A Comparison of Mean Power Prediction with Wide-Band Measurements in the Indoor Environment Zoran Blažević (1) , Igor Zanchi (2) , Ivan Marinović (3) 1, 2, 3 University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture, Split, Ruđera Boškovića bb, 21000 Split, Croatia, tel. + 385 21 30 57 77 fax. + 385 21 56 38 77 E-mail: 1 zblaz@fesb.hr, 2 izanchi@unist.hr, 3 imarin@fesb.hr Abstract: The aim of this paper is to try to exploit the results of impulse response measurements for testing the results of the path-loss predictions obtained by empirical or semi-empirical models. The comparison relies on the fact already confirmed in scientific references that the wide-band path-loss can be efficiently approximated by the minimum time-domain path-loss. An additional assumption that the wide-band path-loss obtained for a wide-band transmission could be approximated by the mean path-loss of a corresponding continuous- wave transmission has been checked by the measurements conducted with a vector network analyser. A comparison of the results shows that, besides various wide-band channel parameters, the impulse response measurements could serve well for testing the predictions of mean power obtained by path-loss models. 1. INTRODUCTION Mobile propagation channel modelling is one of the key tools for a mobile radio system design. The results of particular analysis based on the model can be different parameters that must determine examined radio channel in a useful way and with a sufficient accuracy. Models could roughly be classified as empirical or semi-empirical and physical. Empirical and semi-empirical models are based mostly on intensive measurements in various types of environments and derived statistics. These procedures often offer several relations for the propagation path-loss with various limitations in use, each of them valid for different frequency band or different type of environment. What they do not provide is any information about the receiving signal nature besides the statistically based predictions of mean power regarding estimated influence of parameters such as the frequency, the type of area, the transmitter angle, the average scatterer density etc. Nevertheless, they could be extremely useful in design of the cellular radio networks, e.g. for predicting the coverage of a cell. Measurements of the power of a continuous-wave (CW) reception are usually used for testing some empirical propagation model. However, for the purpose, there also exists a possibility for exploiting the wide-band measurements such like the radio-channel impulse response measurements. Namely, as is shown in [1] and [2], the minimum time-domain path-loss represents a reasonable approximation of the wide-band path-loss. Therefore, it can be estimated directly from the channel impulse responses, simulated or measured by a vector network analyser (VNA) or by some other suitable device, which thus leaves us with an option to make comparisons with the results of some corresponding model for the mean path-loss estimation. The example of a semi-empirical model from [3] taken for the purpose of comparison is explained in section 2, whereas the impulse measurement setup is described in section 3. The results of four measurements conducted in an indoor environment and the comparison with predictions of the mean path-loss obtained by the semi-empirical model is presented and analysed in section 4. 2. PATH-LOSS DEFINITIONS The absolute path-loss L between the transmitter and the receiver is the ratio of transmitted P t and total received power P r and thus represents inverse square absolute of the transfer function H of a radio channel. The absolute path-loss is dependent on frequency, and is commonly expressed in decibels as: ( ) ( ) 20 log 10 log r t P L f f P =− =− H . (1) Consequently, the wide-band path loss L WB in decibels is defined as: ( ) 2 0 10log c c f WB f L f f df + − = + ∫ H , (2) where it is supposed that the bandwidth of 2f c around the central frequency f 0 is examined so that the transfer function is supposed not to exist outside the band. The time-domain path-loss L t is defined via the band-limited channel impulse response, that is via the inverse Fourier Transform (FT) of the channel transfer function as: ( ) ( ) 20 log t L t t =− h , () ( ) { } 1 0 t f f − =ℑ + h H , (3) where sign ℑ -1 denotes inverse FT and h(t) represents the complex base-band representation of the channel impulse response. If a wide-band transmission in a multipath propagation environment is concerned then some sort of averaging is reasonable to expect, since fading at some frequency from the band does not mean the same amount of fading at a different frequency apart in the band. On the other hand, the mean-path loss L AVG is defined as the average of the