Critical ReviewsTM in Biomedical Engineering
Investigation of Brain Potentials
in Sleeping Humans Exposed
to the Electromagnetic Field
of Mobile Phones"
N. N. Lebedeva1, A. V. Sulimov2, O. P. Sulimova3, T. I. Korotkovskaya4,
and T. Gailus5
1Doctor of Biological Sciences, Principal Scientist at the Institute of Higher Nerve Activity
and Neurophysiology, Russian Academy of Sciences. E-mail: N.Leb@relcom.ru;
2Candidate of Medical Science, Senior Researcher at the Institute of Higher Nerve Activity
and Neurophysiology, Russian Academy of Sciences.; 3Researcher at the Institute of
Higher Nerve Activity and Neurophysiology, Russian Academy of Sciences; 4Deputy
Director of the “Extremely High Frequency” Medical and Technical Association; 5Biological
Department Head at the Deutsche Telekom AG (Germany).
ABSTRACT: An investigation was made of 8-hour EEG tracings of sleeping humans exposed to
the electromagnetic field of a GSM-standard mobile phone. To analyze the EEG-patterns, manual
scoring, nonlinear dynamics, and spectral analysis were employed. It was found that, when human
beings were exposed to the electromagnetic field of a cellular phone, their cerebral cortex
biopotentials revealed an increase in the α-range power density as compared to the placebo
experiment. It was also found that the dimension of EEG correlation dynamics and the relation of
sleep stages changed under the influence of the electromagnetic field of a mobile phone.
Normal sleep is made up of cyclic variations in the functional state of the
brain. The most salient and invariable feature of sleep consists in that the brain
shows a decrease in the cerebral cortex activity on the scale “sleep−wakefulness”.
The functional state of the brain is usually studied by means of an
electroencephalogram (EEG). An analysis of electroencephalograms made earlier
demonstrated that sleep is an inhomogeneous process. It has a complex structure
and a certain sequence of alternating patterns. There are many classifications of
sleep due to numerous approaches to the description of EEG patterns.
*Originally published in Russian in “Biomeditsinskaya Radioelektronika”, No. 7, 1999
© 2001 by Begell House, Inc.
One of the first classifications made it possible to analyze EEG patterns in
sleeping subjects . However, the application of this classification can only
show how deep is the subject’s sleep. In other words, this classification is capable
of revealing only two stages of sleep, such as drowsiness and deep (slow-wave)
sleep. The discovery of the rapid-eye-movement (REM) stage of sleep allowed
the development of a new classification . According to this classification, sleep
was divided into orthodox sleep, which does not involve rapid eye movements,
and into paradoxical, or REM, sleep.
The most complete classification of polysomnographic sleep records is given
in . It uses electroencephalograms, electromyograms, electrooculograms, and
some other techniques. This approach made it possible to recognize five
alternating stages of sleep with certain EEG correlates for every stage.
The first stage is marked by a relatively low amplitude and mixed
frequencies with a pronounced activity at 2 to 7 Hz. The duration of this stage is
relatively short and it ranges from 1 to 7 min. This stage occurs mostly when a
person starts to fall asleep or between other stages of sleep. The second stage is
recognized by a relatively low amplitude, “sleep spindles”, and K-complexes
arising either due to responses to unexpected stimuli or in the absence of any
differentiable stimuli. This stage constitutes approximately a half of the total
sleeping time. The third stage is characterized by slow high-amplitude waves,
which account for 20 to 50% of the whole analyzed period. The fourth stage
shows high-amplitude waves with a frequency of 2 Hz. These waves make up
more than 50% of the whole analyzed period. The fifth stage (or REM-stage) is
distinguished by a relatively low amplitude and by mixed EEG frequencies
periodically interrupted by episodes of REM sleep.
Presently, researchers make use of all the enumerated classifications of sleep
stages depending on the solved problem.
In healthy adults, night sleep is made up of 4 to 6 sleep cycles that include
all five stages. Each cycle begins with a period of slow-wave sleep. On the
average, each cycle lasts for 90 min. However, the first cycle lasts longer than the
However, night sleep of healthy people can be affected by many kinds of
things in the environment. For example, sleep is dependent on personality and it
was found to be different in morning types and evening types [4−7]. Apart from
that, sleep depends on the time when a person goes to bed and on the time during
which he or she stayed awake before going to bed [8, 9]. In addition, sleep
depends on the surroundings , age , subject’s geomagnetic orientation
[12, 13], and other conditions.
In view of a constantly aggravating electromagnetic pollution of the
environment, it seems essential to study the effect of various electromagnetic
devices and equipment on the activity of the central nervous system of human
beings , for example, during their sleep .
Our investigation was conducted on 20 volunteers (20 men aged from 20
to 28). We made two runs of experiments, in which the volunteers were exposed
to simulated (sham) and actual electromagnetic fields of a mobile phone in
random order. Two-channel EEG tracings were recorded during an 8-hour
laboratory sleep. The EEG tracings were recorded in the CZ and PZ leads, with the
reference electrode being placed in the FZ lead.
