1176 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 53, NO. 4, JULY 2004

Transcription

1 1176 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 53, NO. 4, JULY 2004 Frequency Domain Characterization of LoS Nonfading Indoor Wireless LAN Channel Employing Frequency and Polarization Diversity in the GHz Band Akram Hammoudeh and David Anthony Scammell Abstract This paper presents results of frequency domain measurements conducted to characterize the distortionless transmission bandwidth (DTB) of indoor nonfading channels employing vertically and horizontally polarized antennas in the frequency band GHz. The mean delay spread ( mean ), root mean square (rms) delay spread ( rms ), and the DTB of the channel are also presented as functions of distance between terminals and are compared for both polarizations. The dependence of DTB on the separation between terminals is modeled as DTB = where is a constant. mean increases linearly with, and its relationship with DTB is characterized as DTB =(1 mean )+, where and are constants. The effectiveness of frequency and polarization diversity in mitigating the effects of multipath fading in indoor channels has also been evaluated. The performance of both diversity techniques when modulated signals with high data rates for multimedia applications are utilized is presented for maximum selection combining. The performance of frequency diversity is also shown as a function of frequency separation between diversity branch signals to determine whether an optimal frequency separation exists. Index Terms Coherence bandwidth, diversity, indoor channels, millimeter waves, multipath propagation. I. INTRODUCTION CONSUMER demand for the ever-increasing multimedia communication technologies, requiring large transmission bandwidths, is increasing exponentially [1]. However, there is a limit to how many users can be squeezed into the already overstretched frequency spectrum utilized for personal communication services. A solution would be to exploit the high-frequency millimeter-wavelength end of the spectrum, which was previously untapped by the communications industry [2] [4], particularly that around the 60-GHz region. Examples of multimedia applications requiring wireless transmissions over short distances together with estimates of required data rates are given in [2]. However, there is no universally accepted model that characterizes the radiowave propagation parameters that help to establish the performance and design of 60-GHz wireless local area network (WLAN) indoor communication channels. Manuscript received May 23, 2003; revised August 8, 2003, November 21, 2003, January 3, 2004, and March 12, The authors are with the School of Electronics, University of Glamorgan, Mid Glamorgan, Pontypridd CF37 1DL, U.K. ( amhammou@glam.ac.uk). Digital Object Identifier /TVT An accurate prediction of the radiowave propagation mechanisms [5] [10] for the development of new techniques, as well as system deployment, would have to be characterized before radio planners can design and implement networks signaling at high data rates. In the European Advanced Communications Technologies and Services (ACTS) program, the 60-GHz band has been addressed in a research project under MEDIAN with a target data rate of 156 Mb/s. In Japan, the Multimedia Mobile Access Communication (MMAC) committee is looking into the possibility of ultrahigh-speed 60-GHz wireless indoor LANs supporting 156 Mb/s. In mobile radio scenarios, multipath propagation causes the amplitudes and phases of the spectral lines over the transmitted bandwidth to vary randomly as the receiver moves away from the base station. In this case, the channel is characterized as a time-variant random channel or a fading channel. One of the key parameters that portrays information on the correlation during fading among different spectral lines over the transmission bandwidth is the frequency-correlation function. The bandwidth at which the magnitude of the frequency correlation function drops to a give value, normally 0.9, is used to characterize the coherence bandwidth of fading channels. This is necessary in order to establish whether such signaling rates are within the constraints of the channel s performance or whether countermeasure techniques such as equalization or diversity would have to be employed to help mitigate the effects of multipath propagation [11], [12] reducing intersymbol interference (ISI). A critical review of the estimation of frequency-correlation functions of a time-varying random channel (fading channels) is reported in [13]. The derivation of the equation relating to a channel s average power delay profile to its frequency correlation function via the Fourier transform is reviewed, with emphasis on the conditions needed for validity of this equation. An alternate and more robust method, free from most of the conditions required for the Fourier transform, is proposed in [13]. Such an alternative method is to directly cross-correlate time variations in the complex amplitudes of different spectral lines within experimental estimates of the complex frequency response against amplitude variations of the line at a reference frequency. This, however, is not valid for characterizing a WLAN radio link where multipath is present, /04$ IEEE

