I. INTRODUCTION
The intention of presented work was to investigate the frequency dependent behaviour of the correlation coefficients of the signals received by the single elements of a linear antenna array. These correlation coefficients can be interpreted as a measure of the similarity of the signals at the different antennas and thus are a direct measure for the performance of diversity or MIMO techniques. The estimation of the correlation coefficients in the presented investigations is based on realistic indoor wave propagation simulations that have been conducted with a ray optical simulator. Different scenarios have been investigated. The conclusion of the full paper will give an overview on the results obtained for different system configurations.

II. WAVE PROPAGATION SIMULATIONS
A fully three-dimensional ray tracer was used for the realistic modeling of wave propagation in a typical indoor scenario. The indoor scenario is described by a polygon object model. This model represents a room equipped with furniture. All materials are assigned suitable dielectric and magnetic properties. The employed ray tracer is able to find all existing propagation paths between two arbitrarily placed antennas that occur due to reflection, diffraction, and rough surface scattering. For each propagation path attenuation and phase shift are calculated with full polarimetric information. At the receiver the arriving paths are added coherently and hence the complex channel transmission coefficient is obtained. By conducting this procedure for different transmission frequencies the transfer function of the channel is determined.
With this simulation tool the correlation of the received signals at the different elements of a linear antenna array with constant element spacing d has been investigated. The simulations have been conducted over the complete frequency range regulated by the FCC for UWB applications from 3.1 to 10.6 GHz. While the receiving array has always been kept in a fixed position, the transmitter was placed at 250 different randomly chosen positions in order to obtain stochastically independent observations for the correlation estimation.

III. SIMULATION RESULTS
The figure shows the obtained results for the power correlation coefficients |rho_ij|² of a linear array with d = 5 cm element spacing and elements with omni-directional directivity. The single elements are denoted with consecutive numbers. The estimated correlation coefficients are strongly dependent on frequency and element distance. The correlation between neighbouring elements (rho₁₂,rho₂₃,rho₃₄) shows very similar behaviour and comparably high values for the lower frequencies. The correlation between the elements with larger distance is rather low over nearly the entire frequency range. The final paper will include a detailed discussion of the simulation results and their impact on diversity and MIMO applications.

The full characterization of a wideband multi-antenna channel requires to state both the temporal and directional properties of the channel. This leads to the double-directional response of the channel that can be seen as the superposition of the contributions from the more significant multipath components. Starting from the SISO impulse response [1]:
(1)
where ô is the delay, rTx and rRx are the transmitter and receiver positions. Ù and Ø the directions of departure and arrival, and N(t) the number of significant multipaths for an specific time. The hl(t,ô,Ù,Ø)is the contribution of the l-th multipath,, and is modeled as: (2)
For the multiantenna case this impulse response is expressed in a matrix form. The proposed compact representation is based on obtaining the spatial scatter distribution by using an imaging technique to identify and locate the most significant sources of scattering. The imaging geometry, shown in Fig. 1(a), uses elements, Nt forming a transmitting array and Nr forming a receiving array. A measurement matrix can be obtained as follows: for every selected transmitting element the receiving array is scanned obtaining a Nr-measurement column. Then the procedure is repeated for the Nt elements of the transmitting array.

The reconstruction algorithm forms every image point by means of the synthesis of two focused arrays [2] (transmitting and receiving arrays), i.e., all the elements of both arrays are weighted by focusing operator so as to be focused on a unique object point. Both arrays consist of 32 elements separated distance dx and dy for the transmitting and receiving array respectively, and operating frequency band is 3 to 10 GHz UWB range. In Fig 1(b) the scatterer imaging reconstruction for regular geometry is shown.
The imaging of the scatterer distribution (position and reflection coefficient) allows to describe the channel characteristics in a compact way in terms of equation (2). The concept can be generalized to arbitrarliy distributed UWB network of sensors [3].

[1] A.F. Molisch, "Wireless Communications", Wiley, 2005
[2] Y.J. Kim, l. Jofre, F. De Flaviis, M. Q. Feng, "Microwave Reflection Tomographic Array for Damage Detection of Civil Structures", IEEE Trans. Antenn. Propagat., vol. 51, no. 111, Nov. 2003, pp. 3022-3032
[3] C. Chang, A. Sahai, "Object Tracking in a 2D UWB Sensor Network", Thirty-Eighth Asilomar Conference on Signals, Systems and Computers, 2004, pp1252-1256

Many studies have been initiated around the world on indoor communication systems based on ultra wide band signals. European projects (Ultrawaves, Pulsers) have been initiated in the past three years. Each project contains at least a sub-workpackage dealing with the "indoor UWB channel characterisation" showing first the importance of the topic and second the fact that this channel has not been fully understood yet. Many channel sounders have been deployed. A few statistical models have appeared. These preliminary results have indicated how complex the topic is.