As soon as electroencephalograms were recorded, sleep stages were scored
manually and then their duration and alternation were analyzed. After that,
nonlinear dynamics methods were employed to calculate correlation dimension
and then FFT-based spectral analysis was used. We left out of account periods
affected by muscular motion or winking artifacts. For normal EEG periods, we
calculated the absolute spectral power in the δ-, θ-, α-, and β-ranges and the
correlation dimension D2 for each lead. The secondary statistical treatment was
carried out with the aid of variance analysis (ANOVA).
The spectral analysis of EEG patterns obtained from sleeping humans
demonstrated that reliable variations in the spectral power indices were observed
only in the α-range for the P lead (Figure 1, Table 1). An increased spectral
power in the α-range can be explained by the fact that this range is most reactive
to the influence of electromagnetic fields [16, 17]. A similar tendency of changes
was also observed in the δ- and θ-ranges. However, the β-range did not exhibit
significant variations or tendencies. It was also found that volunteers exposed to
the actual electromagnetic field of a mobile phone showed a significant decrease
(p < 0.05) in the two-channel correlation dimension D2 during an 8-hour sleep as
compared to the placebo experiment with a simulated electromagnetic field
The mean values of correlation dimension indices D2 for EEG tracings
recorded during an 8-hour sleep from the CZ and PZ leads are presented in
Figure 3 and in Table 2. Although the decrease in D2 was significant (p < 0.05)
for both leads, it was more pronounced for the CZ lead.
Table 1. Mean values of the spectral power indices in the α-range
for simulated and actual electromagnetic fields of a mobile phone
FIGURE 1. Mean values of the EEG spectral power density in the
α-range obtained from the CZ and PZ leads during an 8-hour sleep for
simulated and actual electromagnetic fields of a mobile phone
(pPz < 0.05).
FIGURE 2. Mean values of the two-channel correlation dimension D2
during an 8-hour sleep.
FIGURE 3. Mean values of the correlation dimension indices
D2 obtained from the CZ and PZ leads during an 8-hour sleep.
Table 2. Mean values of the correlation dimension indices D2 for simulated
and actual electromagnetic fields of a mobile phone
The dynamics of the correlation dimension indices D2 during an 8-hour sleep
for simulated and actual electromagnetic fields of a mobile phone is shown in
Figure 4. Similar patterns were observed for every lead.
Manual scoring demonstrated that, when volunteers were exposed to
simulated and actual electromagnetic fields of a mobile phone, they exhibited
insignificant differences in sleep stages. However, one can see that the percentage
of slow-wave sleep during the whole 8-hour sleep revealed a tendency to decrease
in volunteers exposed to the actual electromagnetic field (Figure 5, Table 3).
Table 3. Percentage of the slow-wave sleep for simulated and
actual electromagnetic fields of a mobile phone
FIGURE 4. Dynamics of the mean correlation dimension indices D2 obtained
from the CZ and PZ leads during an 8-hour sleep for simulated and actual
electromagnetic fields of a mobile phone (p < 0.05).
It is known that the structure of night sleep includes deep synchronized
slow-wave sleep (SSWS), which takes a large part of the first half of a sleep
cycle. SSWS percentage then decreases and EEG tracings reveal
desynchronization processes. These processes are typical of nonsynchronized
slow-wave sleep (NSSWS), which includes the first, the second, and the fifth (or
REM) stages .
It is also known  that EEG correlation dimension indices D2 of SSWS
are significantly lower than those of NSSWS. This is associated with the fact that
processes occurring in the brain at the third and fourth stages take a simplified
course. During the second half of night sleep, NSSWS becomes dominating and
the cerebral cortex’s bioelectrical processes get more complicated. This gives rise
to increased values of D2. It is this dynamics of D2 that was seen during sleep in
human beings exposed to a simulated electromagnetic field (Figure 4).
FIGURE 5. Percentage of the slow-wave sleep (the third and
the fourth stages) with respect to the whole 8-hour sleep structure for
simulated and actual electromagnetic fields of a mobile phone.
However, when volunteers were exposed to the actual electromagnetic field
of a mobile phone, they exhibited a different EEG pattern. Although during the
first half of their sleep the dynamics of D2 was virtually identical to that of the
control group, it did not show the expected increase in D2 during the second half
of their sleep. Moreover, it remained invariable to the end of our experiment.
Thus, the dynamics of D2 provided evidence for a changed sleep structure, which
was related to the NSSWS depression. Similar results were reported in , in
which healthy people were exposed to the electromagnetic field of a cellular
phone and they also revealed a suppressed REM-stage (which is one of NSSWS
stages). Apart from that, one can see from Figure 5 that the slow-wave sleep cycle
(the third and fourth stages) revealed a tendency to decrease under the influence
of the actual electromagnetic field, although this effect was not significant.
