Using efficient ray-tracing techniques to predict propagation losses in indoor environments

5. 6.

7.

8.

9. 10.

works, Proceedings of the Sixth International Conference, TELSIKS 2003, October 1–3, 2003, Vol 2, pp. 565–568. G. Niu, Noise in SiGe HBT RF technology: Physics, modelling, and circuit implications. Proc IEEE 93 (2005), 1583–1597. J.P. Roux, L. Escotte, R. Plana, J. Graffeuil, S.L. Delage, and H. Blanck, Small-signal and noise model extraction technique for heterojunction bipolar transistor at microwave frequencies, IEEE Trans Microwave Theory Tech 43 (1995), 293–298. G. Jianjun, L. Xiuping, W. Hong, and G. Boeck, Microwave noise modelling for InP-InGaAs HBTs, IEEE Trans Microwave Theory Tech 52 (2004), 1264 –1272. U. Basaran, N. Wieser, G. Feiler, and M. Berroth, Small-signal and high-frequency noise modelling of SiGe HBTs, IEEE Trans Microwave Theory Tech 53 (2005), 919 –928. S. Haykin, Neural networks, IEEE, New York, 1994. V. Markovic and Z. Marinkovic, Small-Signal and Noise Modelling of Microwave Transistors Based on Neural Networks, In: Asia-Pacific Microwave Conference, APMC 2004, New Delhi, India, pp. 712–713. Invited paper, CD and Proceedings of Abstracts.

© 2007 Wiley Periodicals, Inc. Figure 3 Magnitude of optimum reflection coefficient versus frequency by using the model 2N 7 5 (continual curves) compared to measured values (symbols)

the complete procedure of training and testing, a neural network denoted with 2 N[N]7 5 was selected. Average and worst case errors for this model are presented in Table 1 as well. It can be observed from Table 1 that, by using the modified neural structure, the accuracy of modeling is improved, compared with the basic simple neural structure. Once trained, the ANN noise model provides an instantaneous response for different input vectors covering the whole operating range. As an illustration, Figure 3 shows the plots of the magnitude of optimum reflection coefficient versus frequency, obtained by using the model 2 N[N]7 5, at two different bias points: (1) Vce ⫽ 2 V, Ic ⫽ 5 mA; (2) Vce ⫽ 2 V, fc ⫽ 20 mA where the second one does not belong to the training set. The corresponding measured data are marked in the same figure. An excellent agreement of simulated noise characteristics with measured values can be observed. Finally, it should be pointed out that the developed ANN HBT noise models are implemented within the standard circuit simulator ADS as user-defined library elements, enabling very fast simulation for microwave circuit design.

USING EFFICIENT RAY-TRACING TECHNIQUES TO PREDICT PROPAGATION LOSSES IN INDOOR ENVIRONMENTS F. A. Alves, M. R. M. L. Albuquerque, S. G. Silva, and A. G. d’Assunc¸a˜o Departamento de Engenharia Ele´trica, Universidade Federal do Rio Grande do Norte (UFRN), 59072–970 Natal, RN, Brazil Received 24 August 2006 ABSTRACT: This paper addresses an efficient implementation of the 2.5D ray-tracing propagation model to predict propagation losses in indoor environments. The analysis is based on the Shooting and Bouncing Rays technique and Uniform Theory of Diffraction. Besides the lineof-sight propagation, we consider that the radio waves may experience reflection, refraction, and diffraction (NLOS). The simulated results are validated by the measurement carried out in a teaching building at 1.8 GHz. Validation is also provided by comparisons with published data. © 2007 Wiley Periodicals, Inc. Microwave Opt Technol Lett 49: 854 – 858, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.22266

4. CONCLUSIONS

A new, ANN-based approach can be used successfully for modeling the noise parameters of HBTs. For developing a neural noise model, only a number of measured data is needed. That gives an advantage to ANN approach in comparison with other modeling approaches, especially when the noise generating mechanisms of the device are too complex or not well known. Developed neural models are characterized by high accuracy together with the efficiency and simplicity and therefore are convenient for CAD purposes. REFERENCES 1. S. Lee, Direct extraction of base-collector model parameters for AlGaAs/ GaAs HBT equivalent circuit. Electronics Lett 33 (1997), 815– 817. 2. B. Li, S. Prasad, L. Yang, and S.C. Wang, A semianalytical parameterextraction procedure for HBT equivalent circuit, IEEE Trans Microwave Theory Tech 46 (1998), 1427–1435. 3. Q.J. Zhang and K.C. Gupta, Neural networks for RF and microwave design, Artech House, 2000. 4. K. Munshi, P. Vempada, S. Prasad, E. Sonmez, and H. Schumacher, Small signal and large signal modelling of HBT’s using neural net-