We believe the proposed method is complementary for some aspects to the IEEE 802.15.3a channel model, which was useful for the selection process of the new standard for UWB high-data rate communications [A. Molisch, "Channel models for ultrawideband personal area networks", IEEE Wireless Com., Dec. 2003, Vol. 10, n°6].
The work we present here consists in a study of a deterministic solution to model the indoor UWB channel. The method was first explained and presented in [Y. Lostanlen, G. Gougeon et al. " A deterministic indoor UWB space-variant multipath radio channel modelling.", UWB-SP, vol. 7 ed. F. Sabath & E. Mokole, 2006]. Enhancements on the prediction method have been brought after comparison with measurements.
This work mainly intends to provide a prediction method improving the understanding of the physics underlying the propagation of a signal from the transmitter to the receiver by illustrating the multipath components. Starting from the antennas, that have to be modelled over this wide range of frequencies, the signals interact with many scatterers leading to various behaviours depending on the electromagnetic material composition, frequency, nature (transmission, reflection, diffraction ,...). The strength of the solution is to provide easily the resulting pulse after a simple phenomenon (e.g. a diffraction) as well as a complex bi-directional spatio-temporal channel for a chosen configuration. But we want also to show simple phenomena as they may help understanding the complex channel by discriminating the interactions. Indeed communication system designers and antenna specialists need to clearly understand how the signal behaves to provide the optimal system. This method may be used as an elementary module for any kind of UWB techniques (pulse, multiband OFDM, energy, ...).

The proposed paper will show the novelty and the maturity of the approach by clearly explaining the process. Measurements were carried out in indoor environments. Comparisons between the obtained Impulse Response simulations and measured signals will be described and analysed. It shows good results and gives some ideas to enhance further the modelling. The measurements were specified to analyse specific phenomena like diffraction or reflections in a corridor, multifloor propagation as this will be illustrated in the proposed paper. The simulated result appears to be less dense that the measured one (mainly because the richness of the real scene is not fully represented by the modelled scene e.g. pieces of furniture, interior of cupboards). The amplitude of the simulated cluster may be tuned to adapt the model to the environment. But the main contributions (time and shape of cluster arrival) are well retrieved. The inserted figures show the main rays determined by the proposed model. The measured and non-tuned simulated responses in a corridor are represented.

For the last few years, UWB techniques have been envisioned for low and high data rate indoor communications. The main advantage of UWB techniques is the great diversity provided by the available bandwidth. UWB solutions traditionally investigated are a trade-off between the exploitation of the available diversity and the receiver complexity. To increase data rate and take full advantage of channel diversity without making the receiver more complex, time reversal techniques can be applied to UWB communications. These techniques consists in transmitting the time-reversed version of the channel response. The reciprocity of the time invariant channel ensures both temporal and spatial focusing of the signal at the receiver side. Time reversal techniques have been successfully used in acoustic and underwater areas for many years [1]. More recently, papers have demonstrated the relevance of these techniques in electromagnetic area and their feasibility in a UWB context [2-3]. Theoretically, this system provides similar performances compared to traditional impulse radio coherent systems. In practical cases, time reversal performances are highly dependant on propagation channel properties.

In this paper, we study the characteristics of the propagation channel for transmitted reference combined with UWB communications. First, we set up a general MIMO channel modeling formalism based on the ray propagation assumption. Particularly, this formalism highlights the channel reciprocity principle on which time reversal techniques rely. Then, this formalism is implemented in a deterministic simulation tool that combines a ray tracing approach with optical geometry (OG) and uniform theory of diffraction (UTD). Prior measurements in near field chamber are used to rigorously take into account antennas in the simulator [4]. We use this simulator to verify the channel reciprocity property whatever the type of antennas. We also demonstrate that simulations can be exploited to analyze the spatial coherence of the channel. This characteristic determines, on the one hand, the distance that ensures focusing of the received signal, and on the other hand, the minimum gap between each antenna requisite in a MIMO situation.
The figures illustrate the time and spatial focusing on an example of a simulated channel response including real omnidirectional UWB antennas. Figure (b) shows the time focusing of the time reversed response compared to the channel impulse response of figure (a) where significant amount of energy is spread over a longer delay. The figure (c) shows an example where the focusing is not observed when the receiver is shifted by a distance of four wavelengths, which illustrates the spatial focusing property of the time reversal technique.