Thus, taking into account the results obtained in  and results presented in
Figures 4 and 5, one can postulate that a prolonged exposure to the
electromagnetic field of a mobile phone brings about a pronounced sleep structure
transformation. In this case, sleep is largely made up of the first and second sleep
stages, which involve no dreaming. Bearing in mind that the first stage is rather
short and that it mainly accompanies the REM-stage, one can conclude that, when
human beings were exposed to the actual electromagnetic field of a mobile phone,
a great part of their sleep consisted of the second sleep stage, which is typical of
aged people .
It is known that the fundamental biological role of sleep in live organisms is
directed to providing adaptive regulation. On the one hand, sleep reduces
functional loads on various bodily systems, such as the nervous system, the
cardiovascular system, and the muscular system, with this reduction taking place
during the third and the fourth sleep stages. On the other hand, the adaptive role
of dreaming, which occurs during the REM-stage, is also of great importance
because dreaming helps people to process stored information and to reduce
Hence, the electromagnetic field of a mobile phone affects the sleep
structure and reduces slow-wave and REM-stage sleep percentage, which is able
to decrease the adaptive reactions of human beings and to impair their state of
health as a result of this.
1. Loomis, A. L., Harvey, E. N., and Hobart, G. A., Cerebral states during
sleep as studied by human brain potentials. J. Exper. Phychol., 1937.
2. Dement, W. and Kleitman, N., Cyclic variations in EEG during sleep and
their relation to eye movements, body motility, and dreaming.
Electroencephal. Clin. Neurophysiol., No. 9, 1957.
3. Rechtshaffen, A. and Kales, A., A manual of Standardized Terminology
Techniques and Scoring System for Sleep Stages of Human Subjects.
National Institutes of Health. Publication No. 204, Washington, 1968.
4. Akerstedt, T., Kecklund, G., and Knutsson, A., Manifest sleepiness and
spectral content of the EEG during shift work. Sleep, Vol. 14, No. 3,
pp. 221−225, 1991.
5. Lancel, M. and Kerkhof, G. A., Sleep structure and EEG power density in
morning types and evening types during a simulated day and night shift.
Physiol. Behav., Vol. 49, No. 6, 1991.
6. Kerkhof, G. A., Differences between morning types and evening types in the
dynamics of EEG slow-wave activity during night sleep. Electroencephal.
Clin. Neurophysiol. Vol. 78, No. 3, 1991.
7. Armitage, R. and Roffwarg, H. P., Distribution of period-analyzed delta
activity during sleep. Sleep, Vol. 15, No. 6, 1992.
8. Aeschbach, D., Dijk, D. J., and Borbely, A. A., Dynamics of EEG spindle
frequency activity during extended sleep in humans: relationship to slow-
wave activity and time of day. Brain-Res., Vol. 14, 1997.
9. Gillberg M., and Akerstedt, T., The dynamics of the first sleep cycle. Sleep,
Vol. 14, 1991.
10. McCall, W. V., Erwin, C. W., Edinger J. D., et al., Ambulatory
polysomnography: technical aspects and normative values. J. Clin.
Neurophysiol., Vol. 9, No. 1, 1992.
11. Dijk, D. J., Beersma, D. G. and van den Hoofdakker ,R. H., All night
spectral analysis of EEG sleep in young adult and middle-aged male subjects.
Neurobiol. Aging. Vol. 10, No. 6, 1989.
12. Ruhenstroth-Bauer, G., Ruther, E., and Reinertshofer, T., Dependence of
a sleeping parameter from the N-S or E-W sleep direction. Z. Naturforsh [C],
Vol. 42, No. 9−10, 1987.
13. Ruhenstroth-Bauer, G., Gunther, W., Hantschk, I., et al., Influence of the
Earth’s magnetic field on resting and activated EEG mapping in normal
subjects. Int. J. Neurosci., Vol. 73, No. 3−4, 1993.
14. Lebedeva, N. N., Reactions of the central nervous system to electromagnetic
fields with various biotropic parameters. Biomeditsinskaya Radioelektronika,
No. 1, 1998 (in Russian).
15. Mann, K. and Roschke, J., REM suppression induced by digital mobile
radio telephones. Wein. Med. Wochenschr., Vol. 146, No. 13−14, 1996.
16. Lebedeva, N. N., Sulimov, A. V., Sulimova, O. V., and Kotrovskaya, T.
I., Effect of the electromagnetic field of a mobile phone on the bioelectrical
activity of the human brain. Biomeditsinskaya Radioelektronika, No. 4, 1998
17. Sulimova, O. P., Electrical and Psychophysiological Reactions of the Human
Being on the Peripheral Influence of Low-intensity EHF Electromagnetic
Radiation. Abstract of Thesis for the Candidate of Biological Science.
Simferopol State University, 1992 (in Russian).
18. Fell, J. and Roschke, J., Nonlinear dynamical aspects of human sleep EEG.
Int. J. Neurosci., Vol. 76, No. 1−2, 1994.