MoPro Chapter 8.2

8.2 Satellite Radio Reception Inside Buildings from 700 MHz to 1800 MHz

This section deals with the results of Vogel and Torrence [1993] who arrived at propagation results for measurements inside six buildings comprised of brick, corrugated sheet-metal, wood frame, mobile-home, and reinforced concrete-wall constructions. Their investigation emphasized geostationary satellite transmissions associated with direct broadcasting systems since their receiving antenna had a relatively narrow beamwidth (90°) and was pointed along the line-of-sight to the transmitter. Such an antenna experiences less multipath fading than one that has an azimuthally omni-directional gain characteristic.

8.2.1 Experimental Features

Swept CW signals from 700 MHz to 1800 MHz were radiated from an antenna located on an 18 m tower attached to a van outside these buildings. The part of the receiver system inside the buildings was comprised of the above described antenna on a linear positioner located approximately 1.5 m above the ground and pointed towards the transmitting antenna. Measurements were made for the case in which the receiver antenna was sequentially moved in 5 cm steps (for a total of 80 cm) along any of the three orthogonal directions. The receiver had a resolution bandwidth between 10 KHz and 1 MHz, a carrier to noise ratio of 45 dB, and an overall measurement accuracy of better than 0.5 dB. A summary of the experimental parameters is given in Table 8-1, and a succinct description of each of the site locations (labeled Site 1 through Site 6) is given in Table 8-2.

Table 8-1: Pertinent measurement system parameters of Vogel and Torrence [1993].

Characteristics	Value
Frequency
Coverage, Df	700 MHz to 1800 MHz in 1 s
Span	0 Hz to 1100 MHz
Resolution	10 KHz to 1 MHz
Amplitude
Transmitted Power	10 mw
Range	45 dB
Resolution	0.2 dB
C/N Ratio	64 to 45 dB over Df
Error	< 0.5 dB
Transmitter and Receiver Antennas
Type	Cavity Backed Spiral
Polarization	Right-hand Circular
Beamwidth	90° (3 dB)
Gain	-2.5 to 4.5 dB over Df
Elevation Angle (Receiving)	12° to 48°

Table 8-2: Site descriptions of indoor propagation measurements of Vogel and Torrence [1993].

Site #	Description	Elevation Angle
1	A corner office (6 X 7 m) with two large windows in a single-story building. Walls are of concrete-block masonry with the interior covered with plasterboard. The ceiling is comprised of acoustic tiles suspended at a height of 3 m from metal hangers. A double-glazed optically reflective window is located in the wall toward the transmitter. Roof is flat and consists of concrete panels supported by steel beams. Room is furnished with wooden office furniture. Also referred to as the EERL Building	27.5°
2	A small room in another building (3 X 4 m) with two windows and construction similar to Site 1. Room furnished with metal filing cabinets.	18°
3	A 5 X 5 m corner foyer in another building where a large reflective glass door encompasses half of one outside wall. The external walls are of concrete wall construction, and internal walls have metal frames covered with plasterboard. Also referred to as the Commons.	16°
4	A 3 X 6 m shack with corrugated sheet-metal walls and roof on the outside and plywood on the inside. It has one window on each of the two narrow sides and a metal-covered door centered between two windows on one of the wide sides.	25°
5	An 1870 vintage restored and furnished two-story house with wood siding. The walls are filled with rock wool and covered with plasterboard on the interior and wood siding on the exterior. The gabled roof is covered with wood shingles. Measurements were made in two rooms on the ground floor and one room on the second level. Also referred to as the Farmhouse.	25°
6	A 12 X 2.4 m empty mobile trailer home with sheet-metal exterior and aluminum frame windows with metal screens.	45°

8.2.2 Multipath Interference During Frequency and Position Sweep

An example of maximum and minimum multipath interference (of relative signal level) experienced inside Site 2 is given in Figure 8-1 for a composite frequency sweep and vertical position scan near a window. The set of curves labeled maximum and minimum were derived by executing a frequency sweep between 700 MHz to 1800 MHz for each antenna position and culling out the maximum and minimum signal levels (upper and lower traces) at each frequency. The center thick curve is an example of the signal variability when the antenna position was fixed at an arbitrary position over the frequency interval. The large variability of the signal due to receiver antenna position and frequency changes demonstrates that multipath effects may be significant over the indicated position and frequency intervals. For example, the minimum trace shows signal levels which vary between -5 dB to smaller than -30 dB, the maximum trace shows signal levels which vary between +2 dB and -9 dB, and the fixed antenna position trace shows a variability between 0 to -30 dB.

