UWB communication primarily derives its roots from its forerunner, the Impulse Radio (IR) technology, which has been in use in military radar applications for several years. UWB signals are defined as any signal with a bandwidth greater than 20% of its centre frequency or a bandwidth greater than 500 MHz. The FCC regulation has specifically demarcated the frequency band 3.1-10.6 GHz for UWB communications and measurement systems.

This paper investigates the line of sight (LOS) path loss characteristics of the Ultra Wideband (UWB) channel by explicitly focusing on a very short range of 5 m. A vector network analyzer was used to measure the channel transfer function in the frequency band of 6-8 GHz. It has been observed that the published path loss models for UWB communications report a path loss exponent in the vicinity of the benchmark '2' for LOS indoor measurements for 1-10 m. However, as one of the potential applications of UWB technology will be for Personal Area Networks (PANs) such as wireless desktops, it becomes imperative to characterize the UWB channel for the same.

The measurement setup consists of a Rhodes and Schwarz ZVB 8 Vector Network Analyzer (VNA), a wideband antenna pair, and a PC for storing and post-processing of measured data. Measurements have been taken in a lab environment and were performed on weekends with no movement whatsoever during the recording time. The transmit antenna (Tx) was kept stationed at a fixed position while the receive antenna (Rx) was moved on a measurement grid throughout the lab with the help of a trolley. The height of transmit and receive antennas was kept 1.27 m above the ground. Since the measurements were for short-range indoor scenarios (wireless desktop environment), a transmitter-receiver distance ranging from 1-5 m was selected for LOS paths. The measurements were performed at uniform steps of 0.5 m. Around each local point measurements have been performed on a spatial grid consisting of 20 square blocks of dimension 50x50 mm. A set of 25 frequency response snapshots was recorded for each spatial point. The measured database consists of 20 channel complex frequency responses for each local point and 25 complex frequency responses for each spatial point. Analysis of 25 channel frequency responses at each spatial point verifies the time-invariant nature of the indoor UWB channel

We have calculated the overall path loss exponent, no by fitting a regression line through the spread of measured path loss points such that the root mean square deviation of path loss points about the regression line was minimized. The value of no was found to be 1.85. The value of standard deviation was found to be 1.01 dB. It was also observed that path loss as a function of distance has two distinct regions. In the first region of 0-4 m the path loss exponent value is within value of 2. At distances greater than 4 m the propagation loss increases significantly. The distance dependent path loss exponent values calculated are 1.55, 1.97, 1.98 and 3.35 for 1-2 m, 2-3 m, 3-4 m and 4-5 m respectively.

The electric conductivity of the ground is an important parameter in the prediction of the AM radio broadcasting's covering area and becomes critical, mostly for planning the digitalization of this service. In practice, the value of the conductivity is difficult to obtain, due to the inherent complexities of the measurement process or operational cost involved. In Brazil, the ground conductivity, extracted from the ITU-R P832-2 recommendation, is equal to 1 mS/m in Brasilia's surroundings, and it is been used in all covering area predictions till now. This value, however, is estimated and was neither gotten nor proven by means of field measures of any kind.

In this paper, a methodology for the determination of the electric conductivity of the ground in rural environment is proposed, using the attenuation of the surface wave field. It has been carried the radiated electric field intensity measurements through eight radio stations operating in medium waves, located at the central region of Brazil. The data have been collected from throughout six radial routes, uniformly distributed, with 180 km radius each. For the accomplishment of the measures, it was used a mobile unit of the National Agency of Telecommunications (ANATEL), which is responsible for the regulation of the telecommunications services in Brazil. This vehicle possesses calibrated equipment that makes possible the acquisition of the field intensity samples at each 500 ms, with the respective geographic coordinates.

The proposed methodology is used for the comparison of the electric field intensities with the results obtained by means of the surface wave propagation on spherical earth model. Different values of the conductivity are used during this evaluation. In the theoretical model, the electric permittivity of the ground is considered constant, whereas in practice it is variable. Therefore, the electric permittivity being constant can cause differences in the values of the conductivity. Then, an analysis of the introduced error was carried through, where the value for the estimate conductivity can be validated.

The figure bellow illustrates the result for one of the evaluated Brazilian routes. It shows an average conductivity equal to 2.8 mS/m, different from the ITU recommendation. It means that the area of the current covering AM station is greater than the estimated by ANATEL. We believe that this methodology is important and must be considered in the planning of the implantation of a simulcast broadcasting system in Brazil.

In this paper we use the uniform geometrical theory of diffraction (UTD) to model distortion of a particular UWB pulse in the channel containing hill. The distortion of UWB pulse is difficult to model using statistical measurements, therefore it is better to do this in a deterministic way, as it is in the case of the uniform geometrical theory of diffraction. UTD related to electromagnetic wave scattering by convex surfaces in frequency domain was presented in literature (e.g. the papers of H. Pathak). Basing on this theory we derive in this paper a 2-ray model of the channel containing hill by developing its transfer function H(f). The transfer function was decomposed into two parts H(f)=H₁(f)+H₂(f). The first part H₁(f) can be transformed by means of Laplace transform to the time domain, giving impulse response h₁(t). The second part H₂(f) containing special Fock function, which occurs in the formulation of UTD for convex surfaces in frequency domain, can not be transformed to the time domain in an analytical way. Therefore we proposed to use an approximation of the transfer function H₂(f) by rational function. To do this we used a new, very effective method called "vector fitting". In this way we finally obtain the closed form formula for impulse response of the channel of the following form: h(t)=L^-1[H₁(s)]+∑c_ne^a_nt +dδ(t). The time domain solutions are in very good agreement with the inverse fast Fourier transform (IFFT) of H(f) – based frequency domain solutions.

The present paper is devoted to the study of penetration through finite-thickness cylindrical and planar structures made of artificial materials, which show plasma-type frequency dependence of the effective permittivity eeff . Our main motivation concerns engineering the structures with enhanced frequency and/or angular selectivity. It has been shown in J. B. Pendry et al., Phys. Rev. Lett., v. 76, 4773-4776, 1996 that plasma properties of metals, for which the plasma frequency is typically equal to 103-104 THz, can be scaled down to microwave frequencies, by using periodic wire media. In this case, eeff shows the frequency dependence of the same type as that for a Drude metal. The plasma-type behaviour at microwaves can also be obtained in a disordered wire medium, provided that the long-wave approximation holds. We studied in detail the effect of frequency and geometrical parameters on the penetration through two-dimensional planar slabs and circular rings, which are made of a Drude-type material. Calculations have been performed for both polarizations of the incident wave. For the planar slab, the emphasis was on the possibility of scaling the effect of angular cut-off, which is well-known for x-rays and at the visible (e. g., see B.T. Schwartz and R. Piestun, J. Opt. Soc. Am. B, v. 20, 2448-2453, 2003), down to smaller frequencies. This effect is associated with the phenomenon of total internal reflection, and hence needs 0 Re{ eeff } is (almost) zero, so that the field inside the slab is a (close-to-) static one. In case of the ring shell, the main operation regimes have been studied by varying the parameters in a very wide range. In particular, the abilities of enhancing, weakening, and reshaping the field inside the shell have been studied in subwavelength and resonance regimes. Examples of the field pattern in case of a ring shell with outer radius a and inner radius b, which is illuminated by a plane wave, is shown in Fig. 1 for three sets of values of b/a, eeff, and k0a where k0 is free-space wavenumber. Here the following effects are demonstrated: (left plot) transformation of plane wavefront to symmetric, azimuthally-independent field inside the shell, which takes place due to tunnelling through an almost-zero-permittivity wall, (middle plot) whole-volume shielding, and (right plot) strong enhancement of the field arising simultaneously with the excitation of the surface waves at the ring interfaces. The mentioned effects can appear in the same ring structure for different frequency values, so that very strong frequency selectivity can be realized. Besides, situations have been detected, for which a shielding effect occurs for both polarizations at fixed frequency. The performance of circular and planar structures, in which the above-discussed effects can be obtained, has been discussed in the context of some microwave and quasi-optical applications.

