The Effect of Human Bodies on Path Loss Model in an Indoor LOS Environment

Transcription

1 The Effet of Human odies on Path Loss Model in an Indoor LOS Environment Young-Hoon Kim and Seong-Cheol Kim Seoul Nation University INMC and SOEE Seoul, Korea bstrat This paper deals with the effet of the presene of sitting in an indoor Line-of-Sight (LOS) environment on a Ultra-Wideband (UW) hannel. To assess this, we seleted four senarios with no, 5, 10, and 20 in the room. We reated two regions in the room, one near two side walls and the other in the enter aisle. In eah senario, we measured the hannel at 24 reeiving points from a fixed transmitting point in the room. t eah reeiving point, the reeiver was moved around to 9 loal positions to obtain a loal average. In this paper, the onsidered UW hannel parameters are a frequeny-independent pathloss model and a frequeny-dependent pathloss model. We find that the pathloss exponent for the region of the enter aisle dereases as more are in the room, while the pathloss exponent for the region near the side walls inreases when there are more in the room. We also study the effet of the frequeny on the pathloss harateristi. The results suggest that the effet of the presene of on UW hannels should be onsidered when assessing the performane of UW systems. Keywords-Path loss model; Human odies; UW; Indoor; LOS. I. INTRODUCTION UW systems are ommonly defined as systems that have either more than 20% relative bandwidth or more than 500MHz absolute bandwidth. It is well nown for UW systems to have many advantages, suh as low omplexity, a low ost, resistane to severe multipath fading, and the apability of a fine time resolution. There are numerous appliations of UW systems, suh as personal area networs (PNs), sensor networs, geo-loation sensors, and emergeny ommuniations. UW systems using a relatively large bandwidth have to use low power so as not to ause interferene in the neighboring ommuniation systems. This feature maes UW systems deployable in LOS and wea non-line-of-sight (NLOS) environments in whih the signal undergoes less attenuation. lassroom is an example of a LOS environment, and furniture (e.g., dess and hairs) as well as the in the lassroom are fators that hange the UW hannel. In previous wor, there were three types of hannel variation by : 1) the depth and duration of shadow fading due to pedestrians moving in the viinity of suh lins [1-3], 2) the effet of the presene of humans on wireless personal area networs (WPNs) in whih one end of the lin is loated either lose to or on a person [4-7], and 3) the effet of the presene of a human on wireless body area networs (WNs) in whih both ends of the lin are loated either lose to or on a person [8-10]. The above-mentioned papers deal with hannel variation by only a person. nother reent paper [11] assesses the hannel variation depending on how many seats are oupied by passengers in airplane, but this environment, whih ommonly involves metal material, is different from an indoor environment in a building. This paper addresses UW hannel variation depending on the presene of sitting in an indoor LOS environment. From the measurement data, we obtain a frequeny-independent pathloss model and a frequenydependent pathloss model. The paper introdues a frequeny-dependent pathloss model, whih inludes the effet of the frequeny on the pathloss harateristi. This an be useful for multiband orthogonal frequeny-division multiplexing (M-OFDM), as it divides the entire frequeny band into several sub-bands with a bandwidth of 528 MHz. This paper is organized as follows: Setion II presents the hannel measurement system and the measurement senario. In Setion III, we desribe the hannel parameters, i.e., the frequeny-independent pathloss and frequeny-dependent pathloss. Finally, the paper is ended with a summary and onlusion in Setion IV. II.. Measurement System MESUREMENT METHODOLOGY Figure 1. lo diagram of the measurement system. In this paper, we measured the UW hannel using a frequeny-domain hannel sounding method for hannel haraterization with a vetor networ analyzer (VN). The VN (gilent 8719ES [12]) transmits 801 disrete tones that are uniformly spaed from 3.1 to 4.7 GHz with a frequeny interval of 2 MHz, requiring 400 ms for one sweep. This 152

