can be cosey modeed by theoretica exponentia distribution functions with the mean vaue of K, which is equa to 3.7 and 8.5 for LOS and NLOS scenarios, respectivey. It can be seen that the mean vaue of the number of MPCs per custer is increased from LOS to NLOS scenarios. In order to extract the mean custer arriva rate, Λ, the arriva time of the first MPC in each custer was considered to be the custer arriva time, regardess of whether or not it had the argest ampitude. The arriva time of each custer was subtracted from its successor. The conditiona probabiity distribution given in (6) coud be estimated by appying the east mean square fit of the custer inter-arriva time to an exponentia distribution. The resuted /Λ vaues are 7.95 ns and.49 ns for LOS and NLOS scenarios, respectivey. A simiar method shoud be carried out to estimate λ which is the average ray arriva rate within a custer. The estimated /λ vaues are.3 s and.6 ns in LOS and NLOS conditions, respectivey..9.8.7.6.5.4.3.. K =3.7 Empirica (LOS) Exponentia Fit 5 5 5 3 35 4 Number of MPC per Custer.9.8.7.6.5.4.3.. K =8.5 Empirica (NLOS) Exponentia Fit 5 5 5 3 35 4 Number of MPC per Custer Fig. 8. s of the number of MPCs per custer for LOS and NLOS scenarios. Parameter LOS scenario NLOS scenario L.3.9 K 3.7 8.5 Γ (ns). 34.6 γ (ns).3 6.3 Λ (ns) 7.95.49 λ (ns).3.6 Tab.. Parameters of the custering S-V channe mode for LOS and NLOS scenarios. The custer and ray decay exponents, Γ and γ, can be estimated by considering custers and rays with normaized ampitudes and time deays and seecting their mean decay rates. In order to estimate Γ, ampitude of the first custer arriva in each data set is set to one and its time deay is set to zero. Then, a other custers arrivas in the same data set are expressed reative to this ampitude and time. The estimates for Γ which obtained by the east mean square fit are. ns and 34.6 ns for LOS and NLOS scenarios, respectivey. Appying a simiar approach to estimate γ resuts in vaues of.3 ns and 6.3 ns for LOS and NLOS scenarios, respectivey. For better comparison of the resuted mode in tow scenarios, the obtained parameters are isted in Tab..

RADIOENGINEERING, VOL. 8, NO., APRIL 9 73 3.4 Ampitude Fading Statistics One of the fundamenta parts of the channe characterization is study of the sma-scae ampitude fading. The Rice and Rayeigh distributions can describe the ampitude fading statistics in conventiona narrowband channe modes for LOS and NLOS condition, respectivey. In UWB propagation, the wide frequency bandwidth corresponds to a high tempora resoution capabiity. Therefore, a singe path arriving at a certain deay must be resoved. In order to evauate the sma-scae ampitude fading statistics, the empirica ampitudes over sma-scae areas are cacuated. Depending on the measurement environments and scenarios reative data from taps at specific excess deays were matched to some typica theoretica distributions for ampitude fading statistics such as ognorma, Naagami, Rayeigh, Rice and Weibu distributions [3]. In [6]-[8], the ognorma distribution was obtained to give the best fit for the ampitude fading statistics, whie other measurement campaigns such as [9], [] show that the smascae ampitude fading statistics can be modeed by the Naagami distribution. In [], the Rice distribution has aso best fitted to empirica data. The measurement resuts reported in [3], [], [3] show that the sma-scae ampitude fading have a good fit with Weibu distribution in fuy furnished conference rooms, modern office buidings and high rise apartments. In order to extract sma-scae ampitude fading statistics from the measurement data in the Lab environment, empirica data of the PDPs from different measurement position are gathered and cassified into LOS and NLOS. Ampitudes smaer than db of the pea in each PDP are set to zero in order to get ony the appropriate data for anaysis. Then, data from taps at specific excess deays are coected. Each tap is assumed to contain either one resovabe path or no path. This deay tap is determined by the time resoution of the specific measurement system. In our measurement system, the deay tap width is.5 ps. Extraction of the ampitudes for each tap is carried out by coecting a vector of ampitude vaues having the same