Figure 1. Demonstration of several effects arising in the ring shells made of a Drude material: (left) ee ff=0.000001-i0.0045 and k0a=2.5; (middle) e eff=-38-i24.7 and k0a=2.5; and (right) e eff=-12.84-i5.55 and k0a=6.27. In all plots, b/a=0.8. In the left and middle plots, the electric field vector is directed along the longitudinal axis, in the right plot the magnetic field vector does. Tones of gray from black to white correspond to |Ez| varying from 0 to 1.8|E0| in the left plot, and from 0 to 2.25|E0| in the middle plot, and to |Hz| varying from 0 to 160|H0| in the right plot, where |E0| and |H0| mean amplitude of the incident wave. Dashed lines show the boundaries of the region filled by a Drude material

Inter-symbol interference (ISI) in high data rates communication system and correlation in systems using spatial diversity are the major problem. Indoor Ultra-wideband (UWB) measurements were done from 2 to 4 GHz to determine the spatial correlation and the time delays. These frequency-domain measurements were converted to the time-domain through the inverse Fourier transform (IFT) giving impulse responses. The time delays were then extracted from the impulse responses. A virtual array is also used at the receive side. A spatial resolution of 5 cm was achieved. The spatial autocorrelation was determined by calculating the correlation between positions on the array. Two receive antennas were also measured simultaneously on the receive side to determine the cross correlation to validate the autocorrelation results. It was found that the autocorrelation and cross correlation give comparable results and very low at moderate antenna spacing. Antenna coupling was found to have no noticeable influence on the results. Measurement set up and measurements results-The measurement setup is shown in Figure 1.a including the PC, vector analyzer, antennas, amplifiers, cables and switch. Two medium sized bi-conical antennas on the receive side and one larger bi-conical antenna on the transmit side were used. The two bi-conical antennas are placed in a frame in which they can be moved along one axis. A virtual array was constructed in one dimension by sliding the antennas back and forth. A switch was used to sample the two channels one at a time. For each position this was done 128 times, measuring each channel 64 times for averaging. The channel was kept stationary within the time that it took to complete all measurements. From the indoor measurements the time delay and auto- and cross correlation was determined. Figure 1 shows the measurement set up, time delay and autocorrelation function.

Figure 1 (a) measurement set up, (b) RMS time delay as function of Tx/Rx distance, measured in three different environments, (c) Autocorrelation measured in two lab locations and two shop environments.
Conclusions-It can be concluded that the method introduced in this paper for determining the spatial correlation for a wideband channel is valid. Spatial correlation is usually determined for a single frequency. The method introduced here uses the impulse response derived from frequency-domain measurements through the IFT. This makes it possible to determine the spatial correlation using frequency-domain measurements which are similar to time-domain methods. It was concluded that antenna coupling had no noticeable influence on the correlation. In an indoor environment the spatial correlation can be extracted with a single receive antenna using a virtual array. This was validated by the cross correlation results from the dual channels. It is shown that the correlation is low for moderate antenna spacing. The mean delay spread for different locations was between 22-27 ns.

Automobile access systems gain on importance not only from the convenience point of view but also for security reasons. Remote Keyless Entry (RKE) has become meantime standard for locking and unlocking the vehicle. By means of radio communication the system ensures that the vehicle doors and the trunk are locked and unlocked.

Remote keyless systems consist of a key fob transmitter and a receiver in the vehicle (Fig.1a). For a good functioning of the system, reliable transmission has to be assured. Therefore, it is necessary to characterize the radio frequency propagation channel in order to be able to predict the power at the receiving antenna within the vehicle. The receiver depicts, depending on his position in the car, the directly propagating wave, the wave reflected from the ground and waves reflected within the vehicle. For the evaluation of the received power, the influence of the vehicle nearest environment (e.g. ground properties), the position of the module within the vehicle, the distance d between the key and the receiver, the height of the key above ground have to be condisered. The key issues of interest in this paper are the physics of propagation around the vehicle, implications for modeling and the link performance.

The communication between the key and the receiver takes place in ISM (Industrial, Scientific, Medical) frequency band. The most commonly used frequencies in Europe are 434MHz and 868MHz. Experimental results indicate that surface scattering becomes particularly significant or even dominant when the surface roughness parameters are in order of a wavelength. In above given frequency range with the wavelengths of 69cm and 35,4cm, an assumption of a flat surface has been made. This assumption is not very far away from the car real environment - case of a free parking slot or an empty street. To be able to estimate only the path loss without the influence of the car body the problem has been simplified to the case of two antennas: horizontally and vertically polarized Dipoles Fig. 1b. The transmission factor for grounds with different values for the electrical properties of the ground surface: relative permittivity and the loss factor have been calculated.



Fig 1. a) Path loss diagram for RKE system b) Dipoles arrangement over ground

In the full paper version theoretical formulas and the curves for the path loss over a given distance will be presented, simulation results will be validated with the measurement results for the above problem with different ground types: asphalt, perfect electrically conducting (PEC) ground.

The unquestionable advantages of multiple-input multiple-output (MIMO) systems are having a strong influence on the development of new wireless systems, both in local area networks (WLAN) and in those designed to offer wireless broadband access services in metropolitan areas (WMAN). In fact, the IEEE 802.16d (June 2004) standard for WMAN systems, already acknowledges the option of using two transmitter and one receiver antennas together with the simple and efficient space-time coding system proposed by Alamouti in [1].

The access technology considered in these new systems is Orthogonal Frequency-Division Multiplexing (OFDM). The increase in link throughput and the reliability of such systems in NLOS environments can be reached by the combination of MIMO with OFDM. In previous works [2,3], a 2x2 MIMO channel characterization, focused on the impact of spatial correlation on the attainable capacity, was presented by using a narrowband measurement system. In the present work we extend these contributions to the case of wideband systems, specifically, we characterize the outdoor-indoor 2x2 MIMO-OFDM channel for a 20 MHz band at 3.5 GHz.

The measurement campaigns were conducted in an outdoor-indoor environment at the University of Cantabria. The measurement of the 2x2 MIMO-OFDM channel was carried out in the frequency domain. The measurements were made using linear antenna arrays. The two arrays were located in two different buildings separated by a mean distance of 120m. The transmitter array was in a fixed outdoor position, while the receiving one was moved through three different indoor scenarios. The transmitter is based on a Digital Signal Generator (Agilent ECG4433B) and the receiver consist of a two port Vector Network Analyzer (PNA E8362A), which allows the channel matrix entries be measured using a switching method.

Among other wideband parameters of great interest for the OFDM system design, the measured channel data is used to calculate the capacity CDF in different local areas, as shown for instance in Figure 1. The evaluation of link capacity can provide a valuable measurement of the channel's upper bound limit in achievable data rate. The theoretical limits in Figure 1 are obtained considering spatial uncorrelated channels; the lower limit assumes that the channel is correlated for all the frequency tones, that is the channel is frequency -flat and frequency diversity is not available; the upper limit correspond with the opposite situation in which a very low correlation between different tones is assumed. The experimental capacity curve is nearest to the upper limit showing that frequency diversity provides an appreciable capacity gain in real outdoor-indoor channels.

[1] S. Alamouti,"A simple transmit diversity technique for wireless communications," IEEE Journal on Selected Areas in Communications, vol. 16, Oct. 1998, pp. 1451-1458.

[2] Richard Jaramillo, Oscar Fernแndez, Rafael P. Torres "Empirical Analysis of 2x2 Outdoor-Indoor MIMO Channels for FBWA Applications" accepted for publication in the 13th IEEE MELECON 2006 May Spain.

The objective is to carry out an experimental study of the rapid fading and enhancements, (scintillations), simultaneously experienced by two Ka band receivers 9 metres apart. The signal source is, in this case, the beacon transmission at 20.7 GHz of the UF09 geosynchronous satellite from the US American Defence Department.

The figure illustrates the phenomenological process whereby a satellite plane wave entering the lower part of the turbulent troposphere evolves into a โ€œcrinkledโ€ wave in which, over a plane, the amplitude A of the field changes as a function of the position, i.e. we have an A(r), where r stands for position vector. As time goes-by, A (r) becomes a partly deterministic and partly random time-variant function A(r,t). Therefore, two receivers closely co-located may, or may not, experience simultaneous enhancements (or fades). The process of fading and enhancement at a given location, r, is due to a complex process of random diffraction (often treated also as scattering for convenience) caused by some form of regular โ€œstructurisationโ€ in space of the atmospheric refractive index n(r). These structures are distributions of turbulent eddies generated either by buoyancy (literally warming of the ground air and thus raising the lower atmosphere and mixing randomly upwards) or by instabilities of the horizontal laminar flow of the air.

