MoPro Chapter 8.4

8.4 Slant-Path Building Attenuation Measurements from 0.5 to 3 GHz

In this section we describe the results of Vogel and Torrence [1995c] who executed a comprehensive set of measurements in which they examined attenuation levels over swept frequencies ranging from 500 MHz to 3000 MHz into a set of six buildings. These results extend those of Vogel and Torrence [1993] who examined fading within buildings over the frequency range 700 MHz to 1800 MHz (described in Section 8.2). In the effort described here, Vogel and Torrence used a similar experimental configuration employing an antenna mounted on a tower which simulated transmissions from a satellite platform and which radiated right hand circular polarization signals (RHCP). The receiving antenna received RHCP over a beamwidth of 90° in both the vertical and horizontal principal planes. As in the previous experiment, the receiving antenna inside each of the buildings was located on a positioner that moved it in 5 cm increments over 16 spatial positions (80 cm).

8.4.1 Experimental Description

The major components of the experiment of Vogel and Torrence were a 20 m crank-up transmitter tower mounted on a van, a vector network analyzer (VNA), a personal computer (PC), and a linear positioner. One of the two wideband antennas was mounted atop the tower outside a building and the other at the positioner inside where the relative signal levels (in dB) were determined. As previously described, the relative signal levels were determined by comparing the levels measured interior to each of the buildings with the reference levels measured for the unobstructed line-of-sight case outside. The PC controlled the VNA and linear positioner, and stored the data from the VNA. The relative signal frequency was swept between 500 MHz to 3000 MHz at 1 MHz increments and the data were analyzed at a step bandwidth of 0.2% of the frequency. At the low frequency end, the sampling increment was therefore 1 MHz (500 MHz x 0.002) and at the high end it was 6 MHz (3000 MHz x 0.002). Measurements were made into six different buildings during the Spring of 1995. The names of these buildings and pertinent parameters are summarized in Table 8-13. The Commons, EERL, the Farmhouse and the Motel were sites also used in the L- and S-Band investigations [Vogel and Torrence, 1995a; 1995b] described in Section 8.3. Photographs of the exterior buildings and floor layouts with measurement locations within each of the buildings listed in Table 8-13 are given by Vogel and Torrence [1995c].

Table 8-13: Building names and pertinent characteristics.

Building Name	Year of Construct.	Type	# of Stories	Roof Type	Avg. El. (°)	No. of Locat.
Commons Entry	1987	concrete wall	1	tar	18	7
EERL	1944	block brick	1	tar	38	6
Farmhouse	1880	wood frame	2	metal	33	12
House	1976	wood frame	2	metal	41	10
MER	1992	glass, concrete	2	tar	26	9
Motel	1980	brick	2	composition	37	8

8.4.2 Average Signal Levels over Frequency Band and Positions

Figure 8-13 summarizes the overall relative signal loss results at the various building locations. The mean, standard deviation, minimum and maximum values were obtained by first averaging the signal loss over the frequency band at each position (e.g., average at approximately 900 frequency values). The averages of these means were then taken over all locations (e.g., 16 positions of positioner times number of locations within the building as given in last column of Table 8-13). Table 8-14 lists the mean value and standard deviations given in Figure 8-13. The overall average is shown to be -11.9 dB with an average standard deviation of 7.4 dB (last row of table). The mean relative signal losses vary between -5 dB (in Farmhouse) and -19.5 dB in the House. Although both have metal roofs, the House also has an aluminum heat shield in all exterior walls. The other four buildings fall into the range of approximately -9 dB to -13 dB. These results also include data in which the receiving antenna was located in an open door or window in sight of the transmitter.

Figure 8-13: Relative signal level statistics at each of the building locations.

Table 8-14: Listing of the mean and standard deviations when relative signal loss is averaged over position and frequency (0.5 to 3.0 GHz).

Site	Mean (dB)	Standard Deviation (dB)
Commons	-13.2	10.0
EERL	-9.2	6.9
Farmhouse	-5.0	3.8
House	-19.5	11.6
MER	-11.5	5.4
Motel	-13.1	6.6
Overall Average	-11.9	7.4

8.4.3 Distance and Frequency Dependence

Figure 8-14 shows an example describing the relative signal level versus position at four frequencies extending over approximately 1 GHz to 3 GHz for the Commons. We note the relative signal varies between -15 to -30 dB for scenarios in which the direct path penetrates a concrete wall (e.g., 0 to 3 m). It increases to the range of -5 to +5 dB as the antenna moves across the window area and ends up outside the open door (e.g., 4 to 5.5 m). By contrast, an extreme case is depicted in Figure 8-15 for the House that shows levels ranging from -25 dB to -45 dB. These low fade levels are presumed to occur because the construction includes a metal roof and a tight energy conserving aluminum heat shield under the cedar exterior.