854

Key words: indoor propagation modeling; ray-tracing technique; power prediction; radio propagation

1. INTRODUCTION

The ever increasing demand on mobile communications and the rapid growth in cellular mobile have improved the development of tools that enable an efficient analysis of the power distribution behavior in indoor environments. The reason is that indoor propagation prediction is very important in the design of wireless communications systems [1– 4]. In this paper, we propose a 2.5D ray-tracing propagation model in combination with the Uniform Theory of Diffraction (UTD) to predict radio losses in indoor environment. The implementation of this model is carried out through the development of a graphic program and tools that assist the prediction, allowing versatility to the implementation. We choose the Shooting and Bouncing Rays (SBR) method due to the implementation easiness of the graphic analysis of the rays [5–7].

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 49, No. 4, April 2007

DOI 10.1002/mop

According to the SBR technique, a bundle of rays are launched from the transmitter towards all directions. Each ray is traced and may or may not reach the receiver. For every ray incident on an obstacle (such as a wall), a transmitted ray and a reflected ray are produced, each of which in turn has to be tracked [6]. Wedges, edges, or other discontinuities are treated as secondary sources of radiation, by appropriately calculating diffraction [7]. The propagation mechanism of each ray is treated separately. In line-of-sight propagation (LOS), the main contribution to the received signal comes from the direct ray. When the line-of-sight is blocked, then reflected, refracted, or diffracted signal (NLOS) dominates. Owing to the reflection, refraction, and diffraction of radio waves, several interactions can take place before the transmitted signal eventually reaches the receiver, resulting in a multipath fading [8 –11]. To model the environment, a database is built to insert geometrical characteristics and information on the constituent materials of the scenario. The database works independently of the simulation program, allowing robustness to the implementation and flexibility to model other scenarios. The transmitting channel of the mobile system is simulated by moving either the transmitter or the receiver around the environment. The reliability of the method is verified through simulations and measurements. Propagation characteristics are generated from the simulation, and are compared to measured values at 1.8 GHz frequency. The measurements were carried out for the main corridor and classrooms in a teaching building. A reasonable agreement is observed. The numerical predictions are also compared with published data at 900 MHz and 2.44 GHz frequencies showing good convergence. 2. PROPAGATION MODEL

Figure 1 illustrates the four main types of rays from the transmitter to the receiver. By using the concept of ray-tracing, rays may be launched from a transmitting and can reach the receiver either directly (LOS) or after reflections, refractions, diffractions, or even after any combination of the previous effects (NLOS). In the characterization of indoor electromagnetic propagation, the starting point is to consider the Friis transmission equation to determine the path loss when a transmitter has a line-of-sight unobstructed signal path to the receiver. The equation represents a good model of the attenuation in that path. The Friis transmission equation relates the power, which is fed to the transmitting antenna Pt, to the power, which is received by the receiving antenna Pr. The path loss is proportional to the square of the path length. Thus [12]

冉 冊

Pr ␭ ⫽ GtGr Pt 4␲d

Figure 1

Figure 2 Lay-out of an indoor propagation environment. The figure illustrates locations of receivers. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com]

where Gt and Gr are the antenna gains of the transmitter and receiver, respectively. The final term in the equation accounts for loss of the transmitted wave over the transmit-receive distance d, ␭ being the wavelength of the operating frequency. Following the formulations of reflection, transmission, and diffraction coefficients for vertical polarization of the fields are given, since they will be used to compute the indoor path loss with obstruction. The reflected wave fields are computed according to the well known Fresnel equation for the reflection coefficient. The reflected power is defined as [6]: P reflected ⫽ Pincident⌫.

(2)

The refracted rays can be considered in a fashion similar to the reflected rays. The refracted power is given by P refracted ⫽ PincidentT.