2 frequeny interval allows us to measure a multipath with a maximum exess delay of 500 ns, and the bandwidth of 1.6 GHz gives a time resolution of less than 0.01 ns. The measurement system is desribed in Fig. 1. The same dipole antennas with a gain of 2 di are used on both the transmitting and reeiving sides and are loated on 1.5-mhigh tripods. power amplifier (P) with a gain of 25 d and a low-noise amplifier (LN) with a gain of 27 d are used on the transmitting and reeiving sides, respetively. To eliminate the effet of the antennas, the P, the LN and the ables, all measured data are alibrated in an anehoi hamber.. Measurement Senario To analyze the effet of the presene of on a UW hannel, we use four senarios: 1) a room with dess and hairs, 2) a room with dess, hairs and five, 3) a room with dess, hairs and 10, 4) a room with dess, hairs and 20. We divide a room into two regions, the first near both side walls,, and the seond in the enter aisle,, as desribed in Fig. 2. In Fig. 2, 'Oupied' denotes sit on the hair; otherwise this is 'Unoupied' and the signs of the des and hair is given. ll senarios are arried out in the same room of whih the wall material is ommonly onrete and where there are two large glass windows on one side wall. The loations of the transmitting and reeiving antennas are idential in all senarios, as illustrated in Fig. 2. t eah of 24 loations in the room for eah senario, the hannel responses are measured at nine loal points arranged in a 3x3 square grid, as shown in Fig frequeny responses were olleted at eah loal position. III. PTH LOSS MODEL. Frequeny-independent Pathloss Model In the onventional narrowband system, the pathloss model needs to alulate the lin budget of the system and to minimize the interferene in the neighboring systems. However, the UW system is expeted to require a pathloss model that aounts for the frequeny omponent due to its muh wider bandwidth than a onventional system, as disussed in Setion III.. First, in this setion, we utilize the frequeny-independent pathloss model using equation (1) to show the effet of the presene of on the pathloss alulated with only the distane, PL ( d) PL ( d ) 10n log ( d / d ) S (1) d d where PL ( d ) is the pathloss at the referene distane d d 0 0 (whih is 1m in this paper), d is the separation between the transmitter and the reeiver, n is the pathloss exponent, and S is related to the degree of large-sale fading with a zeromean Gaussian distributed random variable (in d) with a standard deviation of S (also in deibels) [13]. PL ( d ), d 0 and in (1) are averaged over a 1.6 GHz bandwidth, and S () (d) Figure 2. Floor plans and reeiver loations in a room with dess and hairs, a room with dess, hairs and five () a room with dess, hairs and 10, and (d) a room with dess, hairs and

3 n is omputed using the minimum mean square error algorithm. Fig. 3 shows the path loss and their linear regression model in and. The parameters of the regression model are summarized in Table I. Figure 3. Path loss model of and Table I shows the variations of the pathloss exponent with the different senarios in and. efore analyzing the effet of on the pathloss exponent, for two regions with no, in whih the dess and hairs are loated, has a smaller pathloss exponent than, whih orresponds to the enter aisle. This differene is due to the dess and hairs. For, orresponding to the area near both of the side walls, the pathloss exponent inreases when there are more in the room. However, it dereases for in the middle of the room. This differene is aused by the differene in the main propagation mehanism of the reeived signal: In, as the number of inreases, the number of positions in whih the diret path is Senario no TLE I. n EMPIRICL PTHLOSS PRMETERS PL(d0) [d] s [d bloed inreases. In addition, some refleted paths through both side walls and the eiling are affeted by the existing, and some diffration paths exist around them [14]. In ontrast, in, the diret path is not affeted by beause there are LOS paths in all positions, but some refleted paths are bloed or attenuated by. This different propagation