5GHZ WIDEBAND CHANNEL MODEL IN APARTMENT BUILDING

J inwon Choi , DY Kwak, NG Kang, Jaewon Lee*, Hakhoon, Song** and Seong-Cheol Kim School of Electrical Engineering and Computer Science, Seoul National University

Convergence Laboratory of KT*, Samsung Electronic Company** Seoul, Korea

{caesar, sckim, dykwak}@maxwell.snu.ac.kr

ABSTRACT

This paper reports the empir ical 5GHz wideband channel model in apar tment bui ld ing. The channel measurement system is based on the pseudo-noise (PN) correlat ion method. In measurements , t ransmit ter is f ixed at two different posi t ions in a house while receiver moves from the rooms of the house to those of nearby houses in each transmit ter posi t ion. From measurement resul ts , propagat ion loss and wideband channel character is t ics are analyzed. As a resul t , we found that the s ignal ref lected to neighbor ing bui ld ings proffered other c lusters . Especial ly, in cases of the receiver posi t ions where with a b ig window and large away from the transmit ter, th is phenomenon is emphasized. And transmit ter located at the b iased posi t ion, could cause the imbalance of received s ignal level in a s ingle house and the in terference to neighbor ing houses .

Key Words Wideband channel model , Wideband channel character is t ics , Measurement and model ing, ray- tracing s imulat ion and 5GHz WLAN

1. Introduction As demand for the h igh data rate communication system in mobile s i tuat ion increases, the importance of Wireless Local Area Network (WLAN) is a t tended. But, the lower frequency band is fu l ly used for exis t ing communication system, so 5GHz band is suggested for new WLAN system. In Korea, 5GHz ISM band is reserved for new WLAN

system as HYPER LAN system in Europe and Unlicensed National Information Infrastructure (U-NII) band in USA. So, some papers deal with the relat ive works for 5GHz channel modeling [1]~[2] . But , in case of apar tment bui ld ing, the most common dwell ing type in Korea, i t has not to tal ly analyzed yet . Unlike the channel model in any other dwell ing types , that of apar tment is regarded independently because so many houses in s ingle bui ld ing are located closely each other. Fur thermore, in Korea , the res idents l iv ing in apar tment general ly have more in teres t in new communication system rather than people in other dwell ing types . So, i t i s predicted that the demand for new WLAN system design in apar tment wil l be urgent . So, we focus on the 5GHz wideband channel model ing in apar tment for WLAN engineer ing. The wideband channel model is the fundamental resource for eff icient WLAN system in aspects of cel l p lanning, interference problem and performance analysis . For th is purpose, f irs t ly, we constructed channel sounding system using PN method. Then we measured the power delay prof i les while the transmit ter is located in the center of house or in a room, which WLAN wil l be frequent ly used. In each transmit ter posi t ion, receiver posi t ion is located not only in house with transmit ter but a lso in nearby houses. Propagat ion loss and channel character is t ics are analyzed via post processing. Measurement results are compared with the s imulat ion resul ts obta ined by ray- tracing tool [9] . The recentra lder of th is paper is organized as fo l lows: Sect ion 2 covers the measurement system, environment and scenar io. Then, sect ion 3 gives the resul ts and analysis of measurements . Final ly, We conclude in sect ion 4 .

475-066 1

2. Measurement system, environment and scenario

A. Measurement system

Baseband(M-sequence)

Signal Generator(E4433B)

BPFPA BPF

BPF LNA

Signal Generator(E8244A)

SamplingBoard(PC)

a) Transmitter

b) Receiver Fig.1. Block diagram of measurement system

The diagram of measurement system, based on the PN correlat ion method, is i l lustrated in Fig . 1 [3]. In baseband system, 50MHz osci l la tor generates a M-sequence, which wil l be up-conver ted to 5.8GHz by s ignal generator. Then, th is s ignal passes through the power amp (PA) l inear ly operates within 30dB, and is fo l lowed by the band pass f i l ter (BPF) with 100MHz bandwidth. In the receiver system, the received s ignal is dr iven to the BPF and amplif ied by Low Noise Amplif ier (LNA). After LNA, i t is down-conver ted to 300MHz intermediate frequency and sampled with 2GHz sampling ra te to be saved in to laptop computer. The channel impulse response is obtained by cross correlat ion between received s ignal and the or ig inal M-sequence. The antennas used in measurements are omni-direct ional with 4.6dBi gain and located in height 1 .7m. B. Measurement environment , scenar io Measurements were performed at a newly constructed apar tment bui ld ing in Suwon, South Korea . I t is a ferroconcrete bui ld ing of twenty s tor ies with three elevators . The ground plan of a house is i l lus t ra ted in Fig . 2 .