2 HAMMOUDEH AND SCAMMELL: FREQUENCY DOMAIN CHARACTERISATION OF LOS NONFADING INDOOR WIRELESS LAN CHANNEL 1177 Fig. 1. Measurement geometry in a long narrow corridor 35.6 m m m. Fig. 2. Frequency domain channel sounder block diagram. but does not vary with time, as the transmitter and receiver are at fixed positions with no objects moving in the line-of-sight (LoS) path. The channel in this case is characterized as a time-invariant nonfading channel and, more importantly, the coherence bandwidth definition does not apply. Changes in the multipath, if the WLAN scenario permits, are due to the movement of receiver from one location to another. A mistake that is frequently made in the literature is to characterize the coherence bandwidth from snapshot measurements of the frequency response using the complex autocorrelation function given by For nonfading channels, an alternative and more useful parameter for system designers is an indication of the distortionless (1)

3 1178 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 53, NO. 4, JULY 2004 Fig. 3. DTB estimation technique (a) amplitude response and (b) phase response. transmission bandwidth (DTB), over which the medium has a flat amplitude response and a linear phase response with frequency (constant group delay) [14], [15]. Any departure of the amplitude response from a constant value over the transmission frequency band gives rise to attenuation distortion. Similarly, any departure of the phase response from a linear function gives rise to phase distortion, also called delay distortion. An important cause of amplitude and phase distortions is due to multipath propagation, where the signal arrives at the receiver having traveled over more than one path with different time delays. The utilization of transmission bandwidths in excess of the DTB would result in an irreducible bit-error rate (BER). To achieve an improved error performance below, the irreducible BER equalization and diversity techniques are necessary.

4 HAMMOUDEH AND SCAMMELL: FREQUENCY DOMAIN CHARACTERISATION OF LOS NONFADING INDOOR WIRELESS LAN CHANNEL 1179 Fig. 4. DTB measured as a function of distance between terminals in a long narrow corridor using (a) VV and (b) HH antenna configurations together with the best theoretical fit. The snapshot impulse response of the radio channel is calculated from its complex frequency response using the inverse Fourier transform However, since the measurements were performed over a limited frequency band, there was a windowing effect on the results as (2) (3)

5 1180 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 53, NO. 4, JULY 2004 Fig. 5. Measured as a function of distance between terminals in a long narrow corridor (a) VV and (b) HH antenna configurations. In this study, a Hanning window with a sidelobe level (SLL) of 42 db has been used to estimate the channel s impulse response. For this nonfading stationary channel, the power delay profile is given by The rms delay spread is estimated from the power delay profile using (5) (4)

6 HAMMOUDEH AND SCAMMELL: FREQUENCY DOMAIN CHARACTERISATION OF LOS NONFADING INDOOR WIRELESS LAN CHANNEL 1181 Fig. 6. DTB against with best theoretical fit expressing their relationship for a long narrow corridor (a) VV and (b) HH antenna configurations. where is the coefficient of the power received from the th peak while represents its travel time. The mean delay spread is given by (6) This paper reports channel-sounding experiments made to measure the DTB,, and in the frequency band GHz in a long narrow corridor and compares transmission for vertically and horizontally polarized antennas under identical test conditions for nonfading LoS WLAN applications. The effectiveness of frequency diversity to reduce the effect of multipath propagation at different receiver locations

