International Islamic University Malaysia (IIUM), Gombak, Kuala Lumpur, Malaysia

Transcription

1 Comparison o Empirical Indoor Propagation Models or 4G Wireless Networks at 2.6 GHz Al-Hareth Zyoud 1, Jalel Chebil 2, Mohamed Hadi Habaebi 3, Md. Raiqul Islam 4 and Akram M. Zeki Electrical and Computer Engineering Department, Faculty o Engineering, 5 Faculty o Inormation and Communication Technology International Islamic University Malaysia (IIUM), 531 Gombak, uala umpur, Malaysia 1 al7_z@yahoo.com, 2 jalel@iium.edu.my, 3 habaebi@iium.edu.my, 4 raiq@iium.edu.my and 5 akramzeki@iium.edu.my Abstract Indoor path loss models are playing an important role in the design and planning o the 4 th generation o the mobile networks. Moreover, it is an important component o system level simulators used to evaluate and test the network perormance beore it has been established. Many propagation models were proposed or this purpose. Most o these models are or macro and micro cellular networks. Small cell which is known as emtocell has been launched or uture networks and it is ldly deployed by the mobile operators around the world. The available propagation models accuracy is at question when applied to emtocell design and engineering. This paper attempts to quantiy the accuracy o these models by studying and comparing seven dierent propagation models or our dierent implementation scenarios at 2.6 GHz and or dierent separation distances. eywords Indoor propagation models, Femtocell, system level simulator, Free space loss, TE network. I. INTRODUCTION The rapid development o smartphones and tablet PCs increased the interest in indoor communication systems and emtocell technology. Femtocell which is a home access point is an important parameter in uture reless networks. Many researches are conducting annually around the world to study and investigate the challenges o deployment emtocell in the real environment. Thereore, to igure out and reduce the eects and the cost o emotcell deployment in eisting reless networks, system level simulator is necessary. The irst requirement o building a system level simulator is the path loss model. It plays an important role in intererence calculation. Since emtocell is installed indoor, it gained the attention to the indoor propagation models. Many propagation models were proposed in the literature. Most o these models are or macro and microcellular networks. Propagation models that are used in microcellular are not accurate enough to be applied in emtocell propagation calculations. An accurate model that considers more speciic details in addition to the direct path is required. Adding the number o loors and walls between the transmitter and the receiver is more accurate and less errors than using only the direct path [1]. The indoor propagation models that were ound in literature could be divided into two main categories. The irst is the general site models which are based on real measurement and not dependent on the site parameters [2-7].The second category is the site speciic models which proposed based on a speciic scenario and assumptions [8], [9]. The scope o this paper is the general site models. Seven o these models were studied and compared in our dierent scenarios. Most o the studies recently considered the indoor to outdoor propagation [1-12]. ess attention is paid or the in building models. In [13] and [14] the cost231 multi walls multi loors model was applied or the indoor path loss calculations. In this paper a comparison between dierent types o indoor propagation models has been proposed. Most o these models are suitable or three dimensional scenarios, where the number o penetrated walls and loors are considered. The path loss models were applied or our dierent scenarios in three story building. The requency that assigned or TE networks in Malaysia is 2.6 GHz. So, this requency ll be considered during the comparison process. The building consists o three loors, th three rooms each. The ceiling height is 3 m and the room dimensions are 6.5m 5 m. This paper is organized as ollows: seven dierent models are presented in the net section. Section 3 shows a comparison between the available models or dierent scenarios. Finally, a conclusion is given in Section 4. II. INDOOR PROPAGATION MODES Many studies have been conducted around the world that considered the indoor environment. Some are based on real measurements which also called general site models and others based on simulation and called site speciic models. In this section, many types o indoor propagation models are presented which are,, cost231, and multi walls multi loors () models. The scenario, requency bands and the required parameters to run these models are also highlighted. A. D112 V1.2 Model The channel models were proposed or indoor, indoor to outdoor, outdoor to indoor, and outdoor scenarios [2]. Firstly, they were applied at 2 and 5 GHz, then they were etended over requency range 2-6 GHz. The layout o the