The two receivers have a single down-conversion 20.7GHz - 70MHz and use a 15cm diameter horn antenna, giving a receiver temperature of 300K. The receivers are located at the UK Chilbolton Observatory in Hampshire. On-site digital signal processing permits display of the main functions investigated (time series, cross-correlation, power spectral densities, standard deviations, etc.). Fast data storage takes also place on-site, and on-line web control permits remote data monitoring, acquisition and presentation of results. Other data acquired included ground temperature and wind velocity.

As regards results, analysis of the running cross-correlation C(A1(r,t),A2(r+9,t)) indicates that for a non-negligible amount of time there is a significant negative correlation between A1 and A2, permiting to consider some form of dynamic diversity gaim over few dBs. Often however, both signals are either hardly correlated (including the case of rain), or reasonably well correlated.

The main paper will give the salient results of cross-correlation and scintillation intensity within the contex of the inner structure of the random time variant diffraction paterns anlysed.

In conclusion:

In this paper, a wideband indoor MIMO (Multiple Input Multiple Output) system at the 2.26-2.5GHz band is investigated. Our main purpose is to study the potential correlation between two different "slices" (portions) of the transmitted bandwidth, each occupying 60MHz. The semi-sequential digital chirp sounder at University of Durham (UK) offers the desired bandwidth compression for direct data acquisition and a high Waveform Repetition Rate (WRF) that yields adequate unambiguous Doppler coverage. The semi-sequential architecture employs parallel receive channels and an RF switch at the transmitter which switches sequentially between all transmit antennas. A (4 x 4) MIMO system with omni-directional discone antenna elements at both ends of the link was deployed. The transmit elements were linearly mounted (Fig. 1a) at a spacing of half a wavelength (6.7cm), while the receive elements were mounted around a circle of radius equal to 6.7cm too. The transmitter was placed on a moving trolley while the receiver was stationary.

The measurement campaign was conducted on two different floors in the school of Engineering, University of Durham in an area with many office partitions.

The (4 x 4) channel transfer matrix, containing all complex responses between each transmit and receive element, can effectively model a wideband MIMO system. By considering the 4 x 4 = 16 discrete (element-to-element) complex correlation coefficients, the auto/cross-correlation matrices in frequency domain can be constructed at each transmitter location. More specifically, both matrices were generated by calculating the corresponding correlation coefficients (averaged over the time domain) for each pair of frequency bins. Taking into account that 500 frequency samples were recorded, the correlation matrices, containing all possible correlation duets, are of size (500 x 500). In Fig. 1b, the amplitude of the auto-correlation matrix of the upper band is illustrated, for a certain transmitter location (same floor). The matrix is inherently Hermitian with the ones on the main diagonal expressing the auto-correlation coefficients at the same bin.

The data processing also revealed that the degree of correlation varies significantly across the 16 MIMO subchannels (expressed by the transfer matrix elements) as long as they exhibit different propagation characteristics. Moreover, by wheeling the transmitter to the predetermined locations, the overall radio link performance, and consequently the experienced frequency correlation, was extensively affected by the surrounding environment. A detailed analysis of the underlying mechanisms that determine the auto/cross-correlation functions, along with the related mathematical background, are presented in the full paper. A comprehensive capacity study is also conducted by employing the uniform power allocation and the water-filling scheme which yields higher channel capacities. Finally, a complete Ricean K-factor classification of the measurement environment is also included in order to quantify the propagation conditions.

During last years the rapid growth of number of the cellular base stations, satellite and different type of antennas, which are representing radiation of high-powered electromagnetic (EM) field sources were observed. For today the research of their influence on human's health is one of the most common problems in area of EM Compatibility. Except for the human, the given sources of electromagnetic radiation also can essentially affect to the sensitive electronics in cars that in turn can lead to breakdown during driving. Special international committees are organized to minimize, optimize it all over in order to establish allowable limits and health safety standards of ลฬ background level. According to recommendations of the scientists involved in the given researches, most acceptable and optimal would be necessary level uniform distribution of radiated energy of base stations in environment. Given recommendations are intended for realization of stable wireless communication and for reduction the health's risk. For description of a character of EM field propagation in invisible frequency range in urban conditions the computer modeling for simplified scenarios is applied.

In this paper the numerical investigation of a character of EM field propagation in outdoor and coupling indoor structures (i.e. in structures with high probability of strong resonance appearance) when the power density in some areas is capable to increase many times are proposed. Basically the models of room and cars are utilized. Following scenarios for the room model were considered: 1) closed room, 2) room with one window, 3) room with two windows, 4) room with two windows and door, 5) room with outside standing surface. For incident source plane wave was chosen. As car models the model of a jeep and a passenger car with the following scenarios have been considered: the car with incident plane wave falling on it along various directions and the car with the radiating antenna (dipole) placed on its top. Several EM characteristics of objects were investigated: EM field distribution, the behavior of induced currents on the surface, resonance characteristics and so on. The numerical results and their visualization are represented also.

As mathematical tool for solving current problem the Method of Auxiliary Sources (MAS) is applied, as, it is the most optimal for solving the state problems with homogeneous models.

Broadband Fixed Wireless Access (BFWA) networks are a promising wireless solution to connect fixed users to the backbone network instead of broadband-wired networks. They are point-to-multipoint cellular networks operating at millimeter waves in the 20-50GHz. In these frequencies, higher antenna directivities are exploited and greater communication capacities are achieved. At the same time, the radiowave propagation is strongly influenced by rain attenuation, which is a major factor limiting the performance of radio links and the cell coverage of BFWA systems. Intrasystem interference, among adjacent cells, in BFWA networks, limits the capacity in the BFWA cellular network for both the downstream (from base station to subscriber) and the upstream (from subscriber to base station) [Panagopoulos et al, IEEE Trans. on Vehic. Technol., 2006]. In this paper, we will employ a dual frequency and polarization reuse scheme and a TDMA scheme will also be adopted, which is consistent with IEEE 802.16 standard. To measure the performance of intercell interference, the carrier-to-interference ratio (CNIR) must be calculated. In this paper, the worst-case interference scenarios are analytically demonstrated and are taken into account. Our objective is the evaluation of the fraction of the time where the capacity distribution of a broadband fixed wireless access channel under rain fade conditions suffering from co-channel interference, non-exceeds a specified level in (bps/Hz). The proposed analysis examines the statistical properties of the CNIR focusing on the spatial inhomogeneity of the propagation medium and the correlated terrestrial paths. The lognormal distribution is adopted for the rain attenuation random variables, and the convective raincell model is also assumed to estimate the spatial correlation coefficient of the microwave paths. Using the proposed the channel capacity distributions as figure of merit and by employing adaptive fade mitigation techniques (adaptive modulation, adaptive coding) the optimum utilization of the channel capacity can be achieved.

Numerical results obtained from the proposed statistical interference analysis, focus on the impact of frequency of operation, the subscriber's service availability, the cell coverage radius, and the climatic conditions on the capacity distribution of broadband fixed wireless access channels being interfered by other terminal. Some useful conclusions are deduced for the reliable design of broadband fixed wireless access networks.

Microwave propagation on line-of-sight links has been a subject of many investigations. There are many papers dealt with that problem and there are many methods of theoretical and ex-perimental investigations of microwave propagation processes. The most advantageous is homo-dyne method, which is described in some papers (I.B. Shirokov "New Approach to Equipment for Angle-of-Arrival Measurements on Microwave", IGARSS'2002, Toronto, Canada, 24-28 June 2002; I.B. Shirokov "The Determination of Microwave Propagation Mechanism on Line-of-Sight Links", IGARSS'2003, Toulouse, France, 21-25 July 2003). The main advantage of these investiga-tions is the carrying out of phase progression measurements on microwave line-of-sight links. Such approach lets us investigate "thin structure" of electromagnetic field and expand our knowledge about mechanisms of microwave propagation.

In this paper the measuring equipment is described and its technical characteristics are dis-cussed. So, at the first part of testing link there are placed several independent measuring units. Each unit consists of microwave oscillator, circulator, antenna, mixer and several low frequency selective amplifiers with different working frequency. Each unit is represented as a receiver-transmitter device operating with the same antenna. Here the transmitter signal is the heterodyne signal for the receiver. All of microwave oscillators are characterized by the closely spaced but dif-ferent frequencies and are not synchronized. The number of the units can be chosen ad arbitrium and is characterized by the measurement conditions. In addition, their mutual setting is not hard and can be changed even during the measurement process. In each unit there are picked out several low frequency signals and then they are delivered from all units through low-frequency feeders with ar-bitrary phase stability to the processing device, where there are measured amplitude and phase dif-ference (one corresponds phase progression) fluctuations of receiving microwave signals. In a paper the problems of signal conversion and phase difference measuring are discussed. On the other side of the testing link there are placed several retransmission units. Each re-transmission unit receives all off signals from first part of testing link by its antenna, shifts the fre-quency of signals by the same value, which is different for different retransmission unit and then radiates theirs back in the direction of the measuring units. In a paper the problems of frequency shift, transferring of initial phase of low frequency oscillations from one part of testing link to an-other and then transferring one on microwave are discussed. Furthermore, it is discussed problem of synchronization of several low frequency signals.