7 1182 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 53, NO. 4, JULY 2004 Fig. 7. Block diagram of a wide-band frequency-diversity receiver system, using maximum selection combining with diversity branch signals separated by one channels bandwidth (155 MHz) together with the frequency spectrum allocation. Number of channels n=2 and number of diversity branches n. Fig. 8. Normalized total average power of the diversity branch and selected signals versus distance obtained with vertically polarized antennas and 1f = 155 MHz. is also presented and is compared with that measured utilizing polarization diversity. The corridor is 35.5 m long, 1.65 m wide, and 2.76 m high. One of its sidewalls is constructed from plasterboard and had a number of doors leading off to other rooms. The other side of the corridor was also constructed from plasterboard containing four 3.4-m windows, overlooking the foyer to the building. The ceiling was made from suspended fiberboards and the floor was covered in plastic tiles. The corridor environment is considered as a worst-case scenario and is also a good model of other indoor situations. Measurements were made by transmitting a GHz frequency sweep signal from a fixed base to a receiver and recording the average of eight complex frequency responses at different receiver locations. This averaging process reduces the noise floor of the measuring system to a mean noise level of dbm, giving a dynamic range of 30 db at the maximum

8 HAMMOUDEH AND SCAMMELL: FREQUENCY DOMAIN CHARACTERISATION OF LOS NONFADING INDOOR WIRELESS LAN CHANNEL 1183 Fig. 9. Normalized total average power of the diversity branch and selected signals versus distance obtained with vertically polarized antennas and 1f = 155 MHz. separation between terminals in the corridor. The microcell base antenna was placed at one end of the corridor at a height of 1.73 m. The receive antenna was mounted at the same height on a 3-m-long high-precision computer-controlled positioning table. Measurements were made over a number of 2-m sections (Sections I IV) along the center line of the corridor at different distances from the transmitter, as depicted in Fig. 1. For each section, the receiver was automatically moved in increments of (0.58 mm) measuring 3448 complex responses. Measurements were taken with vertically polarized antennas at both the base station and receiver and were repeated under identical test conditions with horizontally polarized antennas. The measurement geometry and location of terminals are given in Fig. 1. The GHz frequency spectrum is divided into a number of channels of equal bandwidths that could accommodate modulated signals with 155 Mb/s. The effectiveness of frequency and polarization diversity techniques, with maximum selection combining, to reduce the effects of multipath propagation have been evaluated and presented for 155-MHz channels. Initially, frequency diversity has been analyzed with diversity branch signals separated by one channel s bandwidth. The performance of this diversity scheme is also characterized as a function of frequency separation between diversity branch channels. This paper is organized as follows. In Section II, the measurement hardware is introduced, followed by characterising and modeling the DTB,, and in Section III for both polarizations as functions of distance. The dependence of DTB on and is also investigated. Section IV presents the performance of frequency diversity for vertically and horizontally polarized antennas. It also compares and contrasts the effectiveness of frequency and polarization diversity schemes, followed by the conclusion in Section V. II. MEASUREMENT HARDWARE The measurement system is based on the sweep-frequency technique, which requires using a vector network analyzer (VNA). A block diagram of the channel sounder is given in Fig. 2. At the transmitter, the VNAs synthesized output is step swept between 1 3 GHz and then mixed with a 62.4-GHz signal producing upper and lower sidebands occupying frequencies between GHz and GHz, respectively. The 62.4-GHz signal is obtained from a phase-locked oscillator (PLO), which is synthesised from a 100-MHz external crystal. The lower sideband is suppressed by a bandpass filter centred at 64.4 GHz prior to transmission. At the receiver, a 62.4-GHz PLO is synthesised from the same 100-MHz external crystal via a 50-m cable. This is necessary in order to ensure that both phase-locked oscillators are coherent to accurately characterize the complex frequency response of the channel. The 1 3-GHz signal is coherently detected, amplified by a low noise amplifier, and then fed back through a second 50-m cable to the receive port of the VNA to measure the channel transfer function. The time resolution that can be achieved with a 2-GHz sweep bandwidth, when a rectangular window is used, is 0.5 ns. Measurements were conducted using vertically polarized horn antennas at transmitter and receiver (VV) and were repeated under identical test conditions using horizontally polarized antennas (HH). The horn antenna, with 10 dbi of gain, has a 69 and 55 - and -plane 3-dB beamwidths.

