Successful design of future wireless communications systems requires a detailed knowledge of the radio propagation channel. 3D ray tracing has meanwhile advanced to a performance that it can provide channel information with high accuracy. This paper demonstrates the capability of 3D ray tracing for defining parameter sets of future wireless communications systems. Moreover it shows that the proposed model can be used in order extract channel parameters for the specification of MIMO channel models. In the first part of the paper the 3D ray tracing model is described. Then comparisons with wide-band non-directional and directional measurements at 2 GHz and 5.2 GHz respectively are presented showing a good agreement.

The 3D ray tracing model consists of a realistic model of the propagation environment and a model to calculate the multi-path wave propagation between the transmitter and the receiver. As environment a digital model of a part of the city of Karlsruhe, Germany, is used. It includes buildings, trees and the street floor. The wave propagation model is based on ray optics. It distinguishes between different multipath components. Each path is represented by a ray, which may consecutively experience several different propagation phenomena, like combinations of multiple reflection, multiple diffraction and scattering. Fig. 1 shows the result of the ray tracing procedure for a single snapshot calculated in the digital Karlsruhe scenario. The receiver is placed on top of an exposed building and the transmitter is positioned at a road crossing 1.7m above ground. It is clearly seen that the wave propagation scenario is very complex and gives rise to numerous mult-path components.

In order to verify the model, it is compared to wide-band non-directional and directional propagation measurements. The MEDAV RUSK ATM vector channel sounder is used as measurement platform. The urban measurements are performed in the city of Karlsruhe, Germany. Two different receiver positions are chosen on top of an approx. 38.5 m high building. The transmitter is placed in a vehicle (van) and is moved along 5 different measurement routes. Characteristic narrow-band and wide-band channel parameters are considered for the non-directional comparison between the simulated and measured channel. The narrow-band comparison includes not only the long-term behavior of the channel but also the statistical properties of the short-term fading component. Fig. 2 shows the measured and simulated receive power for one of the measurement routes. It will be shown in the final paper, that the deviation of the simulated receive power of all other measurement routes is in the same order of magnitude. The time-variance of the channel is compared by the measured and simulated Doppler spectrum and its characteristic properties (mean Doppler shift, mean Doppler spread). The comparison of the wide-band channel behavior is based on the power delay profile, and the resulting delay spread. Finally the spatial behavior of the measured and simulated channel is compared by the angular power spectrum and its characteristic properties (angular spread, mean angle).

The recent emergence of short range applications, e.g. RFID, requires careful investigation of wave propagation and adequate description of the radio channel. In this contribution, we present indoor channel measurements that expose the transition between the near and far field as well as the effects of fading. Furthermore, we develop a statistical single bounce channel model that is parameterized by the characteristics of the scenario and we verify the results by measurement data.

Introduction
In short range applications like RFID the transmission distance ranges from some centimeters to a few meters. The typical environment for these applications is an indoor site where a lot of interacting objects imply rich scattering. Far off the transmit antenna the resulting multipath propagation causes fading that is well investigated and can be modeled statistically.
Unlike in mobile communications where the channel behaviour can be described by models that only consider wave propagation in the far field of the transmit antenna, the near field has to be taken into account for short range applications too. Because of the strong near field component in close vicinity of the transmit antenna that declines very fast with increasing transmission distance, conventional channel models are not able to describe the field within the whole range of operation sufficiently.
In our contribution, we show measurement results of a typical RFID scenario and present a statistical channel model that is derived from simple geometric considerations in conjunction with basic field theory. We show that our model can be used to describe the statistical channel behavior within a user parameterizable scenario. The model is verified by comparison with measured data.