2 indoor scenario is presented in Fig. 1. Two models were proposed, line o site model (OS), where there is no obstacles between the emtocell and subscriber user equipment (UE). The path loss is applied as in Equation (1). P 18.7 log( d) log( c / 5) (1) The second model is non line o site model (NOS) as in Equation (2), where the emtocell and the UE are in dierent rooms. Moreover, the number o penetrated walls and loors are considered. Fig. 1 Indoor scenarios layout or model [2] P 2 log( d) log( / 5) 12n F (2) F 17 4( n 1) (3) where c is the requency in GHz, d is the distance between the emtocell and the UE in meter, n w is number o walls, n is number o loors. Equations 2 and 3 show that the relation o penetrated obstacles is constant and linear th traversed loors and walls. It is increased by 12 db or the wall and 4 db or the loor. B. ITU-R M / 3GPP TR Model Based on real measurement results carried out in China [15-2] the WIINER II model was modiied and accepted by the ITU-R M.2135 [3] and proposed as in Equation (4) or OS and as in Equation 5 or NOS. P 16.9 log( d) log( c ) (4) P 43.3log( d) log( c ) (5) where c is the requency in GHz, d is the distance in meter between the transmitter and the receiver. Fig. 2 Indoor environment (one loor) or ITU-R M / 3GPP TR model [3] c W This model is also adopted by the 3GPP or indoor emtocell scenario [4]. The indoor scenario is presented in Fig. 2. It consists o one loor o 6 m height includes big hall and 16 rooms C. Cost 231 Models Three dierent types o indoor propagation models were proposed in cost231 [5] based on real measurements conducted in European cities at 9 and 18 MHz and were scaling or other requencies. The proposed models are as listed below: The irst is the cost231 one-slope model. This model is only considering the logarithmic distance between the transmitter and the receiver. The path loss is calculated as: P 1n log( ) (6) d Where is the path loss at 1 meter distance (db) and it is calculated as in Equation (7), n is the power decay inde, and d is the distance between emtocell and UE (m). It is clear that this model is equal to ree space loss as in Equation (9) i n is equal to log (7) The second model is attenuation model which indicates that the relation between the path loss and distance is linear. The path loss is epressed as: P FS d (8) where is the ree space losses (db) and it is calculated as FS in Equation (9), is the attenuation coeicient (db/m), and d is the distance between the emtocell and UE (m). FS log( d) 2 log( ) (9) The two previous models are very simple. They are not considering the penetrated walls and loors in the indoor environment between the transmitter and thee receiver as in the real scenarios. Thereore the third model cost231 multi walls multi loors model (Cost231 ) was proposed. In this model, in addition to the ree space loss, the attenuation due to walls and loors is accounted. The path loss is presented as ollows: P FS c I i1 k 2 b k 1 (1) where is the ree space loss between transmitter and FS receiver (db), C is a constant loss which determines rom measurements results and it is close to zero [5], k and are the number and loss o traversed walls o type i respectively, k and are number and loss o penetrated loors respectively, b is an empirical parameter, and I is the number o wall types.

3 Path loss [db] Path loss [db] D. ITUR P Model This model is based on real measurements and over requency range 9 MHz to 1 GHz [6]. The path loss is calculated according to the ollong equation: P 2 log( ) N log( d) ( n ) 28 (11) where N is the distance power loss coeicient, is the requency (MHz), d is the distance between the emtocell and UE (where d >1 m), is the loor penetration loss actor (db), and n is the number o loors separate between the emtocell and UE. This model is considering the number o loors only and proposed as a solution to the requency reuse between loors. E. The Multi Wall Multi Floor Model () This model is based on ray tracing simulation at 5.2GHz. It was veriied using open literature results or dierent requencies [7]. The path loss is computed as in Equation (12). The model shows a nonlinear relation between the traversed walls or loors and the penetration loss. MWF I i1 k1 j 1 nlog( d (12) ) k J j1 k1 where is the path loss in distance o 1m (db), n is the power decay inde and it determined between 1.96 and 2.3 or the considered scenario, d is the distance between the emtocell and the UE (m), k is the loss o k th wall o type i traversed (db), is number o traversed walls o category i, I is the number o wall categories, jk is the loss o k th loor o type i traversed (db), j is number o traversed walls o category j, and J is the number o loor categories. To sum up, all o the previous models could be applied or three dimensional indoor environments. Thereore, net section is presenting a comparison between the previous models or our dierent indoor scenarios. III. COMPARISON OF INDOOR MODES The indoor models which considered in this paper were proposed or dierent requency bands and or dierent scenarios. This section is presenting a comparison between these dierent models or the same requency and the same scenario to check the capability o these models. Four indoor scenarios are considered in this paper A, B, C, and D. The layout o the proposed