(3)

For perpendicular polarization, the reflection coefficient ⌫ and the refraction coefficient T can be expressed as ⌫⫽

␩ 2 cos␪i ⫺ ␩1 cos␪t , ␩2 cos␪i ⫹ ␩1 cos␪t

(4)

T⫽

2 ␩ 2 cos␪i , ␩2 cos␪i ⫹ ␩1 cos␪t

(5)

2

(1)

Ray-paths in an indoor environment

DOI 10.1002/mop

Figure 3 The measured and predicted path losses of 1.8 GHz propagation in the corridor R1. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com]

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 49, No. 4, April 2007

855

Figure 6 The measured and predicted path losses of 1.8 GHz propagation in the room R23. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com]

Figure 4 The measured and predicted path losses of 1.8 GHz propagation in the room R21. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com]

to find out all ray-paths between the transmitter and the receiver [15]. 3. RESULTS

where ␩1 and ␩2 are the intrinsic impedances of the medium one and two, respectively, and ␪ and ␪t are the angles of incidence and refraction, respectively. Furthermore, when the incident ray strikes an edge or a corner or a tip, a different class of rays is produced, called the diffracted or scattered rays. Diffraction modeling in the ray tracing technique is well established through the UTD. The diffracted power is related to the incident power by means of the diffraction coefficient as follows: P diffracted ⫽ PincidentD.

(6)

The diffraction coefficient D is given by [7] ␲

冋

册

1 1 e j4 sec 共␳ ⫺ ␸兲 ⫾ csc 共␳ ⫹ ␸兲 , D⫽ ⫺ 2共2 ␲ k兲 1/ 2 sin共␤兲 2 2

(7)

where k represents the wavenumber. The angles between the incident and diffracted rays and the normal are ␸ and ␳, respectively, and ␤ is the angle between the incident ray and the edge. An advantage of the ray-launching technique is the fact that there are no restrictions imposed on the environment and any shape of object may be present [13, 14]. A search algorithm is used

Figure 5 The measured and predicted path losses of 1.8 GHz propagation in the room R22. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com]

856

The propagation predictions presented are modeled using the raytracing (SBR) model and the UTD. Two indoor environments are taken into consideration. The algorithm enables determination of the power in several reception points, simultaneously, reducing, in this way, the computational time of processing. Besides the LOS, the radio waves may experience reflection, refraction, and diffraction. Figure 2 shows the first indoor environment considered here. The walls are made of bricks, the floor is made of concrete material, and the ceiling is composed of plasterboard. To compare simulations with measurements, the simulated results have been processed in a similar way as the measurements. The prediction accuracy is also evaluated through statistical comparisons of the standard deviation of the error ␴ and the mean error ␦ (the difference between the simulated and measured path losses). Propagation loss measurements were taken in a corridor and adjacent classrooms of a teaching building. Both the transmitting and receiving antennas are half-wavelength dipole antennas, vertically polarized, and placed at a height of 1.5 m above the floor. The transmitting frequency was fixed at 1.8 GHz. Figure 3 shows the power loss as a function of the distance in the corridor R1 (Fig. 2). The measurements were carried out keeping the transmission antenna stationary, while the receiving antenna was moved from position 1 to 29. The corridor has a length of 96 m and the distance between two measured points was taken equal to 20 cm. The results were obtained for LOS. The

Figure 7 The measured and predicted path losses of 1.8 GHz propagation in the room R24. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com]

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 49, No. 4, April 2007

DOI 10.1002/mop

Figure 10 The predicted path loss of 900 MHz propagation in the Route 2. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com] Figure 8 Lay-out of an indoor propagation environment. The figure illustrates locations of transmitters. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com]

measured data are compared with those simulated for the same propagation characteristics. The standard deviation between the prediction and measurements is 4.62 dB and the mean error is ⫺0.81 dB. Although the presence of ventilation rifts, each with 15 cm of diameter, on the wall of the corridor has been neglected in the simulations, the general behavior of the computed and measured fields is quite similar. To take the propagation loss measurements in the classrooms R21, R22, R23, and R24, the transmitter was kept stationary at the corridor R1 (Fig. 2), while the receiving antenna was placed inside each room, moving from position 1 until position 15. The distance between two adjacent positions is chosen to equal 1 m. In the rooms R22 and R23, when the reception antenna is placed next to the transmission antenna, which means that the reception antenna is moving from position 1 until position 5, the results show smaller power loss levels. The antennas can be reached with direct reflections in the wall of the corridor, and the contribution due to the reflected rays will be greater than those of the rooms R21 and R24. However, as the distance between the antennas increases, the measured and simulated path losses take about 60 dB as shown in Figures 4–7. From the mean error and the standard deviation of the error, it can be seen that a good agreement between simulations and measurements has been found. Particularly, in the room R23, the mean error ␦ is equal to 4.04 dB because the measured values are

Figure 9 The predicted path loss of 900 MHz propagation in the Route 1. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com]