Channel Measurement
To determine the radio channel behavior in an indoor scenario, a transmission experiment was set up. At 868 MHz, the signal transmission between a fixed antenna and an antenna mounted on an xy-positioning table was investigated. The channel coefficients were measured with a vector network analyzer while the receive antenna was moved around within an 8 by 2 wavelength area. Fig. 1 shows the magnitude of the measured channel coefficients in dB plotted over the receive antenna position.
In the full paper we will show that the field strength close to the monopole ground plane antenna behaves very accurately like the field of a Hertzian dipole whereas, in the far field the effects of multipath propagation (fading) become dominant. The reason for this is that the near field component around the antenna is much stronger than reflecting signals in the environment.

Channel Modeling
Our model is based on statistical single bounce assumptions which are enhanced by a near field component that is determined by the equations derived by Heinrich Hertz for the Hertzian dipole. The proposed model 1) has enough flexibility to permit reasonably accurate fitting of measured channel data, 2) is simple enough to be used in simulation and analysis, and 3) can be adjusted by a set of parameters to represent different scenarios. The mathematical description will be presented in the full paper.

This paper provides a theoretical yet accurate and flexible proposal of propagation modeling for wireless sensor networks. The proposed modeling approach can be used for free-space and multipath environments and provides means to handle radio irregularities in radio range caused by path losses and heterogeneous communication ranges due to various reasons like hardware calibration and energy depletion over time. Where most of the propagation models are either overly simple and generic or accurate and hardware specific, this model tries to bridge the gap between these two approaches. A basic simulator of a hybrid channel model has been built, both for free space and multipath environments. This model has been imported into the WSN (Wireless Sensor Network) simulator with various simulation setups and topologies to study the effects of channel modeling approach on different layers of wireless sensor networks.
Index Terms-multipath, packet reception rate, radio wave propagation, signals to noise ratio, wireless sensor networks.
II. Methodology and Results
The proposed model is primarily used for analyzing and calculating the packet reception rate and signal strength with varying distance, noise levels, signal-to-noise ratio and various transceiver parameters such as noise bandwidth, data rate etc. Figure 1 shows a case of the methodology adapted to the calculation of the received signals at different nodes in a multipath environment. The received signal at Sink node N4 is given by:

(1)

The satellite to indoor propagation channel in L and S bands is becoming of significant importance as future satellite mobile communication, broadcast and navigation systems (as e.g. S-DMB and Galileo) are aiming at optimum performance in all kind of environments, i.e. even in indoor scenarios. Furthermore, the indoor scenarios are of special relevance for the attractiveness of new MSS services in order to provide ubiquitous coverage over all environments.

The satellite to indoor propagation channel depends strongly on the layout and the material properties, i.e. the construction materials used for walls, windows, and ceilings of the building where the receiver is located. Therefore it is important to first categorise the buildings into relevant types and to define typical material properties for each of the categories defined. A possible classification could be the distinction of residential houses with 1-2 storeys, multi-storey residential buildings (e.g. apartment blocks), office buildings, and commercial buildings as factories, stations and airports. Other important parameters influencing the satellite to indoor propagation are given by the satellite elevation angle and the orientation of the building towards the satellite. Besides the varying attenuation of the direct satellite signal depending on the penetration loss, multipath contributions are introduced by the building walls (e.g. reflection on the floor) leading to Rice or Rayleigh fading distributions depending on the location of the receiver inside the building.

Due to the high amount of parameters influencing the satellite to indoor propagation channel, wave propagation models are mandatory for analysing the various effects in detail, as this can not be achieved by measurement campaigns only. For the accurate simulation of the satellite to indoor propagation channel a ray-optical model for the coverage prediction in terrestrial networks (included in the radio planning tool of AWE Communications) has been upgraded to consider satellite transmitters (in the framework of the ESA funded S-DMB Phase A study). This deterministic model is based on 3D vector data of buildings and considers the different propagation phenomena as direct path, reflection, diffraction and penetration by using the principles of geometrical optics. Accordingly, the ray-optical model allows a site-specific prediction of the satellite to indoor radio channel for each point inside the corresponding building. Numerous simulations of the satellite to indoor channel for various buildings have been performed.