Balcony

Entrance

Bathroom

Elavator

Room 1

Room 2 Room 3

Living room

Bathroom

Fig. 2. Ground plan of single apartment house

The mater ial of wal l is concrete and wall th ickness is f rom 20cm to 30cm. There was no furni ture and glass windows are set i r regular ly. But in the house with transmit ter (central house) , a l l g lass windows in house were se t . In measurements , we assumed that the Access Point (AP) and por table terminal are operat ing together in s ingle house. The transmit ter (Tx) p lays ro le of AP and the receiver (Rx) p lays of por table terminal . To analyze the effect of t ransmit ter posi t ion, we selected two transmit ter posi t ions. The one is the case of in the center of house and the other is the case of in the room which WLAN system is most frequently used. For assuming real es tabl ishment, t ransmit ter is located near a

wal l in each case .

L

K J M N O P

Q D

C

EF G H

IB

TX 1

TX 2

A

Fig. 3. Transmitter/Receiver positions Fig .3 represents the transmit ter and receiver posi t ions in measurement scenar io . We put the t ransmit ter in two posi t ions of the central house then measure the channel response in the central house and neighbor ing 3 houses. The f is t t ransmit ter posi t ion is in center of the house (Tx1 case) . And the second t ransmit ter posi t ion is decided af ter consider ing the frequency of use (Tx2 case) . I t means that th is t ransmit ter posi t ion is in the room where WLAN system is f requent ly used. In 2 cases, wideband channels of 33 receiver posi t ions are to ta l ly analyzed.

2

3. Measurement results and analysis

A. Transmitter in center of central house (Tx 1 case)

(a) Rx A (b) Rx E

(c) Rx F (d) Rx J

Fig. 4. Power delay profiles in Tx 1 case

Fig 4. represents the power delay prof i les in Tx 1 case. To compare the re lat ive power s t rength and delay, we set the maximum received s ignal level equals to 0dBm and delay of the maximum received s ignal to 600ns in each power delay prof i le . In receiver posi t ion A, the l ine of s ight (LOS) s ignal is dominant. But , in posi t ion E also located in the central house, there are two clusters [4] . The one is f rom maximum power s ignal , passing through the wall , and the o ther is f rom the s ignal ref lected by neighbor ing bui ld ing. The supposi t ion is made based on the delay between maximum signal and second one is approximately 381ns, which is calculated from the dis tance between two bui ld ings, 57.2m. And i t is a lso suppor ted by the s imulat ion resul t using ray- tracing tool . Fig. 5 compares the s imulat ion resul ts and measurement resul ts [9]. The f irs t f igure descr ibes the s imulat ion resul t consider ing the exis tence of f ront and back bui ld ings. The second one is f rom the s imulat ion ignor ing the exis tence of two bui ldings. In compar ison, we conf irm that the neighbor ing bui ld ings cause the second clus ter.

a) b) c) Fig . 5 . Simulat ion resul ts : (a) resul t consider ing the exis tence of two bui ld ings (b) resul t ignor ing the exis tence of front bui ld ing and (c) appl icat ion resul t of (a) to measurement resul t

In posi t ion F, i t reveals that the s ignal ref lected from neighbor ing bui ld ings is s t ronger than any other s ignals including the d irect ly received s ignal , which passed through the mult i wal ls . And the th ird cluster f rom the twice-ref lected s ignal exis ts . In posi t ions J ,

Fig . 6 . Propagat ion loss in T

three c lus ters are a lso observed.

x 1 case

Fig. 6 shows the propagat ion loss calculated

se, the received powers in

-110

-90

-70

-50

-30

A B C D E F G H I J K L M N O P Q

Prop

agat

ion

loss

(dB

)

f rom the received s ignal of each receiver posi t ion. Noise threshold is set to 30dB below the s trongest s ignal in posi t ions of central house, 20dB below in another posi t ions. The reason why the threshold is d ifferent due to receiver posi t ion is the received s ignal level and the number of effect ive signals are different in each receiver posi t ion. The received power level in posi t ion B is the h ighest because i t is LOS and the neares t to the t ransmit ter. In central houposi t ions C, D, E (non-LOS, a wall between Tx and Rx) are 10~30dB below that in posi t ion A, B (LOS). Especial ly, received power level in posit ion E is low because a wooden door doesn’t exis t in d irect path between Tx and Rx, while in posi t ions C and D, a door exis ts in

3

direct path. In receiver posi t ions of o ther houses, received power level in posi t ion G, where near a large window facing to the front bui ld ing, is the h ighest though i t is barely –80dBm. In any other posi t ions, i t is about –90dBm. The delay parameters are calculated from

ble . 1 Cluster numbers in s imulat ion and

RX Simul- M e a s u r

- e m e n t RX Simul-ation

M e a s u r- e m e n t

measured data via formulas in [7 ,8] and represented in Fig . 7 . The receiver posi t ions having the high delay parameters are affected by the ref lected s ignal . Those posi t ions are with b ig window to the f ront bui lding or large d is tance f rom the t ransmit ter.