As well as we assume, that on both parts of testing links there are placed several units, then there is a good opportunity to carry out not only angle-of-arrival of microwave measuring, but we can investigate complex mechanism of microwave propagation.

In conjunction with synchronous meteorological measurements, uncial method of amplitude and phase progression investigations let us have a clear idea of microwave propagation in natural medium.

The aim of the paper is models development and experimental validation of millimeter waves narrow beam propagation in shadow zone of the urban building. Establishment dependence of interference fading from parameters for a narrow beam, height of antennae, distance between antennas by propagation above terrestrial cover in rural conditions.

The theoretical task was solved for narrow Gaussian beam with Babinet principle use, i.e. the complex amplitude radiation of the aperture silhouette was subtracted from complex amplitude of the transmitter field on the receiver antenna plane. The transmitter was the parabolic antenna and the conic horn was used as the receiver. The geometrical scheme atmospheric radiochannel is shown on fig.1.

Consider a millimeter-wave beam radiated by an antenna with a circular aperture. When the beam is reflected by terrain, the complex amplitude of the field in the wave zone is the superpo-sition of the complex amplitudes of the primary and reflected beam fields. In this study, experimental data on the observation of two-beam fading when a beam propagated over an asphalt road in the suburbs of Denver are employed. The geometry of the propagation path is schematically shown on fig.2.

Conclusions:

Ultra wideband (UWB) technologies have been expected for use in ultra-high-speed wireless personal area networks (WPAN) and wireless body area networks (WBAN). UWB human electromagnetic phantoms are useful for performance evaluation of antennas mounted in the vicinity of a human body, or when the human body blocks a propagation path. No UWB phantom, however, has been realized so far. The UWB phantoms are not intended to evaluate specific absorption rate (SAR) in a human body, because UWB devices are supposed to transmit very low power and thus no human hazard is expected. This paper describes the development of UWB arm and body phantoms. The liquid phantom materials have been developed to simulate the relative complex permittivity of high-water-content tissues (for example, muscle), whose data were given by Gabriel et al. [Phys. Med. Biol., vol. 41, no 11, 1996]. An exhaustive search was carried out to find the optimal recipe among various sucrose solutions containing sucrose, sodium chloride, various monohydric and polyhydric alcohols. Among the recipes investigated, an aqueous solution of sucrose (C12H22O11, 1.0-mol/l) provided the best result within a UWB bandwidth of 3.1 to 10.6 GHz [Hara et al., 13th Conf. Microwave Tech., Sept. 26-28, 2005, Prague, Czech, Republic]. The errors in the complex permittivity were 3.9% root-mean-square (rms) and 10% at the maximum in the real part and 4.4% rms and 30% at the maximum in the imaginary part. The effects of the errors were numerically analyzed, assuming typical shapes of reflecting and scattering objects (an infinite plane, a sphere, and a cylinder). Calculations revealed that this recipe resulted in negligibly small errors in UWB reflection powers and scattering patterns, while it produced an 18% error in local SAR evaluation in comparison with the real muscle tissues. The arm and body phantoms were developed using this liquid recipe. The container of the arm phantom was an acrylic cylinder, 500-mm in length and 64-mm in diameter; and that of the body phantom was a 20-liter polyethylene nearly-parallelepiped water tank. These containers were filled with the 1.0-mol/l sucrose solution. Comparison was made between three human volunteers and the phantoms in typical propagation scenarios. UWB (3.1 to 10.6 GHz) path losses and delay profiles were measured with a vector network analyzer. The arm phantom was tested on a metal plate simulating typical office desks (desktops will be the most common environment where WPAN is employed). A 500-mm propagation path was partially or fully blocked by the human arm or the arm phantom. The errors between the real arms and the arm phantom in additional loss caused by the blocking were within 1.0 to 1.2 dB rms. The delay profiles were also essentially similar. The body phantom was compared with three volunteers in a situation where WBAN is employed: one antenna (omnidirectional) was mounted on the waist of volunteers or the side of the phantom; and the other antenna (ridged waveguide horn) was placed 3-m apart in a radio anechoic chamber. Propagation losses were measured between the two antennas, while rotating the human bodies or the phantom azimuthally. The rms errors were within 1.0 to 1.4 dB. The delay profiles were also essentially alike. It was concluded from the comparison that the arm and body phantoms properly simulated the respective parts of the human body and were usable for experimental studies of UWB antennas and propagation.

There is a growing interest in time reversal of electromagnetic waves as it was originally developed in acoustics and showed very interesting features. In this paper several simulations and an experiment on time reversal focusing with electromagnetic waves are presented.

For simulations, a FDTD 3D electromagnetic simulator is used. Ten half wavelength dipoles operating at 2.45GHz are simulated. In the first step, one of them is used like a source and the others are sensors. The excitation is a modulated Gaussian pulse of 2.45GHz centre frequency and 2GHz bandwidth. During the first step, all sensors record signal sent by the source. These signals are time reversed and then reemitted to the initial source which becomes the target. This target (dipole) is placed at different positions in order to see the spatial focusing. At the initial position, the signal is higher than at the others positions. However, because of the small bandwidth of dipoles comparing with the initial signal bandwidth, initial signal and focused signal are different. Moreover, following the target position coupling between all dipoles is different. Therefore, the receiving antenna impedance is modified and the received power (05*Re(V*I*)) is not automatically maximum for the first position : 1W (normalised) for the first position, 1.1W for the second and 0.4W for the third.

For experiment, a one-channel electromagnetic time-reversal mirror working around 900MHz is presented. Our experiment takes place in a metallic cavity. Two dipoles, operating at 900MHz, are used. The experiment takes place in a reverberation chamber. The two dipoles are placed in line of sight. An arbitrary waveform generator (from DC to 2.1GHz) and a digital storage oscilloscope (up to 12GHz) are used. The former allows to deliver a short pulse (900MHz central frequency and 200MHz bandwidth) with a 4.2Gs/s sample rate, and the latter is used to observe and to record the received signal. The pulse is composed of about 35 points and its length is lower than 10ns. This modulated Gaussian pulse is sampled at 4.2Gs/s.
The recording signal is time-reversed and then reemitted. A resample is needed to be compatible with the generator sample rate. An algorithm is used to down-sample until 4Gs/s. A 50dB amplifier is placed between the generator and the emitting antenna. Both the reemitted signal (cyan) and the focused signal (yellow) are shown in Figure 1. We can see that the wave is focused on the initial source. Moreover, if one of the antennas is shifted, the amplitude of the recreated signal is lower: there is a temporal and a spatial focusing.
A comparison between the shape of the received signal (after the reversing step) and the shape of the initial signal shows that there is a good agreement between these two signals. This experiment shows that time reversal is able to compensate for complex multipaths of reverberant environments and focus in space and time an electromagnetic pulse.

In Line-of-Sight (LOS) propagation conditions, a GNSS (Global Navigation Satellite System) receiver tracks several satellite signals. Each signal consists of a direct ray, and of diffracted and reflected rays called multipath. This multipath causes an error in Pseudo-Range (PR) measurement, but mitigation techniques in the receiver can reduced this error. When no direct ray is present, these techniques are not efficient and the PR error can become very large. This paper presents the determination of the PR error due to Non-Line-Of-Sight (NLOS)-Multipath in urban canyons.

The urban canyon model consists of two parallel brick walls, modelled by thin dielectric plates. The height profile of the walls is generated from a Rayleigh distribution. No direct power transmission is considered when a satellite is below the first wall. By using all satellite positions during one day in Belgium, a specific ray tracing method is applied in the urban canyon in order to determine diffracted and reflected rays. The receiver is placed in the middle of the urban canyon and is equipped with an ideal omnidirectional RHCP (Right Hand Circular Polarization) antenna. A power threshold of -13dB below which the receiver cannot track the signal has been used. The RHCP Power, the delay and the phase of each ray reaching the receiver are determined and constitute the channel impulse response.

