Report on: Measurement of Cellular Base-station Emissions Using a Newly Developed RF Field Mapping System*

Discussion

Signal Envelope Recordings

From Fig. 4, it can be seen that the logarithm of the envelope has an average component (given as 1.36 V in the oscillograph) along with fast changing and slow changing variations. Because the logarithm function tends to compress the variations in amplitude, they appear to be quite small but are in fact significant. It should be remembered that the envelope shown here is the composite of all the carriers transmitted by the tower in the general direction of the receiver. If the received signal consists of multiple modulated carriers, then a component of the fast changing and slow changing variations may be due to alternate addition and subtraction of the instantaneous carrier amplitudes. This component is expected to give random variations in the envelope.

Another component of the fast and slow envelope variations is due to so-called "multipath" propagation effects. This occurs because the power density received by an antenna is actually the sum of a number of rays of electromagnetic energy. They all originate from the base station but travel over different propagation paths (i.e. bouncing off buildings or passing through trees) to get to the receiver. These different paths are in themselves randomly fluctuating (because of moving vehicles, people, trees, etc.), hence when all the rays arrive at the receiver, their individual power densities sometimes add together or sometimes subtract from each other (also known as destructive or constructive interference) in a random fashion. Thus the component of the fast changing and slow changing envelope variations due to multi-path, is random also.

Given that the variations in the envelope are random, they can be considered as electrical "noise" superimposed on the average value. From inspection of Fig 4., it can be seen that the period of the slowest variation appears to be around 40 - 60 µs, giving a lowest frequency in the 10's of kHz (kilohertz).

It has been suggested by some researchers [RSC 1999, Litovitz et. al. 1993] that digitally modulated (of which PCS is a type) radiofrequency signals are more "biologically active" because of their pulsatile nature and because some of the spectral (frequency) components are quite low in frequency and are close to endogenously generated frequencies (i.e. frequencies generated by our own bodies as a result of brain waves, heart rhythms, etc.). While this may be characteristic of transmissions produced by some types of PCS cellular handsets, it clearly is not applicable to signals from a PCS base station.

Similar oscillographs were observed for Analog-band signals, which certainly consist of a large number of modulated carriers. Again the envelope variations appear noise-like with frequencies very much higher than endogenously-generated ones.

Measured Exposure Levels and Safety Code 6 Limits

In Canada, Safety Code 6 specifies a general public MEL of 5.9 W/m ² for the Analog band (824-894 MHz) and 10 W/m ²for the PCS band (1850-1975 MHz). These values of power density produce a worst-case Specific Absorption Rate (SAR), with respect to all human body sizes, of 0.08 W/kg. To put this number into perspective, the total radiofrequency power absorbed by a 180 lb. (82 kg) person would be 6.6 watts when exposed to SC6 limit-level power densities. This is compared to a resting metabolic rate of 98.4 watts [Polk & Postow, 1996] for the same 82 kg person. Considering the worst-case power densities measured using the GLOBE system of 3 thousand times below SC6, this translates to a worst-case absorbed power of 2.2 milliwatts (a milliwatt is one thousandth of a watt) for the same 82 kg person. This amount of power is roughly comparable to the rate of solar energy falling on the heads of two pins at noon on a sunny day in July [Sliny & Wolbarsht, 1982, p 211]. (This comparison was made only to give an idea of the quantity of energy that is absorbed. It is acknowledged that radiofrequency and solar energy are absorbed differently in tissues.)

In terms of the distribution of power density in the plots, it can be seen that the level does not fall-off regularly with distance from the tower. In fact, adjacent dots sometimes differ by more than a single colour step or power density range. This indicates that two closely spaced locations may have power densities differing by a factor of 10 or more.This may be due to the radiation patterns of the antennas on the tower itself or due to blockage of the direct line-of-sight rays by buildings, trees or other structures. Other contributing factors are multi-path effects caused by static structures and the ground and cellular call traffic which will be discussed later in the section on uncertainties.

Some conclusions that can be drawn from observation of the maps are:

1. measured power densities are generally stronger as you get nearer a base station but vary in a haphazard or chaotic fashion. They do not follow the simple "inverse squared-distance law" as sometimes pointed out by experts.
2. two closely spaced points can have significantly different power densities.
3. accurate mathematical prediction of the power density would involve elaborate description of the terrain and structures and is probably not cost-effective.
4. Generally, PCS-band power densities are lower in relation to the SC6 MELs than are Analog-band ones.