9 1184 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 53, NO. 4, JULY 2004 TABLE I FREQUENCY DIVERSITY GAIN TOGETHER WITH CORRELATION COEFFICIENTS BETWEEN DIVERSITY BRANCH SIGNALS OBTAINED WITH VV AND HH POLARIZATIONS AND 1f =155MHz III. NONFADING CHANNEL PARAMETERS FOR VERTICALLY AND HORIZONTALLY POLARIZED ANTENNAS TABLE II LEVELS ABOVE WHICH k-rician POWER RATIO RESIDES FOR 90% OF RECEIVER POSITIONS FOR VV AND HH POLARIZATIONS WITH 1f = 155 MHz A. Distortionless Transmission Bandwidth Prior to conducting measurements, the channel sounder was carefully calibrated inside an anechoic chamber with the terminals 1 m apart. Calibration was made for VV and HH polarizations. Indoor frequency-domain measurements for a given antenna configuration are then normalized with respect to the relevant calibration file. The DTB is estimated from the complex frequency response measured at each receiver position. All frequency ranges where the amplitudes are within 3 db below that of the resonance frequency, as shown in Fig. 3(a), are identified. The DTB is estimated as the largest possible bandwidth within the selected frequency ranges, where the phase response is linear, as shown in Fig. 3(b). The DTB measured over all sections is shown in Fig. 4(a) and (b) for both VV and HH polarizations, respectively, together with the best theoretical fit. It is clearly evident that the DTB decreases as the distance between the receiver and the base station increases. This dependence is modeled, for 95% confidence interval, as DTB (7) DTB (8) where DTB is in megahertz and in meters. Associated with the fitted equations in (7) and (8) are lower and upper 95% confidence bounds given by DTB and DTB for VV and DTB and DTB for HH, respectively. These results show that the value of the DTB is not influenced significantly by the antennas polarization. At large distances, the DTB obtained with horizontally polarized antennas is about 96% of that measured for vertical polarization. The smallest DTB value would be obtained at the maximum possible separation between transmitter and receiver. In this corridor, these values are 9.38 and 9 MHz for VV and HH polarizations, respectively. Setting the DTB to the appropriate value would ensure flat amplitude and linear phase responses of the channel over this bandwidth for all receiver locations within the microcell. A wireless system signaling at a rate requiring a transmission bandwidth larger than the DTB will experience amplitude and phase distortion. B. Mean and rms Delay Spread The mean and rms delay spread are estimated from the power delay profile. is found to increase linearly with the distance and is characterized for both polarizations as (9) where is in nanoseconds, in meters, and is a constant with values of and for VV and HH polarizations, respectively. The coefficients obtained for the 95% confidence bounds are almost identical to those predicted for the fitted functions. Fig. 5 shows as a function of distance for both polarizations. is observed to increase with separation up to a certain distance and then decreases at larger separations, as shown in Fig. 5(a) and (b) for both polarizations. The behavior of with distance (whole corridor) could not be mathematically modeled. Similar behavior is reported in [16], where measurements in a hallway at 1.6 GHz have shown an increase in with separations up to a distance of 10 m and fluctuates around some mean level at larger separations. It is estimated that in such environment is below 4.8 and 8.2 ns for vertically and horizontally polarized antennas, respectively, for 90% of receiver positions. These values are smaller than those reported by Valera and Sánchez [17], where rms values of up to 20 ns were measured in a corridor at MHz using vertically polarized omnidirectional antennas. This is mainly due to reflections off the back wall, behind the receiver, with long time delays being suppressed by the radiation pattern of the receive antenna. Values below 8.7 and 11 ns were reported for the rms delay spread in rooms using omnidirectional and directional receive antennas at 94 [18] and 60 GHz [19], respectively. C. Dependence of DTB on Mean Delay Spread Using (7) (9), the relationship between the DTB and can be characterized for both polarizations. Fig. 6(a) and (b)