In order to obtain the PR error, a first approximation has been used by considering the maximum power ray as the only one tracked by the receiver. The delay associated to this ray is assimilated to the final PR error. Simulations were carried out in four different urban canyon configurations based on Guildord and Westminster topological data’s. It has been shown that the ray with maximum power is generally the one diffracted by the first wall or the single reflected ray. The PR error distribution typically shows three parts, an important mode below one meter corresponding to diffracted rays, a nearly uniform part between one meter and the half width which mainly corresponds to single reflected rays and a last part corresponding to the other maximum power rays.

Secondly, a receiver model is used in order to determine the true PR error due to NLOS-Multipath. The receiver model uses a coherent Early-Late DLL with a unity space chip and a PLL with a ๐ phase ambiguity. The first results show that the PR error distribution is a mixing of a Rayleigh distribution with a mode below 1m corresponding to the diffraction by the first wall. By calculating the difference between the PR given by the maximum power ray method and the one given by the E-L receiver model, the distribution of this difference is a combination between a normal distribution and a mode on the middle of this normal.

In conclusion, by using the maximum power ray, it is shown that the single reflected ray and the diffracted ray by the first wall have the main contribution in the PR error distribution when no direct ray is present. By using an Early-Late receiver model, the true PR range distribution shows an important mode corresponding to the ray diffracted by the first wall. However, the effect of this diffracted ray is probably smaller when the receiver moves.

Based on the measurement at 4.5 GHz, the relationship between street direction and the propagation mechanisim at a mobile station (MS) in an urban macrocellular environment are analyzed. Figure 1. shows the measurement site near our university.

The measurement was divided into segments of 10 m. The Rx was moved manually in each segment, where consecutive snapshots were measured. The average number of snapshots in one segment is approximately 30. The Tx and Rx act as the BS and MS, repectively, in a mobile macrocellular environment. We roughly classify the propagation mechanisms into four classes by using the azimuth direction of arrival (DOA) of detected incident waves.

The first, second, third, fourth classes represent waves from the front (-45 to 45 degs), left (45 to 135 degs), right (-135 to -45 degs) and back (-135 to 135 degs) of the Rx compared to the moving direction, which is set to be zero and shown by a yellow line in Fig. 1. Figure 2 shows the change of proportion of the total wave power in each class for the streets (see Fig. 1).

For street 1, which is parallel to the Tx axis( i. e. the south-north direction), the environment are NLOS (non line-of-sight) and OLOS (obstructed LOS). Even though all classes have comparable significance, some measurementt periods are occupied mainly by a certain class. For example, for the measurement period betweenn Rx3 and Rx4, the first class is significant.

From simple ray-tracing, we found that the waves of this class arrived at the Rx via the side street shown by a red arrow in Fig. 1. While the second and fourth classes are blocked by a high building at the right side. For street 2, which is perpendicular to the Tx axis, it is characterized by NLOS. The third class showing the direction to the Tx is dominant in this street.

The environment in street 3 are NLOS and OLOS. At about Rx29, the dominance of the second class is caused by a small side street (also shown by a red arrow in Fig. 1) leading to the Tx. For street 4, the environment are LOS with short periods of OLOS. This results in high power from the first class. In summary, we found that for parallel streets, all classes have comparable significance with the difference of less than 25 % in average proportion of power. Some human-made irregularities can create abrupt changes in the DOA profile at the Rx. For both LOS and NLOS dominated perpendicular streets, the DOA profile seems to be mainly determined by the direction of the Tx. Finally, polarization characteristics and their relationships to the DOA profile are also studied in this work.

Over the last decade the world has witnessed explosive growth in the use of wireless mobile communications. Looking around we find users with mobile phones, wireless PDAs, MP3 players, and wireless headphones to connect to these devices - a small testament of the impact of wireless communications on our daily lives. In addition the burst of new technologies such as Bluetooth and other short-range wireless communications are encouraging the further development of a wide variety of distributed wireless devices. Bluetooth is one of those short range wireless communication technology systems which aims at replacing many proprietary cables that connect one device with another with one universal short-range radio link. Recently, many mobile devices (e.g., mobile phones, PDAs, computer mice) with integrated Bluetooth modules have been introduced. Their wireless technology is used to transfer any kind of data onto these devices. Bluetooth devices operate in the industrial, scientific and medical (ISM) band at 2.4 GHz, and use 79 channels each occupying 1 MHz. Propagation of radio waves inside buildings is a very complicated issue, and it depends significantly on the indoor environment (office, factory) and the topography (LOS: line of sight and NLOS: non-line of sight). The statistics of the indoor channel varies with time due to movements of people and equipment. A survey of indoor propagation measurement and models, and electromagnetic propagation effects in will be presented. Analysis of indoor radio channel characteristics is important for the design and development of personal communication systems. In this paper we present FEMLAB simulation for the prediction of wave propagation for short-range wireless communication systems at ISM band in indoor office environments (see Fig. 1). We will assess the fading phenomenon by simulation for LOS and NLOS propagation scenarios, impact on propagation when the doors along the hallway are either opened or closed, and use measurement results to confirm our findings (see Fig. 2). In addition, there is very little in the open literature on the measurements and performance evaluations of antennas for live Bluetooth links. Thus, further investigations about several types of antennas that might be used for Bluetooth devices and their performance in office environments were needed. These results would provide an insight into the performance of antennas for other short-range indoor wireless technologies such as the emerging ultra wideband (UWB) radio communications. Finally, we investigate a SIMO antenna receive diversity system utilizing different combining techniques to improve the performance over a fading radio channel in a NLOS propagation environment. Simulation results show that a substantial diversity gain would be achieved by using this system (see Fig. 3). Fig. 1: Power distribution of the tested indoor environment. Fig. 2: Power profile when doors are open or closed Fig. 3: SIMO system with different combining methods

Abstract Imprecise propagation models lead to networks with high co-channel interference and a waste of power. In this paper, we aim to adapt a propagation model for an urban area of OMAN as we examine the applicability of Okumura-Hata model in Oman in GSM frequency band. The study was carried out for urban area, since measurements provided from OmanMobile were about the urban areas. The study helped to re-design better GSM network for the city area. We will accomplish this by investigating the variation in pathloss between the measured and predicted values, according to the Okumura-Hata propagation model for a cell in Salalah city. Then, we intend to modify the Okumura-Hata model according to the results obtained in our investigation. We will then verify our modified model by applying it for other cells and conclude the results. For the purpose the mean square error (MSE) was calculated between measured path loss values and those predicated on basis of Okumura-Hata model for an urban area. The path loss (after incorporating MSE) is shown for a cell in urban area in Figure: 1. The MSE is less than 6dB, which is an acceptable value for the signal prediction [1]. Therefore, the model gave a significant difference in an urban area that allowed necessary changes to be introduced in the model. That error was minimized by subtracting the calculated MSE (around 16dB) from the original equation of urban area for Okumura-Hata model. Modified equation was also verified for another cell in an urban area in Oman and gave acceptable results. [1] Jianhui Wu and Dongfeng Yuan, “Propagation Measurements and Modeling in Jinan City”, IEEE International Symposium on Personal, Indoor and Mobile Radio Communications, Boston, MA, USA, Vol. 3, pp. 1157-1159, 8-11 September 1998.

When solving electromagnetic rough-surface scattering problems, the effect of shadowing by the surface roughness often needs to be considered, especially as the illumination incidence angle theta approaches grazing incidence. Indeed, due to the surface roughness, only a part of the surface is illuminated especially for grazing angles phi=pi/2-theta. This phenomenon is characterized by the statistical shadowing function which gives the probability that a point on a rough surface is illuminated. In this paper, we propose to calculate the bistatic statistical illumination function for any one-dimensional random rough surface and to analyse its impact on the forward propagation above rough sea surfaces by considering Gaussian statistics and for grazing angles phi of the order of one degree.

The problem of shadowing from rough surfaces was considered in the textbook of Bass and Fuks (Pergamon, 1979) by means of the theory of random function overshoots. The statistical illumination function was then expressed from an infinite Rice series. The shadowing effect was also study by Wagner (JASA, pp. 138-147, 1966) and Smith (IEEE TAP, pp. 668-671, 1967), who retained only the first term of the series. Moreover, Smith used Wagner's approach by introducing a normalization function. They assumed an uncorrelated Gaussian one-dimensional surface. Recently, Bourlier et al. extended the previous formulations to any uncorrelated process and a correlated Gaussian process by considering one- (WRM, pp. 145-174, 2002) and two- (WRM, pp. 27-58, 2003) dimensional surfaces. More recently, the statistical illumination function in reflection was extended in transmission (N. Pinel et al., OL, pp. 2007-2009, 2005) corresponding to the case where the receiver is located below the surface.

