Indoor and outdoor frequency measurements for mm-waves in the range of 60 GHz

Transcription

1 Indoor and outdoor frequency measurements for mm-waves in the range of 6 GHz Dusan M. Matic 1, Hiroshi Harada and amjee Prasad 1 1 elecommunication & raffic-control Systems Group, Department of Electrical Engineering, Delft University of echnology P.O. Box 531, 6 GA Delft, etherlands D.Matic@E.UDelft.L, Fax: , el: Broadband mobile communication section, Yokosuka adio Communications research centre, Communication esearch Laboratory, Ministry of Post and elecommunications, Japan Abstract Using a spectrum analyzer and synthesized signal generator, several measurements for the frequencydomain characterization of the radio channel in the 6 GHz band were performed. In the experiments, a carrier wave (CW) was swept with a constant amplitude across the 1 MHz band, centered around 59.9 GHz. he channel frequency response was measured at the receiver side. he results can be used for the design of an OFDM (Orthogonal Frequency Division Multiplex) system. he measurement environments were a corridor and a big college room in a high-rise office building (indoor) and a parking and a grassfield at the side of that building (outdoor). he spectrum samples were taken in each of the environments. From such samples, it is possible to calculate the ice parameter k and the path loss coefficients. It was necessary to develop an appropriate measurement analysis, stemming from the fact that the spectrum analyzer gives no information on the phase, only amplitude. Measurement analysis method and results are explained. I IODUCIO In order to develop a good mobile communication system, a good understanding of the radio propagation is crucial. he bandwidth is considered as wideband, when it is bigger than coherence bandwidth B C. Within coherence bandwidth, flat fading exists, i.e., there is a large correlation between components of the spectrum (they fade synchronously). If a signal has a bigger bandwidth than B C, drops in the signal level can be observed in some parts of the spectrum. his phenomenon is called frequency fading. he reason for this is that a number of reflected signals approach receiver antenna and sum together. his can be destructive for some frequencies and constructive for other. Frequency fading measurements can give important design information for a transmission technique like OFDM (Orthogonal Frequency Division Multiplex) with its multiple carrier-components, which is a very promising candidate for future MBS (Mobile Broadband Systems). hese characteristics give an idea of the level of signal distortion in frequency domain. 6 GHz measurements were done almost exclusively in the time-domain by using network analyzers [1] and channel sounders. etwork analyzers determine the complex channel response by sending a narrow pulse and by observing the effect of the channel on the received signal. Both amplitude and phase are measured at the same time. While being very precise and sophisticated, they are also very complicated to use. hey are not very transportable as well, because of the difficult supply of the (reference) signal at 6 GHz. Channel sounders are movable and cheap, but the available ones give only a magnitude of time-domain response. he new channel sounder systems with coherent demodulation, enabled by combination of radio and optical technologies, will broaden the possibilities of this method []. his paper presents a way to measure the channel frequency response by sending a sweeping CW and measuring the response with a spectrum analyzer. Usage of this measurement method enables constant power envelope (as opposed to the time-domain measurements). his reduces effect of nonlinearities and allows greater areas to be measured. Also, the set up for the measurements is easier, it is flexible and takes short time for the measurements. his paper presents the results of the frequencyfading measurements over 1 MHz bandwidth, which was centered around 59.9 GHz. For each point a frequency profile was measured. From this data is possible to extract k factor of the ice distribution and path-loss coefficients. Different spatially distributed locations throughout the test area were used by fixing the transmitter and moving the receiver. he results of many points in an environment give a possibility to make statistical distributions. he scenario was that there was a omnidirectional base (transmitter) station antenna. wo different receiver antennas were used for receiving, omnidirectional and directional. his was done in order to compare the difference in reception of the two. In this paper is given a detailed description of the measurement setup, environments, measurement analysis and conclusions.