scenarios is as shown in Fig. 3. The building consists o three loors, th three rooms each. The loor height is 3 m and the room dimensions are 6.5 X 5 m. The applied requency is 2.6 GHz, since it is the requency assigned or TE networks by Malaysian mobile operators. The irst scenario is A, where is the emtocell and the UE are in the same room and OS connection. The maimum distance is the length o the room which is 6.5m. is 2.6 GHz. n in equations (6) and (12) is 4 [5]. N in Equation (11) is 28 [6]. in Equation (8) is.62 [5]. The comparison is as in Fig. 4. Notice that the cost231 model in Equation (1) is jk same to FS as in Equation (9), and the model Equation (12) is same as the one slope model Equation (6). Ү T C Cost 231 Fig. 3 The layout o indoor scenarios Fig. 4 Path loss or scenario A Fig. 4 shows that the and models close to each other and epecting path loss range o 75 db or 6.5 meters distance, while the path loss is 3 db better or cost231, and ITUR M2135. The and cost231 the linear models are predicting 63 db in between A cost 231 D Fig. 5 Path loss or scenario B In the second scenario (B), the emtocell and the UE are in the same loor but in dierent rooms. The maimum distance is the length o the loor which is 19.5m. is 2.6 GHz. n, N B

4 Path loss [db] Path loss [db] and are same as in scenario A. w or cost231 model is 6.9 db [5]. w11, and w12 in the model are 16 and 14 db respectively [7]. Comparison between the seven models is shown in Fig. 5. In this scenario the eect o walls is considered. In Fig. 5 it is clear that, cost231, and models are close to each other and predicting 8 db losses or the two walls. The path loss is increased to 1 db or the and models. Moreover, the path loss is worse and up to 12 db by using the model. In the third scenario (C), three loors are considered. However, the UE in the third loor is eactly above the emotcell. It is one meter above the ground. Thereore, only two penetrated loors are accounted and the distance is 7 m. The ollong parameters are considered: n or two loors is 5.2 [5], N is 28, is 2.8, in cost231 model is 18.3 db, 11 and 12 in the model are 19 and 15.2 db respectively [7] cost Fig. 6 Path loss or scenario C Five o the models show almost the same perormance th some dierence as shown in Fig. 6. The path loss was 78 db or the linear model, while the cost 231 was 13 db worse and pointed at 81 db. In addition, the model shows over estimated result th 12 db loss. Finally, the perormance o the ITU-R 2135 was not stabile because the distance is less than ten meters, hence the NOS part o the model was proposed or distance more than 1 meters. Normally the distance between loors in the residential departments is around 3 meters. So the NOS part o the model ll not be suitable to estimate the penetration losses o loors even though it was proposed by the 3GPP or emtocell scenarios. The last scenario considered the diagonal distance between emtocell installed in the let side o the irst loor and UE in the right side o the third loor as in D. In this scenario the eects o walls and loors are studied. Models parameters are same as in scenarios B and C. Fig. 7 shows that our o the models are close to each other especially or the NOS th path loss around 11 db. However, The ITU-R models predicted less loss th 1 db or the and 8 db. astly, the epected loss using the model was very high th 175 db. IV. CONCUSIONS Seven dierent indoor propagation models available in the literature or 4G emtocell requencies was studied and compared in our dierent scenarios. In OS scenarios like scenario A, the minimum path loss is 55 db and it is increased to 75 db or the and models. In addition, or the NOS and same loor scenario, the loss is 8dB or most o the models ecept the model where the loss is around 12 db. The third scenario considered the loors penetration only. The path loss is 9 db or ive o the models and 12 or the model. The ITUR M2135 has limitations in this scenario since the minimum distance assumed or NOS in this model is 1 m. Finally, in the last scenario, walls and loor are accounted. The path loss is below 12 db or si models while it is 175 db or the model. The overestimation o the model might be due to the value used or the path loss eponent which is between 1.96 and 2.1 while or other models it is considered between 4 and 5.2. As a uture works, real measurement ll be conducted and compared th the previous models to check the best relected model cost Fig. 7 Path loss or scenario D ACNOWEDGMENT The authors are thankul to the International Islamic University Malaysia and to the Malaysian Ministry o Higher Education or supporting this study. REFERENCES [1] Saunders, S. R., Antennas and Propagation or Wireless Communication Systems, Wiley, [2] IST , D1.1.2 V1.2, Channel Models, Part I Channel Models, 28. [3] Report ITU-R M , Guidelines or evaluation o radio interace technologies or IMT-Advanced, 29. [4] Further advancements or e-utra (physical layer aspects), 3GPP TR v9.., Tech. Rep., March 21. [5] Digital mobile radio towards uture generation systems, inal report, COST Action 231, European Communities, EUR 18957, Tech. Rep., 1999.