For instance, for Gaussian statistics, when the shadowing effect is taken into account, the surface height PDF (Probability Density Function) of the illuminated points is not Gaussian any more. In this paper, from works of Bourlier et al., this probability is derived for Gaussian statistics with zero mean value. The special case of the forward propagation is analysed with respect to the height δ_h and slope δ_s standard deviations and the grazing angle psi. In addition, one will show that the illuminated height PDF can be modelled by a new Gaussian height PDF with a mean value and a standard deviation as functions of δ_h, δ_s and psi. Comparisons with a benchmark method obtained from Monte-Carlo simulations shows very good agreement with analytical Smith's formulation. In addition, the impact of the shadowing effect is analysed within the forward propagation factor calculated above a rough sea surface by considering the Ament (IEE Proc., pp. 114-115, 1984) reflection coefficient, in which the surface roughness is taken into account without shadowing effect. The atmosphere will be assumed to be the vaccum.

1. Introduction
Quasi-millimetric or millimetric wave band FWA systems encourage the adoption of broadband Internet access service. However, Line-Of-Sight (LOS) paths between the base station and subscriber stations are often covered by vegetation.
We have reported the attenuation through vegetation [1]. In this paper, to offset the attenuation through vegetation, we focus on the effectiveness of space diversity (SD) and present experimental results of SD effect. The SD effect of wind is also examined by using a fan that controls wind condition in a radio shielded room.

2. Measurements
Table 1 shows the major specifications of the tested trees. Figure 1 shows an overview of the experiment. A burst wave was transmitted from the transmitter, and received by the single receiver with higher level of two receivers. The switch of two receivers used the switch circuit that has 8 dB gain and 4 nsec switched speed.

3. Experimental results
Figure 2 shows time variation of received signal level (RSL) through Castanopsis cuspidate at 5 m/s wind speed. RSL of antenna 1 and 2 change in the order of milliseconds and the change is very large. However, rapid attenuation is reduced by effect of the switch. Figure 3 shows the cumulative probability distribution of attenuation. For a single antenna, the cumulative probability distribution of RSL approximates a Rayleigh distribution. The RSL at the cumulative probability of 1% is improved by about 10 dB.

4. Conclusion
We measured the space diversity effect against attenuation through vegetation. The result shows that space diversity offers improvement of about 10 dB in cumulative probability of 1%.

This contribution is concerned with the diffraction of an electromagnetic pulse, described by a Dirac delta function, by a perfectly conducting half-plane, moving with a constant velocity v < c. It is assumed that the source field is a plane pulse propagating perpendicularly to the half-plane edge. Then the problem can be split into two subproblems, with the source fields fully determined by their electric and magnetic field components in the direction of the screen edge.

For convenience two frames of reference are introduced. Laboratory frame is associated with the source field, and in the moving (or primed) frame the half-plane is at rest. We Lorentz transform the source field from the laboratory to the primed frame, and thus reduce the problem to that of pulse diffraction by a stationary obstacle. In the next step the (primed) source field is Fourier transformed, which further reduces the problem to that of harmonic plane wave diffraction by a stationary half-plane. The well known solutions to this problem, corresponding to the Dirichlet and Neuman boundary conditions, are due to Sommerfeld. We take advantage of those solutions, as represented in the form of contour integrals in the complex frequency domain, and transform them back to the time domain with the use of inverse Fourier transform and the Felsen approach. Finally, we Lorentz transform the fields that follow to the laboratory frame.

The resulting fields consist of three species: the incident and reflected pulses, and the pulse resulting from the moving edge diffraction. The incident and reflected fields are plane pulses, described by Dirac delta functions. They vanish in their shadow regions. The diffracted field has a more complex form and is non-zero behind its cylindrical front. It is interesting to note that the incident and reflected shadow boundaries do not coincide with the direction of propagation of the incident and reflected fields, respectively. This discrepancy vanishes as the velocity at which the half-plane is moving tends to zero.

Applications of this research can be found in object recognition, space science and astronomy.

Since the Federal Communication Commission decision in February 2002 of authorizing UWB emissions for radio applications, Ultra Wide Band (UWB) systems have received a great attention. In fact, UWB radio systems are about to provide, in a near future, high data rate and location awareness in difficult propagation channels. In this context, UWB system designers need an accurate knowledge of the radio propagation channel for systems performance through link level simulations. In particular, the channel knowledge can be accessed by performing a deterministic modelling. For this kind of channel modelling, the authors have already studied in a fastidious way the reflection and transmission phenomena occurring in the propagation of a UWB signal.

In this paper, the authors study in a detailed way the diffraction of a UWB pulse using the Uniform Theory of Diffraction (UTD). For this study, they first use the diffraction configuration illustrated on the left of Figure 1 with the heuristic UTD diffraction coefficient defined by Luebbers. Since UWB signals could have frequencies between 3.1 GHz and 10.6 GHz, each frequency component of the UWB signal is experiencing a different propagation ands, therefore, distortion accounts for a change in the UWB pulse shape which has to be considered when designing the receiver correlator. That’s why the main parametric study is the frequency study, as illustrated by the right of Figure 1 where a cartography of the UTD diffraction coefficient modulus, in function of frequency and diffraction angle, is given. Other parametric studies will be described in the paper, as the influence of dielectric characteristics (permittivity and conductivity) of the diffraction materials, the influence of incidence and diffraction angles or the influence of the interior angle of the diffracting wedge. Examples of shapes of received time domain UWB signals, after being diffracted, will be shown.

This kind of study allows us to understand the way the diffraction interaction modifies the UWB pulse, making a parametric study of this particular indoor interaction. We should also use more advanced diffraction coefficients, taking into account the transmission that can occur inside the diffracting wedge or the multiple reflections and transmissions in the diffracting material (if it can be considered narrow enough). Obtained results with these last two cases will also probably be shown in the paper. This study will help us to carefully characterize the UWB signal propagation channel and will help the receiver correlator designer and improve the UWB communication quality.

Aim of this study is a hybrid optimization model useful to predict the electomagnetic field in a cellular network. The hybrid prediction models [1] use in two different steps the deterministic and statistic algorithms, allowing to obtain accurate estimates in considerable shorter time than that needed to the traditional ray-tracing. These methods mainly consist in an optimization of statistic prediction on limited extension areas, based on deterministic techniques.

Usually the hybrid algorithm works in two different steps [2]: in the former the electric field is calculated in a subset of the original observational points; in the latter the power received in the remaining points is calculated by a statistic analysis of the field values previously simulated.

In this work we propose different methods (linear regression, Minimum Mean Square Error (MMSE), higher order regression curve) to be applied in the statistic analysis in order to obtain a better accuracy in the path loss calculation. The results of the different statistic approaches will be experienced and the electric field predictions will be compared. As an example, in Fig. 1 the results of the application of linear regression technique have been compared with those obtained by approaching the Minimum Mean Square Error technique. We can observe that the two techniques converge to the same performance when the subset number approaches to 100. More interesting results can be investigated by using higher order of regression curves.


Figure Captions:

Fig. 1: Linear Regression and Minimum Mean Square Error techniques comparison.

References:

[1] J. H. Tarng, Wen Shun Liu, Yeh Fong Huang, Jiunn Ming Huang, "A Novel and Efficient Hybrid Model of Radio Multipath-Fading Channels in Indoor Environments", IEEE Transaction on Antennas and Propagation, Vol.51, No. 3, March 2003.

[2] C. P. Michaelides, A. R. Nix, "Accurate High-Speed Urban Field Strenght Prediction using a new Hybrid Statistical Deterministic Modelling Technique", IEEE 2001, pp. 1088-1092.

This paper reports on a study on the field strength prediction for mobile communication networks inside buildings. The considered technique is based on advantages of using "dominant paths" (an alternative to ray tracing), and neural networks, trained with measurements. Some graphic and numeric results are presented.

The paper is mainly focused on studying the influences that the variation of some parameters have on processing time of the training phase for the neural network, and the accuracy of the results.

A geometric database was designed, including the exact position of the elements from each floor of the considered building (the School of Engineers of Bilbao), that is: walls (and their construction material), rooms, edges, etc.

For each radioelectric link defined, the convex-corners method is applied to determine the corresponding dominant path, that is, the path with least free-space, transmission and direction-changing losses.

