IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 47, NO. 1, FEBRUARY 1998 225 Characteristics’ Prediction in Urban and Suburban Environments Nathan Blaunstein, Ran Giladi, and Moshe Levin Abstract— In this paper, the character of the propagation in the microcellular urban and suburban environments in the 902–928-MHz frequency band in line-of-sight (LOS) conditions is investigated both theoretically and experimentally for the purpose of wireless radio local loop (WRLL) prediction. The path-loss characteristics and the range of a break point, at which the polynomial character of field intensity decay along the street level is changed, are analyzed for various parameters of street widths, for different average building heights, and for the actual electrical impedance properties of building walls. A multislit waveguide with randomly distributed gaps between the sides of buildings is considered as a model of straight streets. Results of experiments for VHF/UHF wave propagation along the straight streets in urban environment in the conditions of direct visibility between receiver and transmitter are compared with theoretical analysis of field intensity decay, path-loss distribution, and dependence of the break point on street topography in LOS conditions. I. INTRODUCTION HE PREDICTION of propagation characteristics is an essential part of radio wireless network planning. The development of new radio communication systems like the multigain wireless (MGW) system [1] requires knowledge of propagation conditions for the operation environments. To successfully predict the performance of such systems, detailed experiments and theoretical investigations have been carried out to examine wideband radio channel characteristics and their dependence on the various real urban environment parameters [2]–[9]. The MGW system [1] is a wireless local loop system used in local networks, serving as an alternative to conventional loop distribution networks. To design the system successfully, it is very important to investigate the characteristics of urban radio channels, define optimal locations of the radio ports, and make performance predictions for the individual subscribers. Methods of predicting field intensity attenuation (or path loss) for line-of-sight (LOS) propagation along the streets and the range of a break point , at which the characteristic of energy loss with distance from the transmitter is changed from the mode (for ) to the mode (for ), were introduced earlier on the basis of the “two-ray” model [7], [8], [10]–[12]. But as was noted in [13], using the “tworays” model we cannot explain the experimentally observed changes of the mode of field intensity attenuation from , T Manuscript received June 28, 1996; revised February 12, 1997. N. Blaunstein and R. Giladi are with the Department of Communication Systems Engineering, Ben-Gurion University of the Negev, Beer Sheva 84105, Israel. M. Levin is with Tadiran Communication Ltd., Petakh-Tikva 49104, Israel. Publisher Item Identifier S 0018-9545(98)00691-4. to , beyond the break point, , which m to m [9], [10], [13]–[15]. varied from Moreover, the “two-rays” model cannot describe the absence of a break point, that is, the absence of two-mode law of field intensity decay, in LOS conditions along narrow streets or streets with high buildings in urban cells with a range of 1–2 km [13], [16], [17]. A new multislit waveguide model with randomly distributed slits (gaps between buildings) and screens (walls) for low antenna heights (in comparison with building heights) along the street level has been proposed in [13], [15]. This model considers the multiple reflections on building walls, multiple diffraction on their corners, and reflections from the road surface. This is intended for field intensity analysis in LOS conditions for a regularly planned city with straight streets and with random intersections between them and between buildings along each street. It was shown that the total field is formed basically by the waves reflected from the walls and ground surface as well as by the waves diffracted from the edges of buildings. Both the proposed two-dimensional (2-D) waveguide model [13] and three-dimensional (3-D) waveguide model [15] correctly (in comparison with other models and experimental data) described the transformation with essential of field intensity attenuation from the mode intensity oscillations to the smooth exponential attenuation (which is closed to the experimentally obtained propagation , – up to 1–2 km from the source, the mode existence of a break point at which this transformation occurs, – m. and the range of this break point, However, in these works all new effects were analyzed without taking into account the actual impedance properties of building walls. Moreover, such an important parameter of LOS propagation along the streets as a break point, was not examined in detail according to existing numerous experimental measurements of this parameter. Below we continue to analyze the problem of wave propagation in LOS conditions along straight streets using the concept of a multislit street waveguide according to [13] and [15] and consider walls (screens) with actual dielectric properties. The range of break point is investigated in more detail, taking into account approximate formulas obtained both from the waveguide model and from the “two-ray” model. The field intensity decay and path-loss distribution are compared with experimental data obtained in [2] and from the experiments carried out in the suburban environment with regularly distributed straight crossing streets. 