10 HAMMOUDEH AND SCAMMELL: FREQUENCY DOMAIN CHARACTERISATION OF LOS NONFADING INDOOR WIRELESS LAN CHANNEL 1185 Fig. 11. Correlation coefficient versus frequency spacing measured for horizontally polarized antennas. Fig. 10. Frequency diversity gain measured from cdfs at 0.1% of the time for (a) vertically and (b) horizontally polarized antennas. shows the measured DTB against theoretical fit. together with the best DTB (10) DTB (11) where DTB is in megahertz and in nanoseconds. It has not been possible to characterize the dependence of DTB on. This is mainly because the behavior of with distance, as described earlier, could not be modeled mathematically. IV. CHARACTERIZATON OF FREQUENCY AND POLARIZATION DIVERSITY Results presented in Section III characterize the DTB in a long narrow corridor using vertically and horizontally polarized antennas. At large distances from the base station, the DTB is well below the bandwidth required to transmit data rates of 155 Mb/s, even if spectrum efficient modulation schemes are employed. Countermeasure techniques would have to be implemented to help mitigate the effects of multipath and, hence, improve BER performance [20]. The effectiveness of wide-band frequency-diversity system, employing maximum selection combining [21], is presented for 155-MHz channels that could be used to transmit 155 Mb/s using binary phase-shift keying (BPSK). A block diagram of the frequency-diversity scheme analyzed together with channel allocation is shown in Fig. 7. The diversity branch signals, for each channel, are separated by one channel s bandwidth resulting in maximum utilization of the frequency spectrum, as illustrated in Fig. 7(b). The performance of this diversity scheme is then presented for 155-MHz channels as a function of frequency separation between diversity branch signals. A frequency spacing that is not an integer multiple of the channel s bandwidth would not yield the most efficient spectrum utilization. This, however, allows system planners to determine whether an optimal frequency spacing to achieve a maximum diversity gain could be obtained for the whole environment. Analysis in this section is performed offline for one channel (channel 1) employing vertically and horizontally polarized horn antennas. A. Frequency Diversity Performance for 155-MHz Channels With MHz The total average power in each diversity branch channel [diversity branch 1 (lower frequency branch) and diversity branch 2 (upper frequency branch)] and the selected one have been computed at each receiver position relative to 1-m LoS channel measured inside the anechoic chamber and presented as a function of distance between terminals. Results obtained for VV polarization are given in Figs. 8 and 9. The diversity gain, at each receiver position, is obtained by subtracting the diversity branch envelopes from the selected envelope. The gain is observed to vary significantly with receiver location with the maximum and mean gain values, obtained for all sections given in Table I, for both polarizations together with

11 1186 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 53, NO. 4, JULY 2004 Fig. 12. Block diagram of a wide-band polarization diversity receiver system using maximum selection combining with 155-MHz channel bandwidth together with frequency spectrum allocation. the correlation coefficient between diversity branch signals. High values for the maximum diversity gains are measured. The mean diversity gain varies between 4.1 and 5.2 db for VV and 3.3 and 5.8 db for HH polarizations. It can be seen that the smaller correlation coefficient values do not appear to guarantee a larger average diversity gain. In addition, neither polarization appears to present superior frequency diversity performance over the other.