5 [6] Report ITU-R P , Propagation data and prediction methods or the planning o indoor radiocommunications systems and radio local area networks in the requency range o 9 MHz to 1 GHz, 212. [7] M. ott and I. Forkel, A multi-wall-and-loor model or indoor radio propagation, inproc. VTC 21 Spring Vehicular Technology Con. IEEE VTS 53rd, vol. 1, 21, pp [8] M. Hassan-Ali and. Pahlavan A new statistical model or sitespeciic indoor radio propagation prediction based on geometric optics and geometric probability, IEEE Trans. Wireless Commun., vol. 1, no. 1, pp , 22 [9] C.-P. im, J.. Volakis,. Sertel, R. W. indt and A. Anastasopoulos "Indoor propagation models based on rigorous methods or site-speciic multipath environments", IEEE Trans. Antennas Propag., vol. 54, no. 6, pp , 26. [1] Corre Y, Stephan J, ostanlen Y: Indoor-to-outdoor path-loss models or emtocell predictions. In Proceedings o IEEE 22nd Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC). Toronto, Canada; 211: [11] Rose, D. M., ansen, T. and u rner, T. Indoor to Outdoor Propagation Measuring and Modeling o Femto Cells in TE Networks at 8 and 2 MHz. Presented at IEEE Global Communications Conerence (GOBECOM). Houston, Teas, 5th-9th December 211. [12] A. Valcarce and J. Zhang "Empirical indoor-to-outdoor propagation model or residential areas at GHz", IEEE Antennas Wireless Propag. ett., vol. 9, pp [13] etourneu, F., Corre, Y., Suteau, E., ostanlen, Y. 3D coverage analysis o TE urban heterogeneous networks th dense emtocell deployments, EURASIP ournal on Wireless Communications and Networking (October), 212. [14]. ee, S. Bae, Y. won, and M. Chung, Intererence analysis or emtocell deployment in OFDMA systems based on ractional requency reuse, IEEE Commun. ett., vol. PP, no. 99, pp. 1 3, Apr [15] Dong, W., Zhang, J., Gao, X., Zhang, P. and Wu, Y. Cluster Identiication and Properties o Outdoor Wideband MIMO Channel, Vehicular Technology Conerence. VTC-27 Fall. 27. [16] Fujii, T. Path-loss Prediction Formula in Mobile Communication -an Epansion o SA AGAMI Path-loss Prediction Formula. Trans. IEICE, Japan, J86-B, 1, p , 23. [17] u, Y., Zhang, J., Gao, X., Zhang, P. and Wu, Y. Outdoor-Indoor Propagation Characteristics o Peer-to-Peer System at 5.25 GHz, Vehicular Technology Conerence, 27. [18] Xu, D., Zhang, J., Gao, X., Zhang, P. and Wu, Y. Indoor Oice Propagation Measurements and Path oss Models at 5.25 GHz, Vehicular Technology Conerence, 27. [19] Zhang, J., Gao, X., Zhang, P. and Yin, X. Propagation Characteristics o Wideband MIMO Channel in Hotspot Areas at 5.25 GHz, Personal, Indoor and Mobile Radio Communications, 27. [2] Zhang, J., Dong. D., iang. Y., Nie. X., Gao. X., Zhang, Y., Huang, CH. and iu, G. Propagation Characteristics o Wideband MIMO Channel in Urban Micro- and Macrocell, IEEE PIMRC 28, Invited paper.