0018–9545/98$10.00  1998 IEEE 226 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 47, NO. 1, FEBRUARY 1998 Fig. 1. The simplified scheme of Kefar-Yona houses built on a rectangular street grid and of the first experiments in LOS conditions along the street. II. CONDITIONS OF THE EXPERIMENT The first series of measurements were taken in the small town of Kefar-Yona, Israel, where the MGW system of Tadiran, Ltd. was under trial in conditions of direct visibility along the street (LOS conditions, see Fig. 1). The omnidirectional base-station antenna was located at the same (or lower) level than the buildings’ roofs, at a distance of 4–5 m from the corner building surface as depicted schematically in Fig. 1. The mobile omnidirectional radio port antenna changed its position along the street in the middle of the road in LOS conditions (Fig. 1). The tested MGW system was operated in – MHz and utilized spreadthe frequency band spectrum (frequency hopping) digital radio communication. The tested environment is a typical small urban or suburban region of two- and three-storied brick buildings with – m and with a rightapproximately uniform heights angle crossing straight street plan (as schematically presented in Fig. 1). The base-station transmitter antenna was installed at m; the moving radio port antenna was lower the height – m). The tested cell radius of than rooftop level ( such an area estimated from measurements was approximately 1–2 km. Field intensity measurements in decibels relative to m (to compare intensity in free space at the range km (to compare with our calculations) and at the range with experiments carried out by Hughes [2]) were obtained to estimate the path loss and the field intensity attenuation in LOS conditions along the street. In these estimations, we take into account actual dielectric properties of the brick walls of buildings and the real distribution of buildings along the street level. III. LOS PROPAGATION ALONG THE STREET Instead, the “two-ray” model presented in [7], [11], and [12], which is usually used in LOS conditions, consisting of the direct and ground-reflected rays and 2-D propagation model consisting of multiple reflections from the walls [17], in conditions where both antennas were placed at the street level in LOS conditions, a new multislit waveguide model was proposed in [13] and [15] for describing wave propagation in urban areas with straight streets and with randomly distributed walls of buildings and gaps between them (slits) as shown in Figs. 2 and 3. BLAUNSTEIN et al.: PREDICTION IN URBAN AND SUBURBAN ENVIRONMENTS 227 Fig. 2. A 3-D model of the street waveguide. Fig. 3. A 2-D diagram of the street waveguide in the zy plane. The coordinates of source are y = 0 and z = d, and “a” is the street width. Let us now examine the propagation characteristics in LOS conditions along the street in more detail taking into account the real dielectric properties of building walls, actual distribution of street widths, and building heights along the street level. A. Waveguide Model Let us consider that the buildings on the street are replaced by randomly distributed nontransparent screens with scales , the electrical properties of which are defined by surface 228 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 47, NO. 1, FEBRUARY 1998 ; the distances between the buildings (slits) impedance , , , (see Figs. 2 and 3). The laws we define as , of their differential distribution are postulated as independent and , respectively, and exponential with mean values the probability density functions being given by (1) The impedance parallel plane waveguide with randomly distributed buildings along the street models a city street with receiver and transmitter at street level below the rooftops (see Fig. 2). One waveguide plane is placed at the waveguide side , and the second one at (“ ” is the street width, see Fig. 3). We also assume that a vertical electric dipole as plane with coordinates a real transmitter is placed in the , , and , where 0 on the axis, 0 on the axis, is the transmitter is the average building height. Taking into height, and account that all dimensions are much greater than the radiation wavelength , we use the approximation of the geometrical theory of diffraction (GTD) [19]–[22] for the rays reflected from the walls and rays diffracted from the building corners. Here, we present the full field as a superposition of a direct wave field from the source, fields reflected from the road and walls, and fields diffracted from the building edges [15]. As was shown by [7] and [8], we can exclude in our derivations the twice reflected waves, that is, “road-wall-road” and “wallroad-wall” reflected