12 HAMMOUDEH AND SCAMMELL: FREQUENCY DOMAIN CHARACTERISATION OF LOS NONFADING INDOOR WIRELESS LAN CHANNEL 1187 The diversity gain measured over a section in the corridor has also been estimated from the cumulative distribution functions (cdfs) of the lower diversity branch and selected envelopes. This is realized by determining the amount by which the branch envelope would improve at a given percentage of time after selection combining. With vertically polarized antennas, the diversity gain obtained, for 0.1% of the time where the branch envelopes exhibit severe deep fading, over Sections I IV, is 9.1, 1, 1.8, and 3.9 db, respectively. Gain values of 5.1, 6.7, 2.1, and 4 db are measured with horizontal polarization. Larger gains are measured for HH polarization when the terminals are further apart. However, close to the transmitter the vertically polarized configuration shows higher gains. The effectiveness of this diversity scheme in improving the power ratio of the LoS relative to that in the multipath ( -Rician power ratio) has also been computed as a function of receiver position. This is accomplished by modeling the amplitude variation of the diversity branch signal and selected envelopes to a Rician distribution and determining the values of that yield the best fit. Data has been processed over a long window that is being moved by one. This ensures a large number of points (160) to provide accurate estimates. The value of is found to vary with receiver position and the level above which its value resides for 90% of receiver position before and after selection combining are measured and given in Table II for both polarizations. These results clearly show that maximum selection combining has resulted in larger values of, particularly with vertical polarized antennas. These larger values yield an improved BER performance of digitally modulated signals, depending on the modulation scheme used [20]. B. Frequency Diversity Performance for 155- MHz Channels as a Function of In this section, processing was conducted by keeping the channel bandwidth fixed at 155 MHz and varying the frequency spacing between the diversity branches of channel 1 in steps of 2.5 MHz up to 155 MHz. By doing so, the frequencies of the lower diversity branch channel remain unchanged while those for the upper diversity branch move up the frequency spectrum as increases.the diversity gain is measured from the cdfs of the selected and lower frequency diversity branch signal envelopes, for 0.1% of the time. Results for channel 1, as a function of, are shown in Fig. 10 for both polarizations. It can be seen that the diversity gain and the optimal frequency spacing to achieve maximum diversity gain vary from one section to another. The correlation coefficient between the diversity branch envelopes of channel 1 is computed as a function of. Results obtained with horizontally polarized antennas are given in Fig. 11. Generally, as expected, the correlation coefficient exhibits a decreasing trend with larger. However, the lower correlation values measured with large do not necessarily guarantee larger diversity gain. It can also be seen [Figs. 10(b) and 11] that the sections with higher diversity gains do not necessarily exhibit lower correlation coefficients. Unlike con- TABLE III POLARIZATION DIVERSITY GAIN TOGETHER WITH CORRELATION COEFFICIENTS BETWEEN DIVERSITY BRANCH SIGNALS FOR 155-MHZ CHANNELS TABLE IV POLARIZATION DIVERSITY GAIN OBTAINED AT 0.1% OF THE TIME AND LEVELS OF k-rician POWER RATIO MEASURED FOR 155-MHZ CHANNELS tinuous wave (CW) diversity systems [22], [23], the correlation coefficient between the branch envelopes measured for all section with no frequency separation is below one. This is because each of the diversity branches has a number of frequency components that occupy a bandwidth, not one frequency component as in CW systems. C. Polarization Against Frequency Diversity Performance for 155-MHz Channels The performance of wide-band polarization diversity system, employing maximum selection combining, has been evaluated for 155-MHz channels and compared with that of frequency diversity. A block diagram of the polarization diversity scheme analyzed together with channel allocation is shown in Fig. 12. The maximum and mean polarization diversity gains obtained from subtracting the diversity branch envelopes from the selected one are shown in Table III for all sections in the corridor. These values are comparable to those measured for frequency diversity (Table I). The diversity gain obtained from the cdfs of the selected and vertically polarized signal envelopes, at 0.1% of the time, together with the improvement in the level of the -Rician value are given in Table IV. The gain measured with polarization diversity is generally higher than those for frequency diversity. However, the increase in the values of the -Rician power ratio after selection combining is generally lower in polarization diversity. V. CONCLUSION Results of sounding experiments made in the frequency range GHz to characterize the wide-band parameters as well as the performance of frequency and polarization diver-