It was also developed the code related to defining neural networks, selecting training patterns, and training and testing networks. The selected architecture that has been used to make predictions is a feed-forward multi-layered perceptron, trained with backpropagation.

The following error coefficients have been taking into account to determine the validity of predictions: Absolute Mean Error, Standard Deviation and RMS Error.

So as to optimize the convergence speed of the average error per pattern minimization process, and simultaneously obtaining a greater capacity of generalization for the predictions of the network, two solutions have been considered, the noise factor and the momentum term, and their advantageous effects have been explained.

The so-far obtained results show acceptable accuracy, with a usual error margin between measurements and predictions of up to 4dB. The graphic results presented in the article are the result of testing the neural network with the measurements used to train it, with measurements different from those used to train the network and with different geometric environments.

In the following graphics, a comparison between predictions and measurements for some routes followed in the School of Engineering of Bilbao is presented.

I. Theoretical Aspects

Various numerical methods are computed to solve Maxwell`s equations. Mainly, two kinds of techniques may be distinguished: formalisms in time and frequency domain.

First, a brief introduction on Finite Difference and Finite Volume in Time Domain [1] (FDTD and FVTD) will be achieved. Thus, it will be reminded that if FDTD and FVTD formalisms appear quite similar (due to identical sources and materials modelling in time domain for both formulations), one of the major difference rely on the ability of FV to handle non-structured grids.

The commercial software FEKO will be used in this article for simulation in frequency domain. Based upon the writing of Maxwell`s equation in integral form, the propagation of electromagnetic fields is solved by the method of moments. By enabling the use of unstructured 2D grids, this formulation gives a large geometrical flexibility.

II. Comparisons Of Geometrical Effects According To Resolution Formalisms

Each kind of model presented previously presents advantages and disadvantages. That’s the reason why it is no use to consider superiority of one to the other.

For Cartesian meshing as it is the case in FDTD, mesh definition is straightforward but realistic devices design often appears relatively complex. Thus, stirrer of LASMEA RC needs a particular treatment to improve approximation of metallic plates by using discrete geometry [2]. Although discrete geometry allows to use the better approximation on a structured grid, devices characterized by high curvatures and multi-scale structures need a conformal geometry description as it is the case for FV and FEKO© software. In the following part, we will present results obtained for a simple case.

In order to validate our protocol of comparison, we will detail a simple case extracted from [1] (see Fig.1). Comparisons are led on total field components in frequency domain (Fig.2).

Moreover, some more complex examples will be studied in order to confirm our previous results.

III. MSRC Simulations Vs Experiments

We first have to introduce real MSRC losses for numerical simulations using the measured quality factor of the considered MSRC. The final paper will present the way we use to introduce losses. Moreover, the influence of the source excitation type and position will be present. The considered excitation would be for example a point source, plane waves or electric dipoles.

The simulated results will be compared with experimental measurements.

References

[1] X. Ferrieres, et al., Application of a Hybrid Finite Difference/Finite Volume Method to Solve an Automotive EMC Problem, IEEE Transactions on Electromagnetic. Compatibility, Vol. 46, No. 4, November 2004.

[2] R. Vernet, et al., G้om้trie discr่te pour la mod้lisation d’antennes en FDTD, JINA, 2004.

Fig.1 Shape and FV mesh of the cone

Fig.2 Cone's comparisons with FEKO/FDTD/FVTD

The satellite propagation systems are affected by the presence of precipitation in the path of propagation to a great extent Thus, for the optimum performance of a communication link, the system designer needs to have the knowledge about the slant path attenuation prediction for which effective rain height is an important parameter.

Here we present the results of the effective rain height in statistical form for year 2004-2005.

The results are derived from zeinth path attenuation data along with corresponding rain rate data, derived from zeinth looking radiometer and tipping bucket rain guage , respectively.

It is observed that the effective rain height is found to decrease with increase in rain rate.Again the average effective rain height is estimated to be as 6.41 Kms while ITU-R predicts it to be as 4.78 Kms. Therefore ITU-R slightly overestimates the rain height. Thus this difference should be taken into account while calculation of slant path attenuation.

The aim of this paper is to present PA-CONECTA software. This program, which has been developed under the Matlab programming language, allows calculating electromagnetic field levels over an arbitrary terrain for multiple radiocommunication systems. The kernel of the software implements the PO-ML-BSP algorithm, which is based on: the Physical Optics (PO) approximation (Arias, Rubi๑os and Pino, R.R.D. in Magnetics, vol.1, 2000 pp.43-63. ISBN: 81-86846-89-1) in order to determine the induced currents over the terrain, the Binary Space Partitioning (BSP) technique to distinguish between visible and hidden parts, and a Multilevel (ML) scheme to reduce the computational cost.

The software has been programmed with a modular structure. The user can define the terrain to be analyzed with the geometric module, where two different types of scenario have been considered. The first one is deterministic. Its geometrical preprocessing for the work scene and an open-interface definition offer support to load different geometries, such as digitalized terrain maps, with the purpose of creating two files containing the points and triangles which model the terrain. The second one is a random scenario generated by the application. It takes into account some input parameters for the terrain (maximum height, size of the scenario...)

The next step is to build the PO-ML-BSP structure. Then, it is necessary to choose the number of levels and the number of divisions for every axis (x, y, z).

Once the terrain and PO-ML-BSP have been selected, a new module is used in order to set the base stations considered in the scenario. In this module, the user specifies the number of antennas for each station and their location (PA-CONECTA provides a latitude-longitude to UTM converter). Each antenna has its own radioelectric parameters (radiation pattern, input power...) Once the stations are defined, the program presents a figure where the coordinates associated with the antennas are presented.

The following step is the analysis of the coverage level. This module allows the user to consider reflections produced across the terrain due to the calculus efficiency growth associated with PO-ML-BSP technique (multiple reflections have a high computational cost in a PO-BSP traditional scheme). PO-ML-BSP splits the scenario into different voxels, which are composed of groups of triangles contained in a geometric domain (usually cubes). A voxel of superior level is generated by joining adjacent voxels. This multilevel scheme computes the reflections voxel by voxel instead of facet by facet with significant cost reduction.

The representation module offers the possibility of depicting the terrain (Fig. 1a) and the computed fields over the selected observation points as presented in Fig. 1b.

FERMAT software has been developed by the Electromagnetism and Radar Department of ONERA within the framework of a project dedicated to electromagnetic (EM) modelling. This code capitalises the results of research activities in high frequency scattering methods. A unique tool is currently available for studies dealing with radar analysis, antennas radiation, EM inter-system compatibility and propagation applications. This numerical simulation tool has the ambition to calculate scattered EM fields at high frequencies (i.e. the size of objects is supposed large compared to wavelength), in a virtual 3 D, geometrical and physical complex environment including natural and man made objects. EM fields calculations either on a surface or in a volume could be carried out according to specific applications.

FERMAT associates various techniques and tools:

•Virtual 3D, geometrical databases, composed of a great number of elements, landscapes and objects such as vehicles, buildings,..., modelled by thousands of plane geometrical faces and associated management tools. Specific features and textures relative to electromagnetism scattering are provided for each face.

•An optimised technique of geometrical Shooting and Bouncing Rays (SBR), to calculate the intersections between the rays from the transmitter towards the database and back to a receiving point.

•EM models of propagation, reflection, diffraction and an operating strategy (thanks to SBR) which allows unified calculation for the near or far EM scattered fields from the scenes. These models are the formulations of Geometrical Optics (GO), Physical Optics (PO), Uniform Theory of Diffraction (UTD) and Physical Theory of Diffraction (PTD). Again these models judiciously coupled with the SBR technique makes the computation time slightly dependent on the complexity of the treated virtual scene.

•Generation and management scenario tools to simultaneously simulate, for example, motions of a radar sensor and displacements of mobiles in the landscape according to the type of field calculations.

•Outputs and visualisation tools adapted to investigated applications.

The strength of FERMAT is that it is developed, under an agreement of partnership between ONERA and OKTAL SE. This company brings its competencies in the management and exploitation of databases, and the control of the SBR technique.

The validation mainly consisted in comparison with the EM models calculated either with other methods of reference or with measurements on simple objects known as canonical and allowing to evaluate the various EM interactions. These comparisons take place during specialised national and international "workshops".

Results on complex CAD up to 800 000 faces are obtained and will be shown. The near field calculations capabilities is pointed out through an example of strong EM fields estimation in the vicinity of wireless transmitting antennas in urban area.