rays. In this approximation, the resulting field can be considered as a sum of the fields arriving at , from the virtual image the receiver, placed at height sources as schematically presented in Figs. 2 and 3. In most measurements for VHF/UHF waves, this condition is true for – km [7], [8], [10], [11], [16]. distances According to the approach proposed in [15], we also introand , duce here the “telegraph signal” functions which equal one when reflection or diffraction from the walls (screens) takes place and zero when rays pass through the spaces between the buildings, i.e., fall into the slits of the represent screens waveguide. Thus, segments with represent including their edges, but segments with slits (see Fig. 3). The real electric properties of screens (walls) , are defined by the surface electric impedance , where is the dielectric permittivity of the dielectric constant of vacuum, the the wall surface, the angular electric conductivity of the wall surface, and , where frequency of the radiated wave with denotes the frequency of the radiated wave. The full field inside the street waveguide can be presented as a sum of the direct field from the source and rays reflected and diffracted from the building walls and corners. In order to calculate the full field from the source, we substitute for each reflection from the walls an image source (for the first reflection from the left-hand walls of the street waveguide) and (for the first reflection from the right-hand walls), where is the number of the reflections (see Fig. 3). B. Average Field Along the Street To calculate the average total field along the street waveguide we, as in [13] and [15], take into account the exponential screens and slits distributions (1), treat the transmitter antenna as a vertical electric dipole with field , and the simple evaluations from the GTD in which the formulations of diffracted waves are similar with those obtained for reflected waves and in which the reflection coefficient of each reflected wave from the screen (wall) is simply replaced with a diffraction coefficient for each diffracted ray from the wall’s edge [19]–[22]. Here, is distance from the source , is the electric momentum of a vertical electric dipole, is the wave number, and is the wavelength . Using the procedure of averaging arbitrary order moments of “telegraph signal” functions according to [15] and combining the reflected and diffracted waves with the direct wave (LOS component) from the source and introducing the Fourier transform, direct and inverse, we obtain the total average field in the broken impedance waveguide given in (2) at the bottom of the page. Here, is the parameter of brokenness, , and is the constant introduced above. The coefficient of reflection for each reflected waveguide mode will be presented below. The difference between (2) and that obtained earlier in [15] [see (18), p. 318] lies in the existence of reflection coefficient in the integrand. As was obtained for the perfectly conductive multislit waveguide (when 1), (2) consists of two terms: the first term is obtained from integration of integral (2) along contour in the complex plane, and the second one can be obtained after integration of (2) along path in the complex plane (see Fig. 4). We will now examine these terms for the case of an impedance multislit waveguide. 1) Discrete Spectrum of the Total Field: The first term in (2) represents the discrete waveguide mode spectrum of the total field inside the broken waveguide. It can be obtained from (2) for the case as (3) (2) BLAUNSTEIN et al.: PREDICTION IN URBAN AND SUBURBAN ENVIRONMENTS Fig. 4. A scheme of integration paths for (2) in the complex Here, K 229 plane. , , , , is the constant describing the electric properties of the vertical is the coefficient of reflection from the electric dipole, and impedance walls inside the waveguide (4) is its modulus, and is its phase given in (5) at the bottom of the page. For the case of a perfectly conductive , , and , we obtain waveguide, when from (3) (6) which is the same expression obtained in [15] for the case of a perfectly conductive multislit waveguide. In the broken waveguide, as follows from (3) and (6), the wave modes attenuate inside it exponentially. But in the case of the impedance waveguide, their lengths of extinction attenuate inside the multislit waveguide very quickly; the corresponding lengths of extinction are decreased. In both cases of perfectly conductive and impedance waveguides, an increase in parameter ( ), i.e., a decrease in the gaps between buildings along the street, leads to a decreased attenuation of the reflected waves and to an increase of the extinction lengths [see (7)]. In the limit of an unbroken waveguide ( , ), the normal waves with numbers (the main modes) propagate without appreciable attenuation at large distances (depending on the actual dielectric properties of walls’ surface). In the case of a perfectly conductive unbroken waveguide ( , ), the process of wave propagation inside it continuously limits to the classical one when and normal modes propagate without attenuation [19], [20]. 