Figure 8-14: Relative signal level versus position at four frequencies in the Commons.

Figure 8-15: Relative signal level versus position at four frequencies in the House.

8.4.4 Spatial Autocorrelation Characteristics

The autocorrelations were calculated at approximately 900 frequency samples for lags from 0 m to 3.2 m. Figure 8-16 gives an example of the spatial autocorrelations versus frequency for the EERL Building at a lag of 50 cm. The overall data do not exhibit a clear frequency dependence in spite of the fact that the wavelength at 500 MHz is 60 cm whereas it is only 10 cm at 3 GHz. The median value of the autocorrelations in Figure 8-16 is 0.37. The autocorrelation as a function of distance at 1625 MHz is plotted in Figure 8-17 for the Commons, EERL, and the Farmhouse. In Table 8-15 are listed the median decorrelation distances (i.e., distance at which autocorrelation equals 0.37) for each of the building sites. It is noted that the overall average is 0.71 m.

Figure 8-16: Spatial autocorrelation versus frequency at a lag of 50 cm for EERL.

Figure 8-17: The spatial autocorrelation at 1625 MHz versus distance lag in Commons, EERL, and Farmhouse. The abscissa distance is the indicated number times 0.05 m.

Table 8-15: Median decorrelation distances at each building location.

Building	Median Decorrelation Distance (m)
Commons	1.1
EERL	0.5
Farmhouse	0.85
House	0.55
MER	0.7
Motel	0.55
Overall Average	0.71

8.4.5 Relative Signal Loss Versus Frequency

An example of the trend of the relative signal loss versus frequency over the interval 500 MHz to 3000 MHz at a fixed location in the Farmhouse is given in Figure 8-18. The trend slope of -0.006 dB/MHz is shown to exist over the frequency interval between 500 MHz and 3000 MHz. The trend slope should not be confused with the smaller scale frequency slope FS given by (8-6) which characterizes the change of signal level per MHz of bandwidth. The indicated trend slope in Figure 8-18 implies that over the frequency interval of 2500 MHz (500 MHz to 3000 MHz) the attenuation trend is such that the signal level is diminished by approximately 15 dB. An overall summary of the trend slope statistics is given in Figure 8-19 where all positions within each building are considered. We note that the Farmhouse represents an extreme case for the trend loss as all the other sites have significantly smaller values.

Figure 8-18: Dependence of relative signal level with frequency in the Farmhouse. The trend slope is shown to be -0.006 dB per MHz.

Figure 8-19: Summary of trend slope statistics in the six building.

The autocorrelation of frequency variability reveals information regarding systems which attempt to mitigate fading employing power control on the uplink (mobile to satellite) at one frequency using knowledge of the fade on the downlink (satellite to mobile) at another frequency. An example of the autocorrelation at a given receiver location is plotted for each building in Figure 8-20. At 500 MHz, the lag of 10, for example, implies a bandwidth separation of 10 x 0.002 x 500 MHz = 10 MHz. At 3000 MHz, the lag of 10 implies a bandwidth separation of 60 MHz. The lag of 10 gives an autocorrelation of approximately 0.6 for the Farmhouse and 0.1 for the motel (extremes). Table 8-16 summarizes the median decorrelation bandwidths (decorrelation = e ^-1= 0.37) when all positions are accounted for within each of the buildings. The overall average of the median decorrelation is 2.4% of the frequency for the case in which the farmhouse is not included and 5.4% when the farmhouse is included.

Figure 8-20: The frequency autocorrelation in each building at a fixed location for lags from 0 to 20% of the frequency. The lag of 10, for example, implies 10 x 0.002 x f (MHz).

Table 8-16: Summary of median decorrelation bandwidths when all positions within each building are considered.

Building	Median Decorrelation Bandwidth (% frequency)
Commons	2.0
EERL	3.6
Farmhouse	20.2
House	1.6
MER	1.2
Motel	3.8
Overall Average without Farmhouse	2.4
Overall Average with Farmhouse	5.4

Home

Forward