mehanism results in different tendenies of the pathloss exponent in and.. Frequeny-dependent Pathloss Model In the onventional narrowband system, it is suffiient to represent the pathloss model in formula (1) in setion, but the UW system requires the pathloss model onsidering the frequeny dependeny due to its wide bandwidth. For this reason, the frequeny-dependent pathloss property has been disussed in many studies [15-19]. In this paper, we analyze the frequeny-dependent pathloss property through the frequeny-dependent pathloss model as introdued by Jinwon Choi et al. [15]. The frequeny-dependent pathloss is expressed using the following modified expression of (2) PL ( d, f ) PL ( d ) 10 n( f )log ( d / d ) S (2) d d where n( f ) is the frequeny-dependent pathloss exponent. The other omponents are idential to those in (1). For UW systems in partiular, the M-OFDM sheme proposes that the assigned frequeny bands should be divided into sub-bands having bandwidths of 528 MHz [20]. Figure 4. Twelve sub-bands with bandwidths of 500 MHz To design an effiient M-OFDM system, the effet of the frequeny on UW signals with different frequeny bands should be haraterized. For this, we obtained a pratial pathloss exponent formula to express the loss as a funtion of the frequeny. The pathloss exponent variation with the frequeny is haraterized by taing the average of the pathloss exponent over a 500-MHz overlapped window 154

4 bandwidth whose enter frequeny is inremented from 3.35 to 4.45 GHz in steps of 100 MHz, as shown in Fig. 4. In this model, the pathloss exponent averaged over eah sub-band is denoted as ns at the th sub-band for = 1, 2,..., 12. The variation of ns is expressed as a funtion of the enter frequeny of eah sub-band. TLE II. LINER REGRESSION COEFFICIENTS OF (3) Senarios a a b No Figure 5. ns and their linear regression models in and The measured ns ' values are illustrated in Fig. 5. In near the side walls, ns ' values of the senario with 5 are similar to those of the senario with no, but ns inreases in the senarios with 10 and 20. This is beause the bloed multipath inreases as more are in the region. In whih is the enter aisle of the room, ns ' values of the senario with 5 are larger than those of the senario with no. This situation arises beause the loations whih are near to the transmitter have less bloed multipath by, but the loations far from the transmitter have more bloed multipath. ut ns ' values of the senarios with 10 and 20 are smaller than the other senarios beause the differene of the path loss with the distane. s shown in Fig. 5, ns inreases with the enter frequeny of the subband and it an be regressed as a linear funtion. linear regression model of the pathloss exponent with the frequeny is obtained as follows: where ns( f ) a f b (3) f is the enter frequeny of the sub-band (in gigahertz). The statistial representatives of ns 'values and the linear regression oeffiients a and b of (3) are shown in Table II. In both regions, the pathloss exponents inrease as the frequeny inreases, but shows faster growth of the pathloss exponents than for the same senario. These differenes between and are the result of different main propagation mehanisms, of whih the main differene is that the diret path is bloed by the in, whereas this is not the ase in. In addition, the slope of the pathloss exponents, a, inreases as more are in the room in the two regions. This means that the frequeny response is affeted by the presene of in the room. IV. CONCLUSION To show how presene of affets on a UW hannel in an indoor LOS environment, we seleted four senarios in the same room. The frequeny-domain hannel sounding method was used for hannel haraterization from 3.1 to 4.7 GHz. From the results, the presene of auses hannel variation. For the frequeny-independent pathloss model, the pathloss exponent for the region of the enter aisle dereases as more are in the room, while the pathloss exponent for the region near both side walls inreases as the number of in the room inreases. For the frequeny-dependent pathloss model that is useful for the 155