Fig . 7 . Delay parameters in Tx 1 case

Tameasurement

ation

A 1 1 J 3 3

B 1 1 K 2 2

C 1 1 L 3 2

D 2 1 M 2 2

E 2 2 N 3 2

F 3 3 O 3 2

G 3 2 P 3 3

H 3 2 Q 4 4

I 2 3

Table. 1 compares the c luster number of

parameter values are high.

ameter values

G

simulat ion resul ts and that of measurement resul ts . There is no big d ifference between s imulat ion and measurement . General ly, in case of posi t ions having many clusters , delay

Table . 2 . Grouping receiver posi t ions according to delay par

roupRx

PositionsPropagation

Characteristics Spatial

Description

ⅠA, B, In the central C, D

Dominant direct signal house

ⅡE

Ef front building, G, H,K, M

fective multipaths Window to the

ⅢF, O, P, Q

Many multipaths Delayed Max. Peak

Near the transmitter

ⅣI, J, L, N

Weak Max. Peak Many multipaths

A e way from thtransmitter

From the id ac we

iv ided receiver posi t ions in to 4 groups as

se . So, the s ignal v ia d irect

-

- ransmit ter. But , the mult ip le

- l t ip le ref lected

B osi t ion of central

ouse (Tx 2 case)

mit ter is located in a small oom, biased posi t ion of central house. The

se w eband char ter is t ics ,d

0

400

800

1200

A B C D E F G H I J K L M N O P Q

Rx position

dela

y (n

s)

shown in Table . 2 . - The receiver posi t ions in Group I are located

in the central houpath plays the dominant ro le and the effects of o ther mult ipaths are re la t ively very weak. But, in the case of posi t ion E, which is a lso located in the same house, the ref lected Mean Excess Delay RMS Delay Spread

signal f rom front bui ld ing exer ts huge inf luence on delay character is t ics . I t’s because there is a large g lass window in the room. Any other Rx posi t ions in Group I I (G, H, K, M) also have a g lass window faced the f ront bui ld ing. The Group III posi t ions are not so much away from the t

Maximum Excess Delay

walls makes the d irect s ignal weaker while the effect of ref lected s ignal f rom front build ing gets less weaken. So the ref lected s ignal is s t rongly measured. In Group IV, d irect s ignal works re lat ively s trong. And the weak musignals a lso exis t . So the maximum excess delay value is very high.

. Transmit ter in b iased ph In th is case , t ransrpower delay prof i les from measurements are i l lus t ra ted in Fig . 8 .

4

(a) Rx A (b) Rx C

(c) Rx F (d) Rx Q

F

The receiver posi t ion A, which was LOS in

Ta le . 3 Cluster Numbers in s imulat ion resul ts

RX SIM. M E A RX SIM M E A .

ig. 8. Power delay profile in Tx 2 case

Tx 1 but NLOS in Tx 2, has the second power level c luster 15dB below to the s trongest c luster. As l ike receiver posi t ion E in Tx 1 case, the second level s ignal is the ref lected one from the front bui ld ing. However, in posi t ion C in central house, the ref lected s ignal is more powerful than the direct s ignal passed through two walls . In posi t ion F of neighbor ing house, as the separat ion dis tance of the transmit ter and receiver reduces, the effect of direct s ignal increases. In posi t ion Q, the number of c lus ters grows smaller.

b

and measurement resul ts

A 2 2 J 2 2

B 1 2 K 2 1

C 2 3 L 2 2

D 3 2 M 2 2

F 1 1 N 3 2

G 1 3 O 3 2

H 2 2 P 3 2

I 3 1 Q 3 2

Table.3 designates the increased cluster

Fig. 9 reveals the r ved s ignal power

Fig. 10 . represents the delay parameters in

numbers of receiver posi t ions in central house. As the transmit ter goes away from the

receivers and hides behind the wal l , the effect of direct s ignal decreases and the effects of mult ipaths increase. In receiver posi t ions of o ther houses, no s ignif icant change in cluster number except posit ion F and I . At both posi t ions, unl ikely in Tx 1 case, the d irect s ignal p lays the dominant ro le as the t ransmit ter get c loser.