The ray-tracing method is very attractive because several radio propagation characteristics can be simply predicted based on the knowledge obtained from geometrically tracing the rays from the transmitter (Tx) to the receiver (Rx). However, when taking many structures into consideration, many rays must be traced in order to achieve a high level of prediction accuracy and this is very time consuming. Accelerating the ray-tracing process while maintaining a high level of prediction accuracy is one of the most important problems. In order to address this problem, a ray-tracing acceleration technique is proposed that employs the genetic algorithm. Hereafter, this method is called the GA_RT method.

The GA_RT method is based on the conventional imaging method reported in [1]. In the conventional imaging method, the ray paths from the Tx to Rx are searched for various combinations of all components such as planes and wedges of structures. The number of combinations is M^N when searching the N-reflected and/or diffracted rays, where M is the number of considered components. Note that there are many combinations such that the ray path cannot be geometrically determined or where the traced ray hardly affects the total received power. This means that the conventional imaging method is inefficient. When employing the proposed GA_RT method, optimal combinations are extracted from all candidates by using the genetic algorithm and then the ray paths are searched for only the extracted combinations. The genetic algorithm has a high affinity to the imaging method when each combination is handled as eindividualf. Figure 1 shows the population model used in our study. Here, efitnessf of each individual is received power of traced ray along the corresponding path.

We evaluated the GA_RT method using a simple calculation model. Figure 2 shows one of the evaluation results. The figure shows the relationship between the calculation rate and the calculation error. Note that the calculation rate is the ratio of the number of calculated combinations to that of all candidates, and we set the threshold value of this rate as the finish condition. Figure 2 shows the results of the conventional method, but in this case, the search combinations for the ray paths are randomly extracted from all candidates (hereafter random search). We understand that the calculation accuracy of the proposed GA_RT method is higher than that of the random search, even though the number of considered combinations is equal. In addition, if 4 dB is within the allowable error, the amount of calculation can be reduced to approximately 20%. In the full paper, another model is proposed that can obtain higher performance from the GA_RT method. This model minimizes the calculation time to approximately 6% when there are many calculation points in a wide area.

[1] T. Sarkar, et al., IEEE Antennas and Propagation Magazine, Vol. 45, No. 3, pp. 51-82, June 2003.

After the great success of wireless communications used in land mobile radio systems, wireless communications in time variant scenarios becomes interesting. The steadily increasing demand for multimedia applications in all situations requires to look for new wireless concepts for the planning of wireless systems. Time variant scenarios can be found in each environment. Major applications are car-to-car or car-to-fixed-point communication scenarios, automotive radar systems, GSM in aircraft systems or WLAN hotspots in railroad stations or airports. The main aspect in such applications is the time variance of these scenarios, as the positions of transmitter, receiver and/or obstacles in the scenario change continuously and so the received channel impulse response (CIR) is influenced significantly.

Ray optical propagation models are well known in the domain of indoor and urban radio network planning. Classical scenarios are time-invariant, so moving objects in databases are not possible and there is no consideration of time variant effects like Doppler Shift. Therefore a new concept was developed to generate, edit and accomplish predictions in time-variant scenarios. The objects in the scenario are modeled with polygons. Radar cross section (RCS) can also be used for the modeling of complex objects. These objects are substituted with their RCS to accelerate the prediction. Each element in the database can be either stationary (not moving) or non-stationary (dynamic). Translation and rotation vectors, as well as scalars for velocity or acceleration are assigned to objects for the definition of their time variant behavior. Predictions are then accomplished for arbitrary defined time stamps.

A new 3D Ray Tracing model was developed for these time variant scenarios. An arbitrary number of reflections, diffractions and transmissions/penetrations is possible for each ray path. The attenuations due to reflections are computed with the Fresnel coefficients. The loss due to diffractions is determined with the Universal Geometrical Theory of Diffraction (GTD/UTD). CIR and Direction of Arrival (DoA) are computed for arbitrary points. A combination of CIR and DoA leads to the directional CIR (see fig. 1).

As in time variant scenarios, objects can move, the Doppler shift has to be considered, when a ray interacts with a non stationary object.


An example of a time variant car-to-car communication scenario with the most relevant propagation paths for one snapshot is shown in figure 2. Transmitter and receiver are mounted on the roofs of cars. The scenario consists of several objects (buildings, vegetation, guard rails and other vehicles) having a significant influence on the received channel impulse response (CIR).

In the final paper the requirements for modeling the radio channel for time variant scenarios will be described. A Ray Tracing approach for the propagation modeling in such scenarios will be presented and a concept for the modeling of large scenarios will be shown. A comparison of prediction results to measurement campaigns confirms the high accuracy of the model.

Broadband transmission at high data rates with high mobility speeds over wireless propagation channels is in general subject to both time dispersion and frequency dispersion. Time dispersion (delay spread) caused by the multipath channel arises from scattering, reflection and refraction of the transmitted signal. In addition, frequency dispersion, or Doppler Spread, arises from the mobility of the transmitter and/or the receiver. To evaluate the performance of broadband mobile communication systems over such a time varying multipath channel, it is useful to simulate their performance by computer models.

OFDM, with its variants, is a candidate modulation technique proposed for the air interface of next generation mobile communication systems. The main objective of this paper is to assess the performance of OFDM based cellular mobile communication systems under realistic conditions using a time varying channel model with various indoor and outdoor channel types. Accordingly, the simulation makes use of the CODIT channel model that was refined and validated using measurements made by within the RACE project in the UK.

This model provides the required detailed insight into the wireless mobile radio channel across a range of environment types, enabling the combined effects of Delay Spread and Doppler Spread on an OFDM system to be evaluated. For this purpose, a Matlab/Simulink code has been developed to simulate the performance of OFDM cellular system transmitted over the CODIT channel. The paper presents a detailed evaluation through the BER, Capacity and PSD as well as illustrating the OFDM transmitted and received signals.
The figure below show an example of the output of the simulation comparing the SNR to the BER of QPSK transmitted OFDM signal in the CODIT channel model.

An evaluation is presented of the performance of OFDM, dimensioned appropriately for anticipated high bandwidth next-generation mobile systems, in particular showing the extent to which the effect of Doppler spread is mitigated by channel equalisation.

Key Words: CODIT channel model, OFDM, Cellular communication systems, Doppler Spread, Delay Spread

Although wireless standards are prepared to solve connection falls, mainly by retransmission of data, the increasing number of systems using the same spectrum allocation could force the active LANs to continuously retransmit data, overloading the spectrum bands as well as collapsing their own transmission capacity. Another problem associated to the wireless technology is network safety. These upcoming problems can be mitigated by using different techniques, being site shielding one of them. If radio systems could be safeguarded against radiation from transmitter out of the specific network, the frequency reuse is improved and, as a consequence, the number of WLANs sharing the same area may increase maintaining the required quality standards. The proposal of this paper: the use of trees or bushes as a barrier to attenuate signals from other networks and, so that, to defend the own wireless system from outer interferences.

In order to justify this application of vegetal elements, a measurement campaign has been performed at two different wireless frequency bands: 2.4 GHz and 5.8 GHz, focused on determining the attenuation induced by several specimens of four different vegetal species: karo, Japanese cedar, camellia, and apple tree. Then, the effect of tree rows has been tested.

Narrow band measurements were used to characterise the effect of individual trees on the radio channel, as well as the effect of vegetal barrier. The set up consisted in separate transmitter and receiver. Median has been selected to characterise the central tendencies, as it presents better response than mean against outliers.

Analysis of all data shows that when a tree is obstructing the radio channel line of sight, the received power is reduced and the temporal variability of the signal considerably grows compared to free space reference.

Results show a reduction up to 13 or 30 dB for individual trees, accompanied by increments in temporal variability of received power levels (interquartile ranges grow between 4 and 12 dB), as well as in the number of outlier values. Vegetal species susceptible to conform of the barriers have to fit more conditions that just electromagnetic absorption: they must have everlasting leaves, as well as they have to be tall enough to cut the line of sight between base stations and target areas. Another common characteristic of these species is that present densely foliated specimens.

Once tested that individual trees force attenuation in the radio channel at WLANs bands, the next step is to look for a valuable application of this property. And the shielding among LANs is economically valuable. This application was verified by means of a row of apple trees.

The possibility of using lines of trees to reduce the interference in shadow areas was tested, and a shielding between 8 dB at 2.4 GHz and 15 dB at 5.8 GHz band was detected.