Uncertainties

The question of uncertainty usually arises whenever measurements are made. We may divide the uncertainties into two categories, one being the uncertainty of the exposure level (ie the magnitude of the power density) due to the random effects discussed previously and the other being the uncertainty in the measurement of the exposure level (instrumentation uncertainty). The difference can be illustrated by the hypothetical situation where, even if the power density can be measured by an instrument exactly, there will still be temporal and spatial variations in the power density at a location due to the random effects. When the errors caused by imperfect instrumentation are factored in, the overal uncertainty becomes larger.

As pointed out previously, a major source of exposure uncertainty is multi-path propagation effects and blockage due to structures and terrain. Quantitative assessment of this uncertainty component has not been carried out but it has been observed that closely spaced locations may differ in power density by a factor of ten times or higher.

Another source of exposure uncertainty for a specific cellular technology called AMPS (often referred to as analog cellular or FD MA) arises because the total power transmitted by the base station depends on the number of cell phone calls being processed. This causes the power density measured at a single location to vary in time continuously as callers make and disconnect calls. Spot measurements in the analog band over long time periods have indicated the variations to have a distribution resembling the well known "bell curve". The width of the bell curve is such that 95% of all the measurements lay within a factor of between ½ or 2 times the mean or average power density [Gajda et. al., 1998]. Since GLOBE measurements are made with relatively short averaging times, their values have an inherent random uncertainty of up to ½ or two times the indicated value (for the analog band only). This uncertainty component is non-existent for certain digital technologies, such as TDMA, which transmit a small number of carriers continuously.

The issue of instrumentation uncertainty is more easily quantifiable and is estimated to be of the order of ± 4 dB. This may be interpreted by saying that the instantaneous power density level may be 0.4 times lower or 2.5 times higher than what the instrument indicates. This magnitude of uncertainty may seem high but is typical for this type of measurement.

An important part of the instrumentation uncertainty is the reception coverage of the antenna. As mentioned previously, the reception pattern precludes receiving power densities arriving from directly overhead the system, as would be the case directly under the base station antennas. Also, power densities arriving from angles close to the horizon are also slightly degraded although not as severely. Thus the values of power density very close to and very distant from the base station are probably underestimated. While this is of little concern for areas distant from base stations where power densities are low, it is an important factor for locations say within 40 m of a base station. Steps are being taken in the design of the next generation of GLOBE equipment to address this deficiency.

The overall uncertainty of exposure levels indicated on the maps is difficult to estimate quantitatively especially in the Analog band since it has multiple components. For this case, we can estimate a lower bound by assuming it consists only of the instrumentation uncertainty and the call traffic uncertainty. The result would be a factor of between 0.2 or 5 times the indicated power density. Given that the highest levels measured are about 3 thousand times below SC6, factoring in the overall uncertainty still brings the exposure levels far below the code's MEL.

Smallest Power Density Needed to Operate a Cellular Handset

One aspect to consider when viewing the maps is the magnitude of the power densities in relation to those necessary for functioning of the cellular network. A power density of 1 million times below SC6 may seem impossibly small, however it is more than adequate to maintain communications with a cellular phone handset. An estimate of how little power density is needed by the handset to keep in touch with the base station can be made from basic communications principles. If the operating parameters of Table 2 are assumed for the handset, a minimum power density received by the handset, of approximately 30 billion times below SC6 is required

Table 2:

Assumed communications parameters for a PCS cellular telephone handset.
thermal noise spectral density (kT)*	-174 dBm/Hz
minimum signal-to-noise ratio to maintain link	20 dB
noise figure of handset receiver	3 dB
worst-case channel bandwidth	1 MHz
reception frequency	1900 MHz
antenna gain of handset	1 dB

to maintain a communications link with the base station.

* k is Boltzman's constant and T is the ambient temperature in degrees Kelvin

Of course this estimate is very crude and may be higher or lower by an order of magnitude; nevertheless it illustrates how small a power density is needed to establish a communications link with a cell-phone. In view of this, one may ask why the measured power densities shown in the maps are somewhat higher in relation. The answer is that the power densities presented are the sum of all the power densities or channels propagated by a particular base-station. More importantly, they represent levels measured outdoors in relatively open areas (usually streets). They must be high enough so that the energy-attenuating effects of trees, buildings and structural walls can be overcome in order to allow usage of mobile handsets indoors and in vehicles.