Figure 8-1: Maximum and minimum relative signal levels (thin lines) in a composite vertical scan of 80 cm and frequency sweep over the indicated frequency interval for Site 2. Thick curve corresponds to fixed receiver antenna versus frequency example Vogel and Torrence, [1993].

8.2.3 Time Delay Distributions

Through the execution of a Fast Fourier Transform of the signal level over the frequency interval examined, estimates of maximum multipath time delays were derived. Cumulative distributions of these time delays are given in Figure 8-2 for three different site locations. It is clear from this figure that 90% of the power have delays smaller than 20 ns, 30 ns, and 80 ns for Sites 2, 4, and 3, respectively, and more than 99% of the power have associated delays smaller than 100 ns for Sites 2 and 4. These results were consistent with power loss measurements employing a series of bandwidths between 1 and 90 MHz over the frequency interval from 700 to 1800 MHz in that negligible bandwidth dependence was found in the loss statistics results.

Figure 8-2: Approximate cumulative distributions of time delay at three locations.

8.2.4 Cumulative Distributions of Signal Levels

Figure 8-3 gives the relative signal level versus frequency for a series of fixed probabilities ranging from 1% to 99%. These statistics reflect a data base of vertical and horizontal scans at several locations within Site 2 and averaged over bandwidths of 1, 2, 5, 9, 18, 45, and 90 MHz. We note, for example, that 1% of the signal levels exceeded 0 dB at approximately 0.75 GHz to -4 dB at 1.75 GHz, whereas at the probability of 50%, signal levels exceeded -5 to -14 dB at these respective frequencies. The general trend of the probability contour is such that the losses increase with increasing frequency.

Figure 8-3: Cumulative probabilities of relative signal level as a function of frequency for Site 2. The database reflects measurements taken at 16 positions for 80 cm vertical scans at several locations and seven bandwidths

The cumulative distributions of the relative dB signal losses were found to follow a Gaussian distribution for the cases in which the loss values were binned for the composite of values. The composite of values were obtained at the different frequencies and at the different locations within a particular building (from 8 to 20 locations were considered within a given building). In Table 8-3 are given the mean and standard deviations associated with the best-fit Gaussian distributions for each of the sites (building locations) at defined "average" and "best" cases. The "average" case was obtained by taking the linear average loss over a series of bandwidths ranging from 1 MHz to 90 MHz (1, 2, 5, 9, 15, 45, and 90 MHz) within each 100 MHz bin interval over the frequency span 700 MHz to 1800 MHz. In determining the "average" the results were also combined for each of 16 antenna position (5 cm displacement per position) and at each location within a given site (eight to twenty locations per site). The "best" case corresponds to statistics associated with that location within the position scan that resulted in the smallest losses. Site 6 showed the largest mean loss of approximately -25 dB as this building was a mobile trailer with a sheet metal exterior and aluminum frame windows having metal screens. At all other locations, the measurements of Vogel and Torrence were obtained in a less severe attenuation environment. Neither Ricean nor Rayleigh distributions provided a better fit than the Gaussian distribution. Also given in Table 8-3 are the overall averages (of the averages) including the worst site (Site 6) and excluding it. It is noted, for example, that the average for all sites is 12.7 dB with a standard deviation of 5.5 dB. When Site 6 is excluded, the overall average is 10.2 dB with a standard deviation of 5.8 dB.

Table 8-3: Mean and standard deviation of best fit Gaussian distributions for frequency and spatial averaging.

	Average Case		Best Case
Site Location	Mean (dB)	STD (dB)	Mean (dB)	STD (dB)
1	-7.9	5.5	-4.2	4.2
2	-9.1	4.4	-5.4	4.2
3	-15.4	8.4	-9.7	6.7
4	-9.7	6.3	-5.2	4.9
5	-9.0	4.5	-5.4	3.7
6	-24.9	3.8	-19.8	3.4
Average (Sites 1 to 6)	12.7	5.5	8.3	4.5
Average (Sites 1 to 5)	10.2	5.8	6.0	4.7

Table 8-4 lists the median of the relative power levels at all the site locations for the "average and "best" cases (defined previously) over the frequency interval 750 MHz to 1750 MHz. The smallest and largest median power levels are given over this frequency for each of the cases as well as the overall averages "with" and "without" the worst case Site 6 (last two rows). It is noted that Sites 3 and 6 had the most severe fading. Site 3 experienced more than 15 dB fading when the line-of sight path passed through a reflective glass door. The overall averages of the largest fade medians are less than -14.8 dB and -13 dB with and without Site 6.