DOI 10.1002/mop

greater than the computed values. This can be justified provided the ventilation rifts are not considered in the simulation procedure but perform influences in the measurements. To compare our numerical predictions with published results, simulations have been carried out for the indoor environment shown in Figure 8 [6]. Here, the stationary receiver was placed at beginning of the main corridor (Route 1), while the transmitter, in light-of-sight with the receiver, was moved along the route of 34 m. The path losses are simulated for the frequencies of 900 MHz and 2.44 GHz, and the results are presented in Figures 9–11. In Route 1, our numerical computations are shown in Figure 9 for 900 MHz. The results are compared and agree well with predicted and measured results available in the literature [6]. The values of the mean error ␦ and of the standard deviation of the error ␴ are also presented. To simulate the power prediction along the 20 m of Route 2, the model includes not only the reflection and refraction mechanisms, but also the diffraction for propagation around the corner. Our numerical results are shown in Figures 10 and 11. The simulations were obtained by considering that the distance between two adjacent positions is made equal to 1 m. From the mean error and the standard deviation of the error, it can be seen that our results show better concordance with the measured and simulated values presented in Ref. 6 for 900 MHz than those obtained through measurements and simulations for 2.44 GHz frequency. 4. CONCLUSION

In summary, this paper has performed propagation loss and multipath propagation measurements in indoor environments. The analysis combined the SBR method with the UTD. On the basis of

Figure 11 The predicted path loss of 2.44 GHz propagation in the Route 2. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com]

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 49, No. 4, April 2007

857

the ray-tracing theory, in addition to direct ray, we considered reflected, transmitted, and diffracted rays due to the finite object. The SBR approach was employed as a computationally efficient raytracing procedure to find the ray paths to the field points. The model accuracy was investigated through comparisons between measurements and simulated results. The validation of this model was also verified by means of comparisons with predicted and measured results presented in Ref. 6. This predicting technique can be used in a graphic implementation fashion, allowing flexibility to the model, since the ray behavior can be followed in real time, enabling estimation as well as corrections of the results. It could be seen that ray tracing methods can model propagation paths and the mechanisms by which radio signals propagate from transmitter to receiver, exhibiting accuracy and efficiency even for obstructed environments. ACKNOWLEDGMENT

We thank the Brazilian Research Agencies CNPq and CAPES for partial financial support. REFERENCES 1. T.S. Rappaport, Wireless communications: Principles and practice, Prentice Hall, New Jersey, 1996. 2. S.Y. Seidel and T.S. Rappaport, 914 MHz path loss prediction models for indoor wireless communications in multifloored buildings, IEEE Trans Antennas Propagat 40 (1992), 207–217. 3. C.-M. Chen, C.-C. Chiu, C.-H. Chen, and Y.-C. Chen, A novel propagation-prediction model for small rooms with metallic furniture, Microwave Opt Technol Lett 44 (2005), 281–284. 4. J.H. Tarng, W.-S. Liu, H.Y.-F. Huang, and J.-M. Huang, A novel and efficient hybrid model of radio multipath-fading channels in indoor environments, IEEE Trans Antennas Propagat 51 (2003), 585–594. 5. Z. Ji, B.-H. Li, H.-X. Wang, H.-Y. Chen, and T.K. Sarkar, Efficient ray-tracing methods for propagation prediction for indoor wireless communications, IEEE Antennas Propagat Magazine 43 (2001), 41– 49. 6. J.H. Tarng, W.R. Chang, and B.J. Hsu, Three-dimensional modeling of 900-MHz and 2.44-GHz radio propagation in corridors, IEEE Trans Vehic Technol 46 (1997), 519 –527. 7. D.A. McNamara, C.W. I. Pistorius, and J.A.G. Malherbe, Introduction to the uniform geometrical theory of diffraction, Artech House, Boston, 1990. 8. K.A. Remley, H.R. Anderson, and A. Weisshar, Improving the accuracy of ray-tracing techniques for indoor propagation modeling, IEEE Trans Vehic Technol 49 (2000), 2350 –2358. 9. F.A. Agelet, A. Formella, J.M.H. Ra´banos, F.I. de Vicente, and F.P. Fonta´n, Efficient ray-tracing acceleration techniques for radio propagation modeling, IEEE Trans Vehic Technol 49 (2000), 2089 –2104. 10. M. Hassan-Ali and K. Pahlavan, A new statistical model for sitespecific indoor radio propagation prediction based on geometric optics and geometric probability, IEEE Trans Wireless Commun 1 (2002), 112–124. 11. V. Degli-Eposti, G. Lombardi, C. Passerini, and G. Riva, Wide-band measurement and ray-tracing simulation of the 1900-MHz indoor propagation channel: Comparison criteria and results, IEEE Trans Antennas Propagat 49 (2001), 1101–1110. 12. A. Ishimaru, Electromagnetic wave propagation, radiation, and scattering, Englewood Cliffs, Prentice Hall, New Jersey, 1991. 13. Z. Zhang, Z. Yun, and M.F. Iskander, Ray-tracing method for propagation models in wireless communication systems, Electron Lett 36 (2000), 464 – 465. 14. C. Passerini, A quality measure for ray-tracing algorithms, IEEE Trans Antennas Propagat 49 (2001), 500 –502. 15. F.A. Alves, M.R.M.L. Albuquerque, S.G. Silva, and A.G. d’Assunc¸a˜o, Efficient ray-tracing method for indoor propagation prediction, In Proceedings SBMO/IEEE Intern Microwave and Optoelectron Conf, IMOC’, Brası´lia, Brazil, 2005, pp. 1– 4. © 2007 Wiley Periodicals, Inc.