deays. It is found that these data from taps at specific excess deays can be matched to ognorma, Naagami and Weibu distributions where the parameters of these distributions i.e. the standard deviation of the ognorma distribution, m-parameter of the Naagami distribution and b- shape parameter of the Weibu distribution are a ognormay distributed random variabes [3]. Variations of these parameters as a function of excess deay are extracted from the measurement data for both LOS and NLOS scenarios. The mean vaue and standard deviation of the ognorma distribution of these three parameters are isted in Tab.. The Komogorov Smirnov (K S) and chi-square (χ ) hypothesis tests are used to eaborate the goodness-of-fit for these candidate ampitude distributions. A significance eve of 5% is used to evauate the reiabiity of the fit. Tab. 3 compares the passing rate of the K-S and χ tests for the above distributions. It is found that the Weibu distribution gives the highest passing rate in both tests. Therefore, the sma-scae ampitude fading statistics can be we-modeed by Weibu distribution for both scenarios. The s of the empirica sma-scae ampitude fading fitted to the Weibu distribution are potted in Fig. 9 at 6 ns and ns excess deays for LOS condition. Parameter σ - Lognorma σl σ σl m- Naagami mn σ mn bw b- Weibu σ bw LOS scenario.86.938.4886.895.838.36 NLOS scenario.483.86.3855.89.8553.893 Tab.. Mean and standard deviation of the parameters for candidate distributions in LOS and NLOS scenarios. Distribution LOS scenario K-S χ NLOS scenario K-S χ Lognorma 73.48 38.3 98.65 8. Naagami 69.89 64.44 35.6 36.55 Weibu 97.53 84.7 99.7 89.43 Tab. 3. Passing rate of K-S and χ hypothesis tests..9.8.7.6.5.4.3.. Empirica (6 ns) Weibu Fit (6 ns) Empirica ( ns) Weibu Fit ( ns) -8-7 -6-5 -4-3 - - Path Ampitude (db) Fig. 9. s of the empirica sma-scae ampitude fading fitted to Weibu distribution at excess deay 6 ns and ns for LOS scenario. 4. Concusions The resuts of time-domain UWB channe measurement in the Lab environment were presented. Both LOS and NLOS scenarios were considered and the path oss mode and main tempora dispersion parameters were found. The cacuated path oss exponent was.8 and 3.45 for LOS and NLOS scenarios, respectivey. The of the shadowing fading parameter was obtained. The resuted tempora dispersion parameters were aso presented. The custering CIR parameters were obtained for both LOS and NLOS scenarios and distribution functions of these parameters were investigated in the measurement environment. The sma-scae ampitude fading statistics were extracted for both LOS and NLOS conditions. It was shown that sma-scae ampitude fading can be modeed by the ognorma, Naagami and Weibu distributions but Weibu distribution has maximum passing rate of the K-S and χ tests in both scenarios.

74 N. NOORI, R. KARIMZADEH-BAEE, A. ABOLGHASEMI, AN EMPIRICAL ULTRA WIDEBAND CHANNEL MODEL FOR INDOOR Acnowedgements The authors woud ie to express their sincere thans to the staffs of the Communication Technoogy and Type Approva Lab for their contributions during the measurement campaign. This wor was supported by Iran Teecommunication Research Center under contract number 5/8844. References [] ADAMS, J. C., GREGORWICH, W., CAPOTS, L., LICCARDO, D. Utra-wideband for navigation and communications. In Proc. of IEEE Aerospace Conference,, vo., p. 785 79. [] GEZICI, S., SAHINOGLU, Z., KOBAYASHI, H., POOR, H. V. Utra-wideband impuse radio systems with mutipe puse types. IEEE J. Se. Areas Commun., 6, vo. 4, no. 4, p. 89 898. [3] ARSLAN, H., CHEN, Z. N., DI BENEDETTO, M.-G. Utra Wideband Wireess Communications, John Wiey & Sons, 6. [4] GHASSEMZADEH, S. S., JANA, R., RICE, C. W., TURIN, W., TAROKH, V. Measurement and Modeing of an Utra-Wide Bandwidth Indoor Channe. IEEE Trans. Commun., 4, vo. 5, no., p. 786 796. [5] CICCOGNANI, W., DURANTINI A., CASSIOLI, D. Time domain propagation measurements of the UWB indoor channe using PN-sequence in the FCC-compiant band 3.6 6 GHz. IEEE Trans. Antennas Propag., 5, vo. 53, no. 4, p. 54 549. [6] MUQAIBEL, A., SAFAAI-JAZI, A., ATTIYA, A., WOERNER, B., RIAD, S. Path-oss and time dispersion parameters for indoor UWB propagation. IEEE Trans. Wire. Commun., 6, vo. 5, no. 3, pp. 55 559. [7] HENTILÄ, L., TAPARUNGSSANAGORN, A., VIITTALA, H., HÄMÄLÄINEN, M. Measurement and modeing of an UWB Channe at hospita. In Proc. of IEEE Int. Conf. on Utra- Wideband, 5, p. 3 7. [8] MOLISCH, A. F., CASSIOLI, D., CHONG, C.