2) Continuous Spectrum of Total Field: The second term represents the continuous part of the total field spectrum , which for the case can be presented in the following form [15]: (8) (7) depend not only on the number of reflections “ ,” on the waveguide (street) width “ ,” and parameter of brokenness . It also depends on the actual electric properties of the or, as waveguide walls, which are defined by the parameter . Thus, with follows from (5), by the dielectric impedance ( ), as follows from an increase in parameter (3)–(7), the extinction lengths become smaller, and the normal waves in the impedance multislit waveguide attenuate faster than in the case of a perfectly conductive multislit waveguide. The same picture is observed with an increase in the number of reflections . The normal reflected modes with numbers where is the constant describing the real properties of electric dipole. As can be seen from (8), in the case of a broken waveguide the continuous part of the total field propagates as a spherical wave and reduces to the case of an unbroken waveguide in the limit . But, whereas inside the perfectly conductive unbroken waveguide for the large distances ( ) the continuous part of total field is canceled by the source field, that is, [15], in the unbroken impedance waveguide this term does not vanish. As can be seen from (8), for the case (unbroken waveguide), but , the term differs from zero, i.e., in calculations of field intensity and path loss we (5) 230 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 47, NO. 1, FEBRUARY 1998 must take into account both field spectra, the continuous and discrete, which exist in the unbroken impedance waveguide. This is a new principal result, which is absent in the case of an ideal conductive unbroken waveguide. obtain the normalized average intensity for the numerical calculations C. Field Intensity Attenuation Let us consider as above that the term is the field reflected from the wall with coordinate and the term is the field reflected from the wall with coordinate (see Fig. 3). Then, the average total field intensity inside the waveguide can be presented in following form: (12) (9) where and are wave impedances. The sign (*) denotes the inverse complex field values. Using the differential distribution (1) and the results of the procedure of averaging for the one- and two-order moments of “telegraph” functions presented in [15], we finally obtain the expression for the average intensity inside the multislit waveguide Here (10) Using the Fourier representation of the field (13) and introducing it in (10), we can calculate the average field intensity (11) Here, the polar coordinates , , and new parameters and were introduced, where and . is the coefficient of reflection from the road surface, which is considered also as an impedance (in [15] it was assumed that the ground surface is perfectly conductive, i.e., 1); all other parameters were introduced above. After integrating (11) and taking into account the intensity from a vertical electrical dipole in free space, we finally In (12), the two first terms describe the multiray reflection from the walls and diffraction from the walls’ edges. They present normal modes generated in the multislit waveguide, which transform into normal waves in the unbroken waveguide (when ) and propagate along the waveguide with exponential attenuation at large distances. The third term describes the reflection from the ground and then the reflection from the buildings’ walls. The last term describes the direct wave and then reflections from the road surface and from the walls. In the case of the real impedance unbroken waveguide, when , but , this term does not vanish (as was obtained in [15] for a perfectly conductive waveguide). Thus, in the case of impedance waveguide, we must take into account all waves, direct from the source, reflected from the ground, and then both reflected from the walls (as schematically presented in Fig. 2). On the other hand, for we tend to the case of free space propagation above the ideal conductive flat surface. In this case, the last term is much larger than other terms and describes the interference between the direct waves and those reflected from the road, which is described by the term . The field intensity attenuates as a spherical wave . This case is close to the two-rays model presented, for example, in [11], where the plane wave propagates above the flat ideal conductive surface. Moreover because earlier in [13], BLAUNSTEIN et al.: PREDICTION IN URBAN AND SUBURBAN ENVIRONMENTS [15] it was shown that both the 2-D and 3-D waveguide models give sufficiently accurate results of field intensity attenuation, it is very important for the prediction of experimental data to obtain the approximative expression of average field intensity inside the impedance multislit waveguide. Taking into account that the average intensity can be approximately presented as and after integrating (2) for field in the limits of , , and using the same procedure as in [15], we finally obtain the approximate expression for the average field intensity at a large range from the source 231 which continuously limits to the approximate formula only for the case of , , . In fact, let us determine the “break point” range as the range at which the first term in (14) is equal to the second one, i.e., (15) (14) and . We do not present here the expression of the reflection coefficient because it is sufficiently fully described in the literature for different kinds of radiated field polarization (see, for example, [19]–[21]). Let us now examine expression (14) for various actual experimental situations in the urban street scene. 