In the framework of the EU FP6 project MAESTRO, the Fraunhofer Institute IIS (in Erlangen) has carried out several measurement campaigns of the satellite to indoor channel for different types of buildings. These measurements show different zones of signal penetration depending on the location inside the building with respect to the satellite (see Figure). The measurements have been used to compare and verify the ray-optical model implemented in the radio planning tool.

The aim of the paper is to show the influence of humidity contained in real building materials on diffraction part of the field in the vicinity of a wedge-shaped obstacle. Field measurement around both dry and wet wedge formed by real building material was performed in the frequency range of 4 - 40 GHz to get an idea what the influence of humidity contained in building material is upon the field around a building obstacle. The measured results are compared with theoretical prediction by Uniform Theory of Diffraction (UTD).

The experiments of field measurement around a wedge both in anechoic chamber [1] and around a real building [2] show a good agreement between UTD prediction and measurement. Rather than to compare measured results with UTD prediction we focused here on the electrical properties of the building material. There have been numerous papers published on electrical parameters of building materials but relatively small amount of work has been done concerning the influence of water content within building materials on propagation prediction.

As an illustration, Fig. 1 shows angular dependency of received field at frequency of 40 GHz for several types of wedges and for vertical polarization. The depicted dependencies represent perfectly conducting wedge, dry building material and wet building material. The paper presents the measurement results for the whole frequency bands and draws some conclusions regarding the influence of the material water content on the diffraction.

Fig. 1. Field amplitude around wedge-shaped obstacle

References
[1] P. VALTR and P. PECHAC, "Diffraction calculations and measurements in millimeter frequency band," Radioengineering, vol. 13, no. 3, pp. 18-21, Sept. 2004.
[2] H. R. ANDERSON, "Building corner diffraction measurement and predictions using UTD," IEEE Trans. Antennas Propagat., vol. 46, no. 2, pp.292-293, Feb. 1998.

Time reversal (TR) is a scheme that has its origin in wideband transmission in underwater acoustics and ultrasound: in TR, the signal to be transmitted is filtered through a filter that is the time reversed and phase conjugated channel impulse response (CIR). The resulting signal focuses tightly on the intended receiver in both space and time. Due to these properties, TR has recently appeared as a promising technique for secure, wideband, wireless communications. Specifically temporal focusing is important for systems where delay spread is a fundamental limitation.

TR filtering is commonly applied in Multiple Input- Single Output (MISO) situations, i.e. where there are M transmit antennas and one receive antenna. Let g_m(t)=A h_m^*(-t) denote the TR filter employed at the mth transmitter antenna, where h_m(t) is the CIR from that antenna element to the intended receiver and * is the complex conjugate operator. A is a scaling factor so that the TR filters introduce unity gain. The equivalent C h_eq(t) is

h_eq(t) = A h_m^*(-t) X h_m(t) ,

where X denotes the convolution operation.

In general, the delay spread (ds) is given as the second moment of the average power delay profile (pdp). We can similarly define the ds of the equivalent CIR as the second central moment of the equivalent pdp. [1] investigated the efficiency of TR at reducing the perceived ds of the channel, based on actual channel measurements. It showed that the efficiency of TR depends on the scattering situation around the acting transmitters and receivers. Specifically two situations were defined:

a. Cluttered to clear: the intended receiver is in a less scattering environment than the acting transmitter array,

b. Clear to cluttered: the intended receiver is in the clutter.

The measurements showed that the advantage of applying MISO TR increases as the separation between the two ends of the communication link increases, when the intended receiver is in the clear and the acting transmitters are in the higher clutter situation. The opposite occurs when the receiver is in the cluttered situation and the acting transmitters are in the clear. This trend can be explained in terms of the shower-curtain effect. Indeed [2] has mathematically analyzed this effect in the regime of turbulent propagation. [3] has investigated the influence of the shower curtain effect on the spatial focusing properties of TR.