858

SOME NOVEL DESIGN FOR RFID ANTENNAS AND THEIR PERFORMANCE ENHANCEMENT WITH METAMATERIALS M. Stupf,1,2 R. Mittra,1 J. Yeo3, and J. R. Mosig2 1 Electromagnetic Communication Laboratory, 319 EE East, The Pennsylvania State University, University Park, PA 16802 2 Laboratory of Electromagnetic and Acoustics, EPFL-Station 11, Ecole Polytechnique Fe´de´rale de Lausanne, CH-1015 Lausanne, Switzerland 3 Electronics and Telecommunications Research Institute, Daejeon, Korea Received 24 August 2006 ABSTRACT: In this paper we discuss a number of UHF/RFID tag designs including the hybrid loop (Chen and Hsu, International Symposium on Antennas and Propagation, Korea, Vol. 3, 2005, 1045–1048), dual crossed-dipoles, the dual crossed-dipoles utilizing an inductively coupled feed (Choo and Ling, Electron Lett 39 (2003) 3080 –3081). We have carried out extensive parametric studies in the process of analyzing the characteristics of the earlier-mentioned antennas, and have investigated several techniques (Choo and Ling, Electron Lett 39 (2003) 3080 –3081), including meandering, for size-reduction of these antennas. A design methodology based on the Genetic Algorithm (GA) is presented for the optimization of conformal antennas with Electromagnetic Bandgap (EBG) surfaces (Mittra, Proceedings of International Symposium on Antennas and Propagation 2005 (ISAP2005) Seoul, Korea, 2005, 325–328; Sievenpiper et al., IEEE Trans Microwave Theory Tech 47 (1999), 2059 –2074) to improve the antenna performance. The EBG characteristic is realized by utilizing a Frequency Selective Surface (FSS) placed above a thin dielectric substrate backed by a metallic ground plane to act as an Artificial Magnetic Conductor (AMC). The antenna is then placed above the EBG surface to create a conformal integrated EBG - antenna. The initial design of the AMC is carried out by synthesizing a surface whose reflectivity is “⫹1”, for a normally incident plane wave. However, since the antenna situated above it produces not just a single plane wave but a spectrum of these waves, it becomes necessary to refine the initial design to optimize the performance of the antenna/AMC composite, and techniques for doing this are discussed in the paper. © 2007 Wiley Periodicals, Inc. Microwave Opt Technol Lett 49: 858 – 867, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.22265 Key words: RFID; antennas; metamaterials

1. INTRODUCTION

Radio frequency identification (RFID) is a rapidly developing technology that uses RF signals for automatic identification of objects. An RFID system consists of a reader, a tag (transponder), and a computer connected to the reader. Recently, there has been considerable research into the design of RFID tag antennas in the UHF band. Most applications require that the tags be small, as well as inexpensive. Realizing a good impedance match between the antenna and the chip is of paramount importance in RFID designs. Since tailoring new IC designs to suit a tag is a rather costly venture, we attempt, instead, to design RFID tag antennas for a specific ASIC chip that is readily available in the market. Also, adding an external matching network with lumped elements in RFID tags is usually prohibitive because of cost and fabrication issues. In light of this situation, it is desirable to explore designs in which the antenna is directly matched to the ASIC, which has a complex impedance. In the present study, for instance, we seek to match the input impedance of the antenna to an IC chip (Philips RFID/ASIC), whose impedance was Zc ⬃12⫺j 300 ⍀ at 900

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 49, No. 4, April 2007

DOI 10.1002/mop