[1] Fink, M. , "Time-reversed acoustics," Scientific American, Nov. 1999.
[2] Qiu, R. C.; Zhou, C; Guo, N.; Zhang, J. Q.; "Time reversal with MISO for ultra-wideband communications: experimental results," Invited Paper, IEEE Radio and Wireless Symposium, Jan. 2006.
[3] Strohmer, T.; Emami, M.; Hansen, J.; Papanicolaou, G.; Paulraj, A. J.; "Application of time-reversal with MMSE equalizer to UWB communications," Proc. of the Globecom’04, Nov. 2004.
[4] Tchoffo-Talom, F.; Uguen, B.; Plouhinec, E.; Chassay, G.; "A site-specific tool for UWB channel modeling," UWBST & IWUWBS’04, May 2004.

Due to its potential applications like high data rate communication and positioning capabilities, UWB is gaining more interest in the field of wireless communication. However, the analysis and design of UWB communication systems need an accurate knowledge of the propagation channel characteristics. Therefore, extensive and accurate channel measurements are required. To date a limited number of measurement campaigns and channel modeling efforts have been reported to characterize the UWB channel when compared to narrowband measurements.

A large frequency band of 3.1 to 10.6 GHz is devoted to UWB wireless communication in indoor environments according to FCC. Propagation mechanisms (e.g. reflection, diffraction, scattering...) are very frequency dependent. Because of the large frequency band occupied by UWB signal, channel parameters in the whole band are frequency dependent. This important issue should be well incorporated in the modeling of the UWB channel.

A large set of UWB measurements has been performed. The measurements are carried out in indoor office environments at the campus of Delft University of Technology using a time domain technique. The generator fires a Gaussian-like pulse with duration of 50 ps. As a result, the measurements can be performed in the bandwidth of 12 GHz. The maximum measured distance was about 10 m for both LOS and NLOS situations. To characterize the small-scale and large-scale fading parameters of the sub-channels, a grid of 7x7 with a spacing of 5 cm between adjacent points was designed and used in the measurements.

In this paper the total measured band is divided into 15 sub-bands of 500 MHz. For each sub-band statistics of channel parameters are obtained. Furthermore, the frequency dependency of these statistics is investigated. The results of the statistical analysis address the frequency dependency of: amplitude fading statistics, large scale path loss, time dispersion and correlation. Major application of this study is the MB-OFDM which divides the band into several bands.

Vladimir SCHEJBAL, Zdenek NEMEC, Dusan CERMAK, Pavel BEZOUSEK
University of Pardubice, Studentska 95, 53210 Pardubice, Czech Republic
P21 - UWB channel modelling
vladimir.schejbal@upce.cz, zdenek.nemec@upce.cz, dusan.cermak@upce.cz, pavel.bezousek@upce.cz
Prime Author: Vladimir Schejbal

The UWB concept is very useful for radars and communications. A UWB engineer needs to be familiar with both the time domain and frequency domain, able to switch seamlessly from one domain to the other as the nature of problem demands. The UWB radar output signals can be substantially affected due to electromagnetic wave propagation through walls and multipath effects. The program, which allows the calculations for both frequency spectra and time responses for combinations of various receiving and transmitting antennas, input signals and obstacles (walls), has been coded using MATLAB® language. The numerical simulation of UWB radar propagation effects is briefly described.

Several combinations of receiving and transmitting antennas and input signals have been calculated and compared. It can be concluded that UWB radar output transmitted signals are formed both with transmitters and antennas. The transmitting transient responses of an ideal antenna are proportional to the time derivatives of the receiving transient responses of the same antenna. Therefore, UWB antennas should be considered as an integral part of the whole systems. Moreover, the output transmitted signals should be formed according to UWB system demands. That means that analyses should be done for very wide frequency spectrum and simultaneously, the effect of input signals should be considered both for transmitting and receiving antennas.

The propagation of electromagnetic waves through obstacles has been analyzed, where wall parameters are given both for brick and concrete walls with various thicknesses and S11 and S21 can be found using numerical simulations for these cases. The responses (input and output signals calculated using IFFT) have been extensively analyzed as well. The ringing (due to boundary multiple reflections) can be clearly observed. Naturally, the interferences of multiple reflections are much smaller for very short pulses than for CW and narrow-band applications.

The method can be used for analyses of multipath propagation due to reflections (such as ground or wall reflections). Multipath effects are analyzed and simulated numerically for various cases with several heights and distances. The output transmitted signal is usually formed according to UWB system requirements. Various cases show delays (due to propagation through wall and various paths of direct and reflected rays) and the ringing (similar to UWB propagation through wall). Certainly, these phenomena are much more pronounced for reflected rays. On the other hand, the interference effects of multiple reflections and multipath effects are much smaller for UWB signals than for CW narrow-band applications as interference minima and maxima do not occur for the same frequencies. Moreover for very short pulses, the individual pulses are received at various times and can be distinguished more easily.