5 M-OFMD sheme in whih the entire frequeny band is divided into several sub-bands with a bandwidth of 528 MHz, the pathloss exponent inreases as the frequeny inreases, but the region of the enter aisle in the room experienes faster growth than the region near both side walls. In summary, this paper shows that the presene of substantially affets radio-wave propagation in an indoor LOS environment and should be onsidered when haraterizing the performane of UW systems. This finding will be helpful to those who want to validate the results of software simulations of radio-wave propagation in an indoor LOS environment. CKNOWLEDGMENT This wor was supported by the National Researh Foundation of Korea (NRF) grant funded by the Korea government (MEST) (No ) REFERENCES [1] R. Ganesh and K. Pahlavan, Effets of traffi and loal movements on multipath harateristis of an indoor radio hannel, Eletron. Lett., vol. 26, no. 12, 7 Jun. 1990, pp [2] K. I. Ziri-Castro, W. G. Sanlon, and N. E. Evans, Predition of variation in MIMO hannel apaity for the populated indoor environemnt using a radar ross-setion-based pedestrian model, IEEE Trans. Wireless Commun., vol. 4, no. 3, May 2005, pp , [3] K. I. Ziri-Castro, N. E. Evans, and W. G. Sanlon, Propagation modeling and measurements in a populated indoor environment at 5.2 GHz, in Pro. uswireless 2006, Mar. 2006, pp [4] S. L. Cotton and W. G. Sanlon, Charaterization and modeling of the indoor radio hannel at 868 MHz for a mobile bodyworn wireless personal area networ, IEEE ntennas Wireless Propag. Lett., vol. 6, De. 2007, pp [5] T.. Welh et al., The effets of the human body on UW signal propagation in an indoor environment, IEEE J. Sel. reas ommun., vol. 20, no. 9, De. 2002,pp [6]. Kara, Human body shadowing variability in short range indoor radio lins at 3-11 GHz, Int. J. Ele., vol. 96, no.2, Feb. 2009, pp [7] J. Karedal,. J. Johansson, F. Tufvesson, and. F. Molish, Shadowing effets in MIMO hannels for personal area networs, IEEE VTC 2006 Fall, Sep. 2006, pp [8]. Fort, J. Ryaert, C. Desset, P. De Doneer, P. Wambaq, and L. Van iesen, Ultra-wideband hannel model for ommuniation around the human body, IEEE J. Sel. reas Commun., vol. 24, no. 4, pr. 2006, pp [9] S. L. Cotton and W. G. Sanlon, statistial analysis of indoor multipath fading for a narrowband wireless body area networ, in Pro. IEEE PIMRC 06, Sep. 2006, pp [10]. lomainy, Y. Hao,. Owaldally, C. G. Parini, Y. I. Nehayev, C. C. Coonstantinou, and P. S. Hall, Statistial analysis and performane evaluation for on-body radio propagation with mirostrip path antennas, IEEE Trans. ntennas Propag., vol. 55, no. 1, Jan. 2007, pp [11] Chiu S. and Mihelson D.G., " Effet of Human Presene on UW Radiowave Propagation within the Passenger Cabin of a Midsize irliner," IEEE Trans. ntennas Propag., vol. 58, no. 3, 2010, pp [12] 8719ES S-parameter Vetor Networ nalyzer, [Online]. vailable: %3epsg%3pro-pn-8719ES/s-parametervetor-networ-analyzer-135-ghz?=C&l=eng, [retrieved: Jun., 2013] [13] T.S. Rappaport, Wireless Communiations: Priniples and Pratie, 2nd ed. Englewood Cliffs, NJ: Prentie-Hall, [14] Zasowsi T., Meyer G., lthaus F., and Wittneben,.," UW signal propagation at the human head," IEEE Trnas. Miro. Theory and Tehnol., vol.54, no. 4, 2006, pp [15] Jinwon Choi, Noh-Gyoung Kang, Yu-Su Sung, Jun-Sung Kang, and Seong-Cheol Kim, "Frequeny-Dependent UW hannel harateristis in offie environments," IEEE Trans. Veh. Teh., Vol. 58, Sept. 2009, pp [16]. F. Molish, K. alarishnan, D. Cassioli, C.-C. Chong, S. Emami,. Fort, J. Karedal, J. Kunish, H. Shantz, U. Shuster, and K. Siwia, IEEE a hannel model Final report, Teh. Rep. Do. IEEE a, [17] C. C. Chong, Y. E. Kim, S. K. Yong, and S. S. Lee, Statistial haraterization of the UW propagation hannel in indoor residential environments, Wireless Commun. Mobile Comput., vol. 5, no. 5, ug. 2005, pp [18] R. C. Qiu and I. T. Lu, Wideband wireless multipath hannel modeling with path frequeny dependene, in Pro. IEEE ICC, Dallas, TX, Jun. 1996, pp [19] R. C. Qiu and I. T. Lu, Multipath resolving with frequeny dependene for broadband wireless hannel modeling, IEEE Trans. Veh. Tehnol., vol. 48, no. 1, Jan. 1999, pp [20] MO, Multiand OFDM Physial Layer Proposal for IEEE Tas Group 3a, Sep [Online]. vailable: [retrieved: Sept., 2010] 156