In this paper we reproduce the experimentally observed temporal focusing results using simulations of the wireless channel. Specifically we use the ring of scatterers modeling approach. We investigate the effects of bandwidth, number of transmit antennas, and transmit antenna spacing on the efficiency of TR as a means to reduce the perceived ds of the channel.

References

1. P. Kyritsi, G. Papanicolaou, P. Eggers, and A. Oprea, "MISO Time Reversal and Delay-Spread Compression for FWA Channels at 5 GHz", in Antennas and Wireless Prop. Letters,Vol. 3, No 6, 2004, pp.96-99.

2. A. Fannjiang, "Information Transfer in Disordered Media by Broadband Time Reversal: Stability, Resolution and Capacity," e-print: arxiv.org/abs/physics/0509158

3. A. Ishimaru, S. Jaruwatanadilok, and Y. Kuga, "Time reversal in random media and super resolution with shower curtain effects and backscattering enhancement," in Proc. URSI General Assembly Meeting 2005.

In principle, electromagnetic wave propagation is governed and fully described by Maxwell equations. However, in several fields of applied science, such as mobile communications, remote sensing and radar engineering, we often have to deal with very complex propagation media, whose characteristics fluctuate randomly in time and space, and providing an analytical solution by applying the classic electromagnetic theory turns out to be an impracticable or, at least, very time-expensive way to solve the problem. In order to overcome such drawbacks, a possible approach consists in developing random models of the area of interest, characterized by a few parameters. Waves in such environments vary randomly in amplitude and phase and are accordingly described in terms of statistical averages and probabilities densities.

In such a framework, we consider the problem of optical ray propagation in a half-plane percolation lattice [1]. The idea of applying such model for characterizing a medium of disordered scatterers was proposed for the first time in [2], where urban environment is described in terms of a uniform random lattice. We consider a far field scenario, where a plane monochromatic wave impinges on the grid with a prescribed incidence angle. The wave is modeled in terms of a collection of parallel rays that undergo specular reflections on the occupied cells. The aim is analytically estimating the probability a ray penetrates up to a prescribed level inside the lattice before being reflected back in the above empty half-plane.

We will present and compare two different mathematical approaches, by taking into account the most general case of an inhomogeneous grid where scatterers’ density varies with the lattice depth. The first approach is an extension of the one proposed in [2] and is based on the theory of Martingale random processes. We will show limitations of such an approach when dealing with widely varying obstacles’ density distributions and we will propose as a solution an innovative approach based on the theory of the Markov chains [3]. Selected numerical tests and experiments carried out in a real controlled environment will assess the validity of the proposed solutions. Finally, we will present recent advances, where electromagnetic phenomena not considered yet are included in the propagation model, allowing application of the percolation approach to a wider ensemble of real propagation problems.

References
[1] G. Grimmet, Percolation. Springer-Verlag, New York, 1989.
[2] G. Franceschetti, S. Marano, and F. Palmieri, "Propagation without Wave Equation towards an Urban Area Model," IEEE Trans. Antennas Propag., AP-47, 1393-1404, 1999.
[3] A. Martini, M. Franceschetti, and A. Massa, "Ray Propagation in Non-Uniform Random Lattices," J. Opt. Soc. Am A, in press.

The prediction of the electromagnetic field in the urban environment is a fundamental step for developing radio network coverage and estimating the human exposure levels. The urban environment modeling may have different complexity levels depending on whether the model is highly detailed or not. We present in this paper a simplification procedure of a building wall model using homogenization. The process is divided in two steps and must be repeated for each observation point.