1) Let us consider that the street width is larger than the average building heights and both antenna heights, that is, , , . In this case, at distances less than the “break point” in the approximate formula (14), the second term, which describes the direct wave and the waves reflected from the ground and which attenuates as a spherical wave , is larger than the first term, which describes the attenuation of the normal reflecting modes along the multislit street waveguide. Beyond the “break point,” conversely, the first term in (14) is larger, and field intensity attenuates exponentially. This law of attenuation is close to that obtained experimentally in most measurements, where the attenuation mode of field intensity beyond the “break point” was , – . According the “two-ray” model [10]–[12], one can obtain only two modes of field intensity decay: before and beyond the “break point.” Moreover, no clear physical explanation of such a rapid (with – ) field intensity attenuation in the farthest zones from the transmitter existed until now. This effect can be clearly understood using the waveguide street model and following from it the exponential attenuation of field intensity (which is close to mode , – ) at the distances beyond the “break point.” As was shown in [15], the waveguide model continuously tends to the “two-rays” model in the case of wide streets. We can also show that the “break point” range presented in [11], , can be used to estimate the “break point” range only for urban areas with wide streets (avenues). Moreover, the 3-D waveguide model allow us to obtain a stricter expression for “break point” range estimation, Here, After the expansion of the exponent in the right hand of (15) into the series for and and taking into account the fact that because , and 1 all terms with their product are smaller than one, we finally obtain the approximate formula of “break point” range for the waveguide street model (16) which continuously (with constant in order with one for and ) tends to presented, for example, in [11] for the case when and , i.e., for the case of wide streets. 2) In the inverse case of urban areas with narrow streets ( ) the approximate waveguide model (14) can be successfully used to describe the field intensity attenuation along the street in LOS conditions (see Fig. 5, curve 2). In fact Fig. 5 depicts the field intensity attenuation relative to the intensity in free space at the distance 100 m from the source, using the approximate model according to (14) (curve 2), and the strict model according to (12) (curve 1) for , , and 0.8 for the case of a wide street ( m, m, m, and m). As can be seen from the illustration presented in Fig. 5, both models [strict (12) and approximate (14)] predict two modes of field intensity attenuation from to exponential and the existence of a break point at the range – m [according to (16)] for – MHz. Additional estimations have showed that for the case of narrow streets, the range of the break point is farther from the transmitter than in the case of wide streets. For example, for the case of a narrow street with m, m, m, and – MHz, the break point has been observed at the range – m for m and – m for m according to (16). Thus, as can be seen from illustrations and as follows from (16) for and , the range of the “break point” tends to infinity for the observed wavelength band – m with a decreasing street width or with an 232 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 47, NO. 1, FEBRUARY 1998 Fig. 5. The total field intensity attenuation along the street waveguide for M = 0:5 and a = 50 m. The curve 2 represents the calculations according to the approximate model (14), and the curve 1 represents the calculations according to the strict model (12). increasing building height. In the inverse case of a wide street, the approximate model (14) tends to the “two-rays” model, and (16) transforms into that presented in [7]–[12]. So we have a good transition from the waveguide model to the “two-ray” model in the particular case of wide avenues or canyons with building heights less than the street width. On the other hand, the waveguide multislit model, more generally than the “tworay” model, predicts the propagation characteristics in LOS conditions along straight streets. Because the approximate waveguide model gives closed results with the strict waveguide model (see Fig. 5), we can use the simple formula (14) to obtain the path loss in the LOS conditions along straight streets with great accuracy (17) This formula is more general than the approximate formula obtained in [13] for the case of the 2-D waveguide model without