2 II MEASUEME SYSEM he block diagram of the measurement system [3] used for the frequency-domain characterisation of the radio-channel is shown in Fig. 1. ASMIE SWEEPIG SIGAL GEEAO 59.9 GHz BW : 1 MHz SPECUM AALYZE Fig. 1. Measurement setup ECEIVE GPIB LAPOP PC he two main components are synthesized signal generator and a spectrum analyzer. he CW signal generated is fed directly to the intermediate frequency (IF) of the transmitter. he output of the F power amplifier is radiated through an omnidirectional antenna. he energy collected at the receiver antenna is translated to the IF of the receiver. hat signal is fed to the spectrum analyzer to determine the frequency response of the channel. he measured data is then imported by the PC controller and stored for later analysis. he measurement setup is illustrated in Fig. 1. he used equipment were namely, synthesized signal generator (Marconi ), 6 GHz transmitter and receiver and spectrum analyzer (Anritsu MS651A). From the spectrum analyzer, data was transferred to a laptop PC via GPIB (General Purpose Information Bus). he frequency response consists of 5 amplitude samples at a frequency spacing of. MHz, as shown in Fig.. For each point, the distance was also measured. At the transmitter side was used a flat omnidirectional antenna ( dbi, 1 ). At the receiver side, omnidirectional (1 ) and patch directional antenna (pencil beam, 19.5 dbi, 15 ) were employed. Measurements with both were done in order to see the difference in performance, because omnidirectional antenna allows for more reflected components to enter the receiver. III DESCIPIO OF MEASUEMES In this paper, the frequency response measurements are spatially distributed throughout the test area, by fixing the transmitter in a likely place and moving the receiver to different locations. he locations selected are based on the assumption of ubiquitous possibility of mobile multimedia communication using a small terminal, indoor and outdoor. he objective of experiments is to determine the effect of location to the frequency fading. In all experiments line of sight (LOS) existed. he area covered in each case is on the order of picocell (smaller than 5 m radius). Antennas were still during measurements. he results from four environments are presented. he first set of measurements was obtained from a corridor (indoor environment I1), shown on the Fig. 3, located on the 19 th floor of a floor building, situated in the centre of the Delft University campus. here are no other high buildings around it. he corridor is outlined with walls and office doors, except on the ends, where are glass surfaces. A total of 18 points was taken along 5-11 m distance in the middle of the corridor. A step of cm was chosen in order to avoid correlation between samples. he measurements were done two times, first with a directional and second time with a omnidirectional receiver antenna in order to see the difference in received signal. he transmitter s main ray was in both cases radiating along the corridor. he antenna heights were 1.8 m. eceived power [dbm] Frequency [GHz] Fig.. An example of a frequency response measurement made with the spectrum analyzer (the magnitude in [dbm]) m 1m5 m8 11m cm 5m door 5 m Window Figure 3. Wideband measurements in the I1 ( corridor, 5-11 m in steps of cm) he second experiment took place in the faculty s amphitheater (indoor environment I). his was done in order to see the propagation in large rooms, such as large lecture halls, airports, factories, etc. Here were two types of measurements. In both was used an omnidirectional receiver antenna. What was changed was the base station (transmitting) antenna. In one case it was placed at the bottom (on the cathedra), in the second it was in the last, highest row of seats. he signal was measured in 5 seat positions

3 7m 1 m Bushes Car Car scattered round the amphitheater, as shown for the case of the transmitter up on the Fig.. he third group of measurements took place outside, along the building (outdoor environment O1), environment shown on Fig. 5. he receiver antenna was moved away in a straight line from 1 to 7 m distance with a step of 1m. he height of antennas was 1.59 cm. he final, fourth experiment took place on an empty parking (Fig. 6) which was bordered with trees (outdoor environment O). he antennas height was 1.59 m. he experiment was done with both types of receiver antennas, on distances of about 3 to 5 m, with 5 points in a straight line and 1 on the sides of the parking. 18m m or 3m 15m 3 1m m5 m1 9 m18 Fig. Wideband measurements in I, in 5 receiver positions with transmitter up at college room A, omni-directional receiver antenna E-building 3 m Bushes 5m 8m m 1m 8m ree Lamppost Fig. 6. Map of the parking (O) Only straight line measurements are shown. Scattered points are not in the figure. here was no rain in case of outside measurements. IV MEASUEME AALYSIS he main objective of the indoor radio wave propagation measurements is to determine the radio-coverage and data-rate limitations. [,5,6]. he radio-coverage is related to the distance -power relationship in the area, and data rate is limited by the frequency selective fading multipath characteristics of the channel. In this paper are derived the statistics of the k factor and the path-loss coefficients. A. ice factor and the power of the direct wave Container m8 8m5 9m he sum of the direct and reflected waves can be expressed as follows in time and frequency domain. 1m9 m5 Grass field 1m ree 31m5 1m6 j h( t) = s e δ ( t τ ) H( jf ) = β e m θm F m m j( πτ f m+ θm ) m83 Footway Fig. 5. Wideband measurements with omni-directional antennas in a straight line at the grassfield (O1) where s m is mth signal reflection, β m is the amplitude gain, θ m is the phase, τ m is the time delay. he spectrum analyzer can measure only the power of the reflected signal. he average received power is given by:

4 H( f ) = P = k ice parameter k is defined as [7] k = β 1 β β (1) m where β is the direct path amplitude. By using (1) and () is obtained k + 1 P = k () β (3) In this equation there are still two unknown variables. o make a system of two equations, in order to find k, the following expression is used ( ) σ P ( f ) = P ( f ) P( f ) () where σ P( f ) is the variance of P(f). In order to simplify the derivation and solve the system, the assumption was taken that the power of the LOS wave is much stronger than the sum of the powers of reflected waves, because measurements were always done with aligning of the antennas, i.e. βm β (5) Using (5) to simplify the solution of () gives σ P( f ) k + 1 = β k From the system of (3) and (6), it is possible to calculate k and β. Being the solution of a quadratic equation, there are two roots for k, one approximately, and the other with a large value. he physical meaning of k was used as a criterion to choose one of them. Since there was always a LOS and one root is always approximately zero, it was assumed that it was the large one. Fig 7 shows changes of the k factor in the corridor (I1), when the directional antenna was used. From all the samples in an experiment were calculated average values and deviation of k. able I summarizes the results. (6) ice Factor K (lineair) distance (m) Fig 7. ice factor k versus distance in the corridor. Directional receiver antenna used. he k scale is (,1) ABLE I Average value k and standard deviation σ k of ice factor k over all locations in an experiment environment Experiment k σ k I1 (corridor) dir. antenna omnidir. ant. I (large room) up down O1 (grassfield) O (parking) omnidir ant. dir. antenna B. eceived power vs. Distance For a fixed transmitter power (P t ) the received power (P r ) decreases with distance (d) as P ( r d ) = Ad where α is the exponent of the power-distance relationship and A is a constant set by transmitted power and measurement system gain [8]. When logarithm of (7) is taken, the linear form is 1 log 1[ Pr ( d)] = 1 log 1[ A] 1α log 1[ d] (8) In this paper, the power is averaged all 5 points of the measured frequency response. In the previous section was explained how was extracted β, the power of the direct ray from the data. In the able II are given values for α for different environments and for the direct ray. he r is the correlation between the measured and modeled. α (7)

5 ABLE II Values of α, A[dB], derived from the linear regression of power [db] on the 1 log 1 of distance for each of the four global experiments and calculated from β Experiment I1 omni dir I up down α A[dB] r α( β ) A (β ) r , O O omni dir Fig 8.shows the trendline for the environment O (parking). he α is almost, as for the free-space. Correlation r is.9. eceived power[db] y = -.973x = distance [db(m)] Fig. 8. A scatter of the plot of the measured power [db] versus the distance on a log scale for experiment O (parking) with omnidirectional antenna. Also shown is the line with the minimum mean square error fit to the data. Fig 9. Shows the changes of the level of the wideband average of the received power with distance, with omnidirectional antenna used. he step was cm. As it is seen, the average decay with distance is small. his can be explained with the fact that there are many reflections and that they sum up together. However, this leads to small k factor and to the fact that the modeling with the eq. (8) is not so good (r is low). V COCLUSIOS Indoor radio propagation measurements in the GHz frequency band were conducted using a spectrum analyzer. he objective of the experiments was to determine the amount of frequency fading of the received signal. hese experiments were performed in a corridor on the 19 th floor and a big college room in a -floor office building of the Department of Electrical Engineering, Delft, he etherlands (indoor) and a parking and a grassfield at the side of that building (outdoor). he measurements were done in a possible locations for the mobile multimedia communication. A method is developed for the extracting the ice factor and path-loss exponent from the frequency response data base.. received power [dbm] distance [m] Fig. 9. Broadband average received power in the corridor with omnidirectional receiver antenna used ACKOWLEDGMES he authors would like to thank Gerard Janssen for the fruitful discussions. hey would also like to thank student Arthur Mank for his contribution in the measurements. hey would like to express their gratitude to EC and WO (Dutch foundation for the development of science) for their support of this research. It is a part of the MMC (Mobile Multimedia Communication) project. More information can be found on ( EFEECES [1] P.M.F. Smulders, Broadband wireless LAs: A feasibility study, Ph.D. thesis, Eindhoven university of echnology, he etherlands, ISB X, 1995 [] P.E. Leuthold and P. ruffer, Indoor radiowave propagation measurements and stochastic channel modeling Proc. of the IC colloquium on indoor communications, Delft, he etherlands, [3] D. Matic, H. Harada, H. eijmers,. Ueno and. Prasad, Measurements of the indoor spatial distribution of the radio-field level for mm-waves in the range of 6 GHz", Proc. IEEE fifth symposium on communications and vehicular technology in the Benelux, pp. 8-88, Oct ,1997. Enschede, he etherlands, ISB [] W. Jakes, "Microwave mobile communications", IEEE Press, reprint 199, ISB [5] W.C. Lee, Mobile communication design fundamentals,. Ed, John Willey & Sons, 1993, ISB [6].S. appaport, "Wireless communications; principles and practice", Prentice-Hall, 1996, ISB [7] P. Vasconcelos and L.M. Correia, Fading characterization of the mobile radio channel at the millimetre waveband, Proceedings. of the VC 97, pp , Phoenix, Arizona, USA, May 1997 [8] S.J. Howard and K. Pahlavan, Measurement and analysis of the indoor radio channel in the frequency domain, IEEE rans. Instrumentation and measurement, vol. 39,pp , Oct. 199