Fig . 9 . Propagat ion loss in Tx 2 case

eceidispar i ty of received s ignal in central house and the in terference to neighbor house. Because of b iased transmit ter posi t ion, the received s ignal levels in receiver posi t ions of the centra l house are lower than those of Tx 1 case. But in posi t ion F of the neighbor house, i t grows bigger s ince the transmit ter becomes closer. In o ther posit ions, received s ignal powers show a s l ight decl ine .

Fig. 10. Delay parameters in Tx 2 case

Tx 2 case. Delay parameters are l ikely increased in A, B, C, D, G posi t ions in which ref lected s ignals involved more than Tx 1 case. But in posi t ions F and I where direct wall penetrat ion s ignal works more br i l l iant ly, delay parameters are decreased. In o ther

-50

-130

-110

-90

-70

A B C D E F G H I J K L M N O P Q

Prop

agat

ion

loss

(dB

)

1250

0

250

500

750

1000

A B C D E F G H I J K L M N O P Q

dela

y (n

s)

Mean

RMS

Max.

5

posi t ions delay parameters normally d imish. In contras t to Tx 1 case, the room has the smaller window to the front build ing. So, the effect of ref lected s ignal d iminishes .

4. Conclusion

The 5GHz wideband channel model, which

his research was par t ly suppor ted by

wil l be used for new WLAN communication system, is the fundamental resource to design overal l communicat ion system. And i t is expected that 5GHz WLAN wil l be widely used in the apar tment. So in apar tment bui ld ing, we analyzed the power delay prof i les obtained by measurement system using PN correla t ion method. As a resul t , the ref lected s ignal f rom the front bui ld ing has to be considered as an important factor for systems engineer ing. The ref lected s ignal provides the unignorble effect on the wideband channel character is t ics , especial ly in the posi t ions having big window to the f ront bui ld ing or being far f rom the transmit ter. I f the t ransmit ter locates in the center of house, receiver posi t ions could be grouped due to the channel character is t ics . However, i f the t ransmit ter moves to the biased posi t ion of house, received s ignal power d ispar i ty of s ingle house and in terference problem to the neighbor ing house can be occurred. Based on th is work, the channel character is t ics var ia t ion due to the effect of wall , g lass window, furni ture and dwel lers wi l l be s tudied. TUnivers i ty IT Research Center Project , Brain Korea 21 and Korea te lecom.

References

[1] Chia-chin Chong, Chor-Min Tan, David I. Laurenson,

J.

Stephen McLaughlin, Mark A. Beach, and Andrew R.Nix, “A new statistical wideband Spatio-temporal

channel model for 5-GHz and WLAN systems,” IEEESelect. Areas Commun., vol. 21, no. 2, pp.139-150, Feb. 2003.

[2] Chia-chin Chong, D.I. Laurenson, and S.McLaughlin, “Statistical characterization of the 5.2 GHz wideband directional indoor propagation channels with clustering and correlation properties.” In Proc. IEEE Vehicular Technology Conf., vol.1, Vancouver, BC, Canada, September 2002, pp. 629-633

[3] Donald C. Cox, “Delay Doppler Characteristics of Multipath Propagation at 910 MHz in a Suburban Mobile Radio Environment”, IEEE Transactions on Antennas and Propagation, Vol. AP-20, NO. 5, September, 1972.

[4] A.A.M. Saleh, R.A.Valenzuela, “A Statistical Model for Indoor Multipath Propagation”, IEEE J.Selected Areas in Comm., Vol. SAC-5, No.2, Feb.1987, pp.128-137

[5] J.D.Parsons, The Mobile Radio Propagation Channel 2nd/E, John Wiley & Sons, ,2000.

[6] Lee, W.C.Y., Mobile Cellular Telecommunication Systems, McGraw Hill, New York, 1989.

[7] Theodore S.Rapaport, “Wireless communications: principles and practice 2nd /E”, Prentice Hall

[8] H.L.Bertoni, “Radio Propagation for Modern Wireless Systems”, Prentice Hall

[9] Do-Young Kwak, Nohgyoung Kang, Jaewon Lee, Seong-Cheol Kim, Joonsoo Choi, "Characterization of spatial channel model based on ray path analysis in high-rise urban environment", IEEE PIMRC'2003, vol. 1, pp. 955 - 959, 2003.

6