The first step is to determine for a given observation point which zone of the wall reflects most of the energy (e.g. 95 %). It is natural that this zone depends on the incident wave, on the building and on the observation point position. In this paper, our analyses are limited to the study of a flat wall built with different materials. It is illuminated by a normal incidence plane wave. Computations are run using a physical optics solver. The active zone analysis is based upon a decomposition of the wall into several concentric discs around the specular point, as shown in Fig.1.
We calculate the active power received through a square section at a varying distance from the plate, scattered by the surfaces within the concentric discs. From these results we determine which radius corresponds to the 95 % total power. This heuristic approach presented here is compared to the first Fresnel's ellipsoid.
The second step is homogenizing the whole wall with the equivalent material inside the active zone. The equivalent material is computed using a typical weighted algebraic average over the active zone, which takes into accounts the different material proportions.
This approach is dynamic since the homogenization zones depend on the observation points. Consider the scattering profile of Fig. 1 composed of glass (ε_r = 6) and concrete (ε_r = 9). The spatial profile has been generated using a pseudo-2D Markov algorithm [1] in order to obtain a spatial regularity, and not a complete chaotic wall representation. The equivalent material calculated along the Z axis inside the active zone (represented as the red concentric discs) is highly dependent on the distance h of the observation point to the surface, as shown in Fig. 2 below.
The E field scattered by the associated homogenized plate, and for each observation h is compared to the heterogeneous plate, as illustrated in Fig. 3
Due to the reasonably good accuracy between the homogenized field and the heterogeneous one within a 20 % uncertainty, it is of interest to neglect details on the wall in the propagation model, which will be a gain for computing time.
Fig.1. Fig.2. Fig.3.

[1] L. R. Rabiner, "A tutorial on hidden Markov models and selected applications in speech recognition," Proceedings of the IEEE, vol. 77, pp. 257-286, 1989.

Background: In the year 2000, the Swedish Telecom regulator, Post&Telestyrelsen (PTS) granted four licenses for 3G mobile operations in Sweden. All four licensees committed themselves to build networks that cover a population of 8.860.000. The definition of coverage specifies that the field strength outdoor at a height of 1.7m above ground from the primary common pilot channel, CPICH, measured over 5MHz should be 58 dBµV/m with an area probability of 95% [1].

To verify the coverage PTS has developed a test procedure where the field strength is measured in a drive test. However, designing such test constitutes a number of challenges:

First of all the requirement is for covered area while a drive test only measures along a route. In order to convert measurements data from drive testing to probability of area being covered, one need a statistical model based on population density and geography.

For practical reasons, PTS needs to verify the coverage through measurements using vertically polarized antennas. This is the standard way of verifying radio networks.

Typically one antenna for each operator would be mounted on each of the corners on the roof of a standard car. The antennas are then connected to receivers for each operator’s individual frequency band. This set up makes the drive test very easy to conduct. However, one needs to be aware that such a set up is filled with errors due to e.g. the size limitations of the car roof which is assumed to act as a perfect ground plane. In GSM this does not constitute much of a problem since the coverage requirements are not so stringent. However, in 3G the accuracy in the measurement needs to be extremely high since even a small systematic error of 1dB could have the consequence that each operator would have to build an extra ~1000 sites at a staggering cost of 1bilion SEK!

Results

The present paper gives an overview of the considerations behind the design of a highly accurate test method for verification of 3G licence requirements. In particular we analyse: 1. The relationship between the requirement set for the CPICH outdoors and data rates indoors;

2. The power needed to be allocated to the CPICH.

3. A statistical model based on population density and geography to be used in order to convert data from drive testing to 95% area probability is presented and discussed.

4. The antenna designed for the measurements and the influence of the ground plane on polarization, frequency response and gain.

5. The use of polarization diversity at the BTS antennas and its influence on downlink coverage [3].

References

[1] Tillståndsgivningen för UMTS i Sverige, Stockholm, Sweden: Swedish Post- and Telecom Agency (PTS), 2001.

[2] David Ribbenfjärd, Björn Lindmark, Bo Karlsson, Lars Eklund. "Omnidirectional Vehicle Antenna for Measurement of Radio Coverage at 2 GHz". IEEE Antennas and Wireless Propagation Letters, November 2004.

[3] Björn Lindmark and Claes Beckman "Recommendations for compensation of polarization mismatch in 3G networks" R-S3-SB-0325. Royal Institute of Technology, Stockholm Sweden, 2003