by W.A. Steer PhD
About...
GSM phone signal analysis
After coming by some new information, my curiosity was aroused with
regard to what exactly a digital GSM mobile phone transmits. This
document summarises what I found out, and actually measured.
Background
Contemporary 'digital' mobile phones in Europe operate using the GSM
(Global Systems Mobile) system. There are two frequency bands allocated
to GSM mobile phones, one at 900MHz, and one at 1800MHz. GSM uses a
combination of frequency division multiple access (FDMA) and time
division multiple access (TDMA). What this means in reality is that
within each band there are a hundred or so available carrier
frequencies on 200kHz spacing (the FDMA bit), and each carrier is
broken up into time-slots so as to support 8 separate conversations
(the TDMA bit). Correspondingly, the handset transmission is pulsed
with a duty cycle of 1:8; and the average power is one eighth of
the peak power. Once a call is in progress, the phones are designed to
reduce the radiofrequency (RF) output power to the minimum required for
reliable communication - under optimum conditions, the power can be set
as low as 20mW (one hundredth of full power). Battery consumption and
radiation output of the handset is further reduced by using
'discontinuous transmission' (DTX); the phone transmits very much less
data during pauses in the conversation.
| Base station transmit | Handset transmit | Peak handset power | used in the UK primarily by |
| GSM900 | 935-960MHz | 890-915MHz | 2 watts | BT Cellnet and Vodafone |
| GSM1800 (PCN) | 1805-1880MHz | 1710-1785MHz | 1 watt | Orange and One2one |
The basic handset transmission consists of carrier bursts of
0.577ms duration, repeating every 4.615ms, giving a
repetition rate of 216.7Hz. The data is encoded using a phase modulation
scheme known as 'Gaussian Minimum Shift Keying' (GMSK) and a symbol duration
of about 4µs, which doesn't affect the carrier amplitude. Owing to the
coding and control protocols, every 26th pulse is omitted during a
conversation, leading to a component in the output modulation at 8.33Hertz.
Experimental
I rigged up a simple circuit (crystal-set type non-tuned diode demodulator)
and recorded the signal amplitude using a standard PC soundcard line-in
socket.
The figure below represents the amplitude of the a GSM mobile phone (Vodafone)
transmission on reception of an incoming call.
The interpretation of the call progress is my own, based on limited
knowledge. Note how the transmitter power is reduced once the call has
been set up and the conversation begins. Note also the 'gaps' during the
silence in the conversation, where the transmitted signal returns to a
'holding' signal -- which looks superficially similar to the signal during
ringing.
The slow and smooth variation in amplitude is caused by moving the phone
relative to my receiver during the measurement, and is not significant!
This represents a section of the voice transmission, the timescale
of the whole plot being a little over 1 second.
Expanding the timescale further, the detail of the pulse structure
is revealed. The pulses come at
4.62 millisecond intervals (approx. 217Hz frequency), each lasting
0.57 milliseconds. This gives a mark:space ratio of 1:7, allowing
up to eight calls to be time-multiplexed (TDMA) onto the same carrier
frequency. Every 26th pulse is omitted, causing an 8.3Hz periodicity in
the signal.
The plot below shows some detail of the effect of the discontinuous
transmission.
Possible health issues
I'm trying to continue to 'sit on the fence', and not give an opinion
on the possible health-implications of mobile phones. The general
consensus of the scientific community and the relevant radiation-
protection bodies is that there is no significant evidence of a
health risk from mobile phones. Nevertheless, some people still claim
to suffer headaches and other symptoms which they blame on their
phone. Long term effects, of course, can only be observed after a
long time. The official line is basically that they are safe, but
some caution wouldn't go amiss. The emitted RF energy will be much
reduced if you use the phone in a good signal area (eg line of sight
to the base-station), whereas use in a poor signal area like inside
a lift (elevator) or a tunnel will result in the phone using a much
higher power.
You might want to consider the following points:
- With a 2 watt transmitter power the RF field strength, within a
couple of centimetres of the aerial is quite high (around 400V/meter). For
comparision, most electronic equipment is usually only guaranteed to operate
normally in fields up to 3V/meter. Use of a phone within a few metres
of electronic equipment can cause interference, and possible malfunction.
- The pulsed structure of the output has, rightly or wrongly, also
been a source of concern for human well-being. The component at 8Hz
is close to brainwave frequencies.
- Besides the magnetic component of the RF field, a mobile phone would be
expected to emit a weak, low frequency magnetic field. This will be
generated by the power wiring inside the phone as the 2watt transmitter
(around 1amp from the 2volt battery) is switched on and off. Although the
strength of this field is much less than the Earth's constant
magnetic field, recent studies have indicated that even very weak
switched magnetic fields are capable of affecting neurons in the brain,
and of aborting epileptic fits. Perhaps they can also cause other, less
desireable effects?
-
What is for sure is that RF energy to which the public are exposed from
base stations is typically less than one thousandth the strength of that
from holding a handset to the ear. Technically very much higher power
television transmitters have been operated on similar frequencies (to GSM900)
for many years (albeit with different modulation structure), and yet have
not caused such an outcry. With fewer TV transmitters, and not generally
sited in the heart of suburbia, they haven't generated 'not-in-my-back-yard'
sentiments in the same way. The visual impact of 'masts' is still
undesireable though!
Further reading
Created: February 2001
Last modified: 23 March 2002
Source: http://www.techmind.org/gsm/
©2001 William Andrew Steer
andrew@techmind.org