-C., EMAMI, S., FORT, A., KANNAN, J., KAREDAL, J., KUNISCH, B., SCHANTZ, H. G., SIWIAK, K., WIN, M. Z. A. Comprehensive Standardized Mode for Utrawideband Propagation Channes. IEEE Trans. Antennas Propag., 6, vo. 54, no., p. 35 366. [9] MUQAIBEL, A. H. Characterization of Utra Wideband Communication Channes. 3, Ph.D. Dissertation, ECE Dept. at Virginia Tech. [] RAPPAPORT, T. S. Wireess Communications: Principes and Practice. nd ed. Prentice Ha,. [] RAPPAPORT, T. S. Characterization of UHF mutipath radio channes in factory buidings. IEEE Trans. Antennas Propag., 989, vo. 37, no. 8, p. 58 69. [] FOERSTER, J. R., LI, Q. UWB Channe Modeing Contribution from Inte. Inte Corporation, Hisboro, OR, USA, Techn. Report P8.5 /79 SG3a, IEEE P8.5 SG3a contribution,. [3] CHONG, C.-C., YONG, S. K. A generic statistica based UWB channe mode for high-rise apartments. IEEE Trans. Antennas Propag., 5, vo. 53, no. 8, p. 389 399. [4] SALEH, A. A. M., VALENZUELA, R. A. A statistica mode for indoor mutipath propagation. IEEE J. Se. Areas Commun., 987, vo. 5, no., p. 8 37. [5] MOLISCH, A. F., KANNAN, B., CHONG, C.-C., EMAMI, S., FORT, A., KAREDAL, J., KUNISCH, J., SCHANTZ, H., SCHUSTER U., SIWIAK, K. IEEE 8.5.4a channe mode fina report. IEEE 8.5-4-66--4a, 4. [6] HENTILÄ, L., TAPARUNGSSANAGORN, A., VIITTALA, H., HÄMÄLÄINEN, M., Measurement and modeing of an UWB Channe at hospita. In Proc. of IEEE Int. Conf. on Utra- Wideband, 5, p. 3 7. [7] FOERSTER, J. R., LI, Q. UWB Channe Modeing Contribution from Inte. IEEE P8.5-/79r-SG3a,. [8] LI Q., WONG, W. S. Measurement and anaysis of the indoor UWB channe. In Proc. of IEEE Vehicuar Technoogies Conf. (VTC 3-Fa), 3, pp. 5. [9] CASSIOLI, D., WIN, M. Z., MOLISCH, A. F. The utra-wide bandwidth indoor channe: from statistica mode to simuations. IEEE J. Se. Areas Commun.,, vo., no. 6, pp. 47 57. [] ZHU, F., WU, Z., NASSAR, C. R. Generaized fading channe mode with appication to UWB. In Proc. of IEEE Conf. UWB Systems and Technoogies,, p. 3 8. [] KUNISCH J., PAMP, J. Measurement resuts and modeing aspects for the UWB radio channe. In Proc. of IEEE Conf. UWB Systems and Technoogies,, p. 9 3. [] ALVAREZ, A., VALERA, G., LOBERIA, M., TORRES, R., GARCIA, J. L. Utrawideband channe characterization and modeing. In Proc. of Int. Worshop on Utra Wideband Systems (IWUWBS 3), 3. [3] PAGANI P., PAJUSCO, P. Experimenta assessment of the UWB channe variabiity in a dynamic indoor environment. In Proc. of IEEE Int. Symp. on Persona, Indoor and Mobie Radio Communications (PIMRC 4), 4, pp. 973 977. About Authors Narges NOORI (B. Sc. 998, M. Sc., Ph. D. 6 degrees (with honors) from Iran University of Science and Technoogy (IUST), Tehran, Iran, a in Eectrica Engg.) was with RF/Microwave and Photonics Group, Univ. of Wateroo, Ontario, Canada as a visiting schoar (4 to 5). In 5, she joined Iran Teecommunication Research Center (ITRC), Tehran, Iran, now as a research assistant professor. In 8, she received the Best Researcher Award from Ministry of Communication & Information Technoogy (ICT) of Iran. Her research interests incude radio propagation and channe modeing, broadband and utra wideband communications, and numerica methods in eectromagnetics. Roghieh KARIMZADEH BAEE (B. Sc. from Khaje Nasir A-Deen Toosi Univ. (KNTU), Tehran, Iran, in, M. Sc. from Urmia Univ., Iran, in 3, both in Commun. Engg.) joined ITRC as a Researcher in Radio Commun. Group in 3. She is currenty woring toward the Ph.D. degree at the E. and Computer Engg. Dept. of Tarbiat Modares Univ., Tehran, Iran. Her research interests incude propagation channe modeing, design and anaysis of practica antennas, phase array antenna, metamaterias and passive microwave devices. Ai ABOLGHASEMI (B. Sc. 99, M. Sc. 995, both in E. and Teecommun. Engg. from KNTU, Tehran, Iran) wored on switching and signaing in fixed and mobie communication system (997 to 7). He is now a facuty member in Radio Communication group of ITRC. His research interests incude wireess communication systems, especiay utra wideband, 3G and mobie communications.