taking into account the reflection from the road and actual dielectric properties of building walls. IV. COMPARISON WITH EXPERIMENTAL DATA Let us now compare the theoretically obtained (17) with Tadiran’s experimentally measured path loss along straight streets in the urban area investigated. As was mentioned above, the first experiments were carried out in LOS conditions when the transmitter and receiver antennas were placed at the street level with direct visibility below the roof tops, and the moving radio port changed its distance from the stationary base station in the range 10–300 m (see Fig. 1). In Fig. 6 the normalized average field intensity decay in decibels (relative to the intensity of the vertical electric dipole in free space) is presented versus the distance “ ” from the transmitter along the street waveguide for the same conditions as presented in Fig. 5 by curve 1 only for the strict model (12). The solid points correspond to experimental measurements at 920 MHz. As can be seen, the 3-D impedance waveguide model (strict and approximate, because there is not sufficient difference between them, as follows from Fig. 5) gives close results with experimental data and can be used for predicting the path-loss distribution and the range of the break point along the street in LOS conditions. The influence of real electric properties of building walls can be seen in more detail from numerical calculations of (14) presented as a family of curves in Fig. 7 for parameters , , , , and , respectively, and , where is the field intensity normalized to the wave BLAUNSTEIN et al.: PREDICTION IN URBAN AND SUBURBAN ENVIRONMENTS 233 Fig. 6. The field intensity (in decibels) versus the distance r from the transmitter (in meters). The solid curve represents the numerical calculations according the strict model (12) for the same case, as in Fig. 5, and the solid circles represent the experimental data of signal attenuation (its maximum and minimum values). Fig. 7. The normalized field intensity versus the normalized distance r=a from the base station for M = 0:8, a = 20 m, and j0n j = 0:1, 0:2, 0:4, 0:6, and 0:8. Signs “+” represent measured data in a city region with multistoried ferroconcrete buildings; signs “o” represent measured data in a city region with moderate-storied brick buildings. from illustrations presented in Fig. 7, the curves with are closer to experimental data measured in a city area with multistory ferroconcrete buildings (depicted as “o” in Fig. 7). The curves with are closer to the experimental data measured in an area with buildings of moderate height usually constructed from bricks (depicted as “ ” in Fig. 7). Both theoretically and experimentally obtained results give exponential attenuation of radio waves in the farthest zones from the source in LOS conditions along the straight streets in regularly straight street planned urban areas up to 2–3 km. Moreover, for experimentally observed ranges in the conditions of direct visibility, when both receiver and transmitter antennas are placed at the street below the rooftop level, we can use the nonregular multislit waveguide model and with great accuracy approximate formulas (14), (16), and (17) to estimate the propagation loss and “break point” range at street level. V. CONCLUSIONS intensity in free space at the distance 1 km from the transmitter, at frequency MHz versus to normalized distance ( m) along the street. In Fig. 7, signs “ ” and “o” correspond to experimental measurements carried out in [2] at frequency 936 MHz, using two mobile station moving in two different areas of city center. As can be seen The strict and approximate 3-D multislit waveguide models constructed above are generalized waveguide models obtained earlier in [13] and [15] in the case of building walls along the street and road surface with the actual dielectric properties. They give a good explanation of experimentally observed twomode VHF/UHF wave attenuation with the transformation of 234 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 47, NO. 1, FEBRUARY 1998 to exponential attenuation, field intensity attenuation from the existence of a break point at which this transformation takes place, and the range of this break point – m. Moreover, these models can explain the essential changes of break point location relative to the transmitter, and the absence of this point in some urban conditions observed experimentally. Using simple formulas (14)–(17), we can with great accuracy predict the radio-wave-propagation characteristics, that is, field intensity decay, path-loss distribution along the street and the break point range, which defines a microcell size in LOS conditions along straight streets. [18] [19] [20] [21] [22] , “A theory of propagation path loss characteristics in a city streetgrid scene,” IEEE Trans. Electromagn. Compat., vol. 37, pp. 333–342, Aug. 1995. C. A. Balanis, Advanced Engineering Electromagnetics. New York: Wiley, 1989. L. B. Felsen and N. Marcuvitz, Radiation and Scattering of Waves. Englewood Cliffs, NJ: Prentice-Hall, 1973. R. G. Kouyoumjian and P. H. Pathak, “A uniform theory of diffraction for an edge in