Understanding of local environmental conditions near the terminals is important for determining system performance and possible interference in UWB channels[1,2]. Objects like desks and cabinets scatter the radio waves and thereby cause spatial fading as a result of multipath as well as shadowing due to blockage. When people move in the vicinity of the link they cause additional fading, which may be the result of body blockage or multipath interference due to body scattering. On line of sight (LOS) links body blockage is found to produce deep fades, while on obscured links the effect of body scattering are found to be most significant. This difference is observed both in the time dependence of the excess path loss as people move, and in the statistical distribution of the fading for indoor propagation [3,4,5]. In this study, measurement results for local enviromental effects in UWB channels are presented. Effecs of objects in and around the radio link particularly human body shadowings are studied for UWB channels. Fully computer controlled measurement setup includes a signal generator, two identical triangular monopoles and a spectrum analyzer that is connected to a PC (Figure 1). All measurements including outdoor calibrations are performed in Atilim University campus area. Measurement setup and preliminary results involving human body blockage when crossing the radio link at different frequencies are presented in Figure 2-4. 1.Roy, S. et al.,"Ultrawideband Radio Design: The promise of high-speed, short-range wireless connectivity", Proceedings of the IEEE, vol.92, No.2, 2004. 2.UWB Channel Modeling, IEEE P802.15 Working Group Document. 3.Obayashi, S., and Zander, J., "A body-shadowing model for indoor radio communication environments", IEEE Trans. on Antennas and Prop., vol. 46, no.6, pp.920-927, 1998. 4.Hafezi, P. et al., "An experimental investigation of the impact of human shadowing on temporal variation of broadband indoor radio channel characteristics and system performance", IEEE VTC-Fall Vehicular Technology Conference, vol.1, pp.37-42, 2000. 5.Kara, A. and Bertoni, H.L., "Blockage/shadowing and polarization measurements at 2.45 GHz for interference evaluation between Bluetooth and IEEE 802.11 WLAN", IEEE APS/URSI Symposium, vol. 3, pp. 376-379, 2001.

Abstract:
In order to combat fading in wireless networks and to increase the capacity of a system, relaying technique is an interesting solution. In this paper, we develop an ultra wide band (UWB) channel model for multihop transmission using the non-regenerative relay. The signal-to-noise ratio (SNR) and the consumption energy are the two important factors to compare a direct transmission with a relayed one in ad hoc or sensor networks.

Context:
Ad hoc and sensor networks are self configuring. Some, if not all, nodes can assume router functionalities when needed. This enables terminals to communicate with each others when they are out of range, providing they can reach each other via intermediate relaying terminals. This relaying also allows to reduce power consumption, which results in extended battery life time and lower level of interference.
Non-regenerative systems use relays that simply amplify and forward the received signal. A fundamental question is whether it is advantageous to route over many short hops or over a smaller number of longer hops. If energy consumption is reduced, two questions are addressed: what is an acceptable number of hops that can be made and what is the position of an acceptable relay.
This paper presents a model for a multihop UWB channel. The multipath behaviour and the thermal noise plus multi-user interference are discussed.

Multipath evolution with the number of hops:
In multipath transmission, the energy transmitted by different paths is affected by the number of the hops. A linear filter is used to represent the channel. It is deduced from measurement data within the IEMN laboratory [1] in the 57 to 59 GHz band. A deterministic part representing the multipath behaviour of the channel is extracted and some randomness is added by the phase of the paths and their probability of presence.
The energy carried by the main path is reduced when the number of hops is increased. Fig. 1 gives the percentage of the total transmitted energy carried by a number of paths in the line of sight 60 GHz channel and given a number of hops.

SNR evolution with the number of hops:
The noise mean power Pnoise at the input of the destination terminal after n hops can be written as follows:

where Gi is the gain of the ith relay, β is the free space attenuation in the nth hop and No is the noise variance. Equation (1) shows that the noise increases when the number of hops increases. Besides the interfering users arriving at each intermediate nodes will add and the global signal to interference plus noise ratio is reduced.

Conclusion:
In the final paper, we will propose a channel model, including Multipaths and power considerations that depends on the number of hops. We study cases when a multihop communication allows energy savings but keeping a satisfying signal-to-interference noise ratios.

References:
[1] A. Bendjaballah, H. El Ghannudi, N. Deparis, A. Boe, L. Clavier, ‘Channel Model and Performance of ad hoc Networks Based on IR-UWB at 60 GHz’, 4th ESA Workshop Millimetre Wave Techn. Applic., Feb. 2006.