Table 8-4: Relative median power levels over frequency interval 750 MHz to 1750 MHz.

Site #	Average Case (dB)		Best Case (dB)
Site #	Smallest	Largest	Smallest	Largest
1	-5	-11	-2	-6
2	-5	-14	-2	-5
3	-17	-18	-12	-13
4	-9	-11	-5	-6
5	-5	-11	-3	-5
6	-20	< -24	-16	-22
Average (1-6)	-10.2	<-14.8	-6.7	-9.5
Average (1-5)	-8.2	-13.0	-4.8	-7.0

8.2.5 Frequency Dependence of Probability

The dependence of exceedance probability on frequency (f) and relative signal loss (SL) at frequencies from 750 MHz to 1750 MHz has been found to follow the least square relation

(8-1)

where

(8-2)

(8-3)

and where a₀, a₁, b₀, b₁ are tabulated in Table 8-5 for the various site locations. Also tabulated in Table 8-5 is the RMS error of P in percent. Example distributions are given for Site 2 at 750 MHz, 1.25 GHz and 1.75 GHz in Figure 8-4.

Table 8-5: Best fit coefficients for the model cumulative distributions (8-1).

Site	a₀	a₁	b₀	b₁	RMS Error in P (%)
1	66.2	5.3	16.0	-0.54	4.0
2	74.1	7.0	20.3	-1.2	8.4
3	105.7	3.1	-7.8	-0.34	2.8
4	97.8	6.2	-3.8	-1.5	3.4
5	67.2	5.2	24.9	-0.1	6.1

Figure 8-4: Example cumulative distributions of relative signal loss for Site 2 based on model described by (8-1) through (8-3).

8.2.6 Space Diversity Considerations

Vogel and Torrence measured the position spacing between the minimum (troughs) and maximum (crests) signal levels for all buildings and found the median value to be between 35 cm and 45 cm and to be independent of frequency. An example of power level versus spacing at a series of six frequencies spaced 5 MHz apart is given in Figure 8-5. It is noted that the relative power loss varies approximately between 0 dB and -30 dB from crest to trough. Taking averages over frequency of the distances between signal minima and maxima, this distance was found to be Gaussian distributed with a mean near 41 cm with a standard deviation of 17 cm. No significant differences were observed for movement along the vertical or horizontal directions. The above results suggest optimal separation distances of antennas when space diversity concepts are employed.

Figure 8-5: Relative signal level versus vertical measurement distance at six frequencies spaced approximately 5 MHz apart. (Site 2).

8.2.7 Bandwidth Distortion Considerations

At relative signal levels down to approximately -15 dB, the multipath delays tended to be less than 100 ns consistent with the results in Figure 8-2. We may imply from this result that contemplated systems with bandwidths of up to about 1 MHz should not be adversely affected by signal distortion. Vogel and Torrence executed an investigation to establish the expected levels of signal distortion one might encounter by examining the losses at various bandwidths ranging from 2 MHz to 90 MHz. The signal distortion was described in terms of the standard deviation of the deviations relative to the mean signal loss at each bandwidth. Since the standard deviation was found to be frequency insensitive, the results were combined over each frequency interval considered (e.g., 700 to 800 MHz, 800 to 900 MHz, …, 1700 to 1800 MHz), and are shown plotted in Figure 8-6 for the "average" and "best" cases for Site 4. It is noted that at a 2 MHz bandwidth the standard deviation is approximately 0.5 dB, whereas the deviations become considerably larger at 90 MHz, ranging from 2.5 to 3.5 dB.

Figure 8-6: Standard deviation versus bandwidth for Site 4 for the "average" and "best" cases.

The standard deviation was found to follow the regression relations for each site as given by:

Average Case

,
(8-4)

Best Case

,
(8-5)

where a, b, g, are tabulated in Table 8-6 for the various site locations, STD is the standard deviation given in dB, and BW is the bandwidth given in MHz (2 to 90 MHz).

Table 8-6: Parameter values for standard deviation versus bandwidth as given by (8-4) and (8-5).

Site	a	b	g
1	1.6	0.12	0.61
2	1.6	0.03	0.56
3	2.6	2.82	-0.31
4	1.7	0.42	0.42
5	1.4	-0.02	0.55
6	2.4	1.9	0.31

Home

Forward