a perfectly conducting surface,” Proc. IEEE, vol. 62, pp. 1448–1461, Nov. 1974. G. Lampard and T. Vu-Dinh, “The effect of terrain on radio propagation in urban microcells,” IEEE Trans. Veh. Technol., vol. 42, pp. 314–317, Aug. 1993. REFERENCES [1] “Multigain-wireless,” Special Rep., Tadiran, Israel, p. 18, May 1993. [2] K. A. Hughes, “Mobile propagation in London at 936 MHz,” Electron. Lett., vol. 18, pp. 141–143, 1982. [3] R. Steele, “The cellular environment of light weight handheld portables,” IEEE Commun. Mag., pp. 20–29, July 1989. [4] P. Harley, “Short distances attenuation measurements at 900 MHz and 1.8 GHz using low antenna heights for microcells,” IEEE J. Select. Areas Commun., vol. 7, pp. 5–11, Jan. 1989. [5] E. Green, “Radio link design for microcellular systems,” British Telecom Tech. J., vol. 8, pp. 85–96, Jan. 1990. [6] K. Stewart and D. Schaeffer, “The microcellular propagation environment,” in Symp. Microcellular Technology, Mar. 1992, pp. 19–26. [7] H. H. Xia, H. L. Bertoni, L. R. Maciel, A. Lindsay-Stewart, and R. Row, “Radio propagation characteristics for line-of-sight microcellular and personal communications,” IEEE Trans. Antennas Propagat., vol. 40, pp. 1439–1447, Oct. 1993. [8] L. R. Maciel, H. L. Bertoni, and H. H. Xia, “Unified approach to prediction of propagation over buildings for all ranges of base station antenna height,” IEEE Trans. Veh. Technol., vol. 42, pp. 41–45, Feb. 1993. [9] U. Dersch and E. Zollinger, “Propagation mechanisms in microcell and indoor environments,” IEEE Trans. Veh. Technol., vol. 43, pp. 1058–1066, Nov. 1994. [10] A. J. Rustako, N. Amitay, G. J. Owens, and R. S. Roman, “Radio propagation measurements at microwave frequencies for microcellular mobile and personal communications,” in IEEE ICC’89, Boston, MA, pp. 482–488. [11] L. B. Milstein et al., “On the feasibility of CDMA overlay for personal communications networks,” IEEE J. Select. Areas Commun., vol. 10, pp. 655–667, May 1992. [12] J. Walfisch and H. L. Bertoni, “A theoretical model of UHF propagation in urban environments,” IEEE Trans. Antennas Propagat., vol. 36, pp. 1788–1796, Dec. 1988. [13] N. Blaunstein and M. Levin, “Prediction of UHF-wave propagation in suburban and rural environments,” in Proc. URSI Symp. CommSphere’95, Eilat, Israel, Jan. 22–26, 1995, pp. 191–200. [14] S. T. S. Chia, “Radiowave propagation and handover criteria for microcells,” British Telecom Tech. J., vol. 8, pp. 50–61, Aug. 1990. [15] N. Blaunstein and M. Levin, “VHF/UHF wave attenuation in a city with regularly distributed buildings,” Radio Sci., vol. 31, pp. 313–323, 1996. [16] A. J. Rustako, N. Amitay, G. J. Owens, and R. S. Roman, “Radio propagation at microwave frequencies for line-of-sight microcellular mobile and personal communications,” IEEE Trans. Veh. Technol., vol. 40, pp. 203–210, Feb. 1991. [17] S. Y. Tan and H. S. Tan, “UTD propagation model in an urban street scene for microcellular communications,” IEEE Trans. Electromagn. Compat., vol. 35, pp. 423–428, Nov. 1993. Nathan Blaunstein received the B.Sc. and M.Sc. degrees in radiophysics from Tomsk University, Tomsk, Russia, in 1972 and 1976, respectively, and the Ph.D. and D.Sc. degrees in radiophysics from the Institute of Geomagnetism, Ionosphere, and Radiowave Propagation (IZMIR), Academy of Science USSR, Moscow, Russia, in 1985 and 1990, respectively. Since 1993, he has been a Senior Scientist of the Department of Electrical and Computer Engineering and a Visiting Professor in the Wireless Cellular Communication Program at the Ben-Gurion University of the Negev, Beer Sheva, Israel. His research interests include problems of radio-wave propagation, problems of diffraction and scattering in various media for purpose of mobile-satellite, and terrestrial and mobile communication systems performance and services. Ran Giladi received the B.Sc. degree in physics and the M.Sc. degree in biomedical engineering from the Technion, Israel Institute of Technology, Israel, in 1980 and 1983, respectively, and the Ph.D. degree in management information systems from Tel-Aviv University, Tel-Aviv, Israel, in 1992. He is the Head of the Department of Communication Systems Engineering at the Ben-Gurion University of the Negev, Beer Sheva, Israel. He cofounded a multinational data communications firm and is involved in the Israel data and telecommunications industry. His research interests include computer and communications systems performance, data networks and communications, and network management systems. Moshe Levin received the B.Sc. degree in electrical engineering from the Technion, Haifa, Israel, in 1981, and the M.Sc. and Ph.D. degrees in electrical engineering from Tel Aviv University, Tel Aviv, Israel, in 1986 and 1989, respectively. From 1986 to 1989, his research at Tel Aviv University focused on signal processing in radar systems. From 1989 to 1992, he worked at the EltaIsraeli Aircraft Industries in signal and radar system. He is currently Chief Scientist of Wireless Systems at Tadiran Telecommunications, Ltd., Petakh-Tikva, Israel. His research interests include signal processing, radar system, and wireless communication systems.