13 1188 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 53, NO. 4, JULY 2004 sity in an indoor nonfading channel are reported. It is shown that the coherence bandwidth definition does not apply for nonfading channels and an alternative parameter is proposed. This parameter characterizes the DTB where flat amplitude and linear phase responses with frequency are measured. It has been shown that, for both polarizations, the distortionless transmission bandwidth of the channel decreases as distance between transmitter and receiver increases and is modeled as DTB where is a constant. The channel is free of amplitude and phase distortion over a bandwidth of 9.38 and 9 MHz at all receiving positions for vertically and horizontally polarized antennas, respectively. This is determined by the value of DTB obtained at the largest possible separation between transmitter and receiver in the corridor. These values are too small to accommodate modulated signals with data rates in the region of 155 MB/s even if spectrum efficient modulation schemes are employed. Wireless systems requiring transmission bandwidths larger than the channel s DTB experience amplitude and phase distortion. The performance of frequency and polarization diversity schemes to mitigate these effects has been measured and compared. The mean delay spread is independent of antennas polarization. It increases linearly with distance and its relationship with DTB is characterized by a function of the form DTB, where and are constants and their values are polarization dependent. However, the rms delay spread is found to increase with distance up to a certain separation and then drops at larger distances; hence, its relationship with DTB is not clearly identified. The 90% values measured for vertical and horizontal polarizations are below 4.8 and 8.2 ns, respectively. The effectiveness of frequency diversity to mitigate the effects of frequency-selective fading and multipath propagation is characterized for both VV and HH polarized antennas using maximum selection combining. Separating the branch signals by one channel s bandwidth yields comparable performance and, more importantly, maximum spectrum efficiency. The diversity gain measured for 155-MHz channels as functions of receiver positions within the microcell varies significantly. The mean diversity gain measured with VV polarization is between 4.1 and 5.8 db with the maximum gain of 22.6 db, whereas the mean gain measured for HH polarization varies between 3.3 and 5.8 db with a 19.6-dB maximum gain. Results have also shown that selection combining yields higher values of the -Rician power ratio, which leads to an improved BER performance. Neither polarization has shown consistent superior performance over the other. The performance of polarization diversity is also found to be comparable to frequency diversity. The diversity gain, measured for 0.1% of the time, is generally higher than those for frequency diversity. However, the improvement in the -Rician power ratio is generally lower. These results show no clear advantage of using one diversity technique over the other. However, the primary factors that should be considered when utilizing a diversity scheme are system complexity and user density within the allocated frequency spectrum. REFERENCES [1] L. M. Correia and R. Prasad, An overview of wireless broadband communications, IEEE Commun. Mag., pp , Jan [2] P. Smulders, Exploiting the 60 GHz band for local wireless multimedia access: Prospects and future directions, IEEE Commun. Mag., vol. 40, pp , Jan [3] P. W. Huish, I. T. Johnson, T. A. Li, and M. J. Nicholls, MM-wavelength micro-cellular mobile radio systems: A solution to spectrum congestion for high user densities?, presented at the Int. Conf. Land Mobile Radio, Cambridge, U.K., Dec [4] J. P. McGeehan and K. W. Yates, High capacity 60 GHz microcellular mobile radio system, Telecommun., vol. 20, no. 9, pp , [5] X. Hao, V. Kukshya, and T. S. Rappaport, Spatial and temporal characteristics of 60-GHz indoor channels, IEEE J. Select. Areas Commun., vol. 20, pp , Apr [6] Hammoudeh and G. Allen, Millimetric wavelengths radiowave propagation for line-of-sight indoor microcellular mobile communications, IEEE Trans. Veh. Technol., vol. 44, pp , Aug [7] H. Droste and G. Kadel, Measurement and analysis of wideband indoor propagation characteristics at 17 GHz and 60 GHz, presented at the Proc. Inst. Elect. Eng. Antennas Propagation Conf., Apr [8] T. Manabe, Y. Miura, and T. Ihara, Effect of antenna directivity and polarization on indoor multipath propagation characteristics at 60 GHz, IEEE J. Select. Areas Commun., vol. 14, pp , Apr [9] P. F. M. Smulders, Frequency domain measurement of the millimeter wave indoor radio channel, IEEE Trans. Instrum. Meas., vol. 44, pp , Dec [10] K. Sato, T. Manabe, T. Ihara, H. Saito, T. Tanaka, K. Sugai, N. Ohmi, Y. Murakami, M. Shibayama, Y. Konishi, and T. Kimura, Measurements of reflection and transmission characteristics of interior structures of office building in the 60-GHz band, IEEE Trans. Antennas Propag., vol. 44, pp , Dec [11] G. A. Kalivas, M. El-Tanany, and S. Mahmoud, Millimeter-Wave channel measurements with space diversity for indoor wireless communications, IEEE Trans. Veh. Technol., vol. 44, pp , Aug [12] G. Allen and A. Hammoudeh, Frequency diversity propagation measurements for indoor 60 GHz mobile radio link, presented at the 7th Int. Conf. Antennas and Propagation (ICAP 91), Your, U.K., Apr [13] R. J. C. Bultitude, Estimating frequency correlation functions from propagation measurements on fading radio channels: A critical review, IEEE Select. Areas Commun., vol. 20, pp , Aug [14] I. Otung, Communication Engineering Principles. Basingstoke, U.K.: Palgrave, Houndmills, [15] B. P. Lathi, Modern Digital and Analog Communication Systems. Oxford, U.K.: Oxford Univ. Press, [16] H. Zaghloul, Comparison of indoor propagation channel characteristics at different frequencies, Electron. Lett., vol. 27, no. 22, pp , Oct [17] M. S. Varela and M. G. Sánchez, RMS delay and coherence bandwidth measurements in indoor channels in the UHF band, IEEE Trans. Veh. Technol., vol. 50, pp , Mar [18] A. Kajiwara, Effects of polarization, antenna directivity, and room size on delay spread in LOS indoor radio channel, IEEE Trans. Veh. Technol., vol. 46, pp , Feb [19] T. Manabe, K. Sato, H. Masuzawa, K. Taira, T. Ihara, Y. Kasashima, and K. Yamaki, Polarization dependence of multipath propagation and high-speed transmission characteristics of indoor millimeter-wave channel at 60 GHz, IEEE Trans. Veh. Technol., vol. 44, pp , May [20] R. P. Torres, B. Manteca, S. Loredo, and L. Valle, Site-specific BER simulation for indoor wireless communications, presented at the IEEE Vehicular Technology Conf. (VTC 01), Rhodes, Greece. [21] W. C. Jakes Jr, Microwave Mobile Communications. New York: Wiley, [22] H. J. Thomas, R. S. Cole, and G. L. Siqueira, An experimental study of the propagation of 55 GHz millimeter waves in an urban mobile radio environment, IEEE Trans. Veh. Technol., vol. 43, pp , Feb [23] M. G. Sánchez, A. Hammoudeh, J. P. Kermoal, and E. Grindrod, Coherence bandwidth characterization in an urban microcell at 62.4 GHz, IEEE Trans. Veh. Technol., vol. 49, pp , Mar

14 HAMMOUDEH AND SCAMMELL: FREQUENCY DOMAIN CHARACTERISATION OF LOS NONFADING INDOOR WIRELESS LAN CHANNEL 1189 Akram Hammoudeh received the B.Sc. degree in electrical and electronic engineering from Yarmouk University, Irbid, Jordan, in 1987 and the Ph.D. degree from the University of Bath, Bath, U.K., in He joined the School of Electronics, University of Glamorgan, Pontypridd, U.K., in 1991, as a Postdoctoral Research Fellow in radiowave propagation at microwave and millimeter wave frequencies, before being appointed as a Senior Lecturer in 1993 and a Principal Lecturer in His research interests include studies of indoor and outdoor channel characterization at millimetric wavelengths, mobile radio propagation, multipath countermeasures, channel-modeling techniques, and radiowave propagation through woods and forests. David Anthony Scammell received the B.Sc. degree in information technology and the M.Sc. degree in electronic product engineering in 1996 and 1997, respectively, from the University of Glamorgan, Pontypridd, U.K., where he is currently working toward the Ph.D. degree on wide-band millimeter channel modeling and countermeasure techniques. He was an Electronic and Embedded Systems Design Engineer and in 2001 he was appointed as a Senior Lecturer in the School of Electronics, the University of Glamorgan, where he teaches telecommunication, programming, and embedded design courses.