Wi-Fi and Health: Review of the Current Research

1Moulder, HPS, June-2013

Wi-Fi and Health: Review of the Current Research*

John Moulder, Ph.D.Medical College of Wisconsin

Milwaukee, WIKenneth R Foster, Ph.D.

Bioengineering, University of PennsylvaniaPhiladelphia, PA

*Health Physics (in press)


Disclaimer

• This presentation is based on a review that was funded by the Wi-Fi Alliance (Washington DC) and Mobile Manufacturers Forum (Brussels, Belgium).

• Neither organization had any role in the research for, or preparation of, the review.

• Neither organization had any knowledge of the contents, or conclusions, of the review prior to submission for publication.

• Neither organization had any input into this presentation.

• The opinions in this review are those of the present authors only.


What is Wi-Fi?

• Wi-Fi© is a trademarked name for wireless networking products that are certified to be compliant with IEEE 802.11 family of standards • Certification is by the Wi-Fi Alliance (an industry group).

• Wi-Fi devices contain low-powered radiofrequency (RF) transceivers that support wireless local area networks (WLANs). • Their most familiar (but not only) use is to provide access to the Internet by

laptop computers • But IEEE 802.11 protocols are used for other communications devices

including some electric utility meters.

• Initially developed as a wireless replacement for Ethernet cable to connect computers to LANs, IEEE 802.11 is now the basis of virtually all WLANs present in homes and offices.


What is Wi-Fi?

• Wi-Fi© is a trademarked name for wireless networking products that are certified to be compliant with IEEE 802.11 family of standards • Certification is by the Wi-Fi Alliance (an industry group).

• Wi-Fi devices contain low-powered radiofrequency (RF) transceivers that support wireless local area networks (WLANs). • Their most familiar (but not only) use is to provide access to the Internet by

laptop computers • But IEEE 802.11 protocols are used for other communications devices

including some electric utility meters.

• Initially developed as a wireless replacement for Ethernet cable to connect computers to LANs, IEEE 802.11 is now the basis of virtually all WLANs present in homes and offices. • John: “4 Wi-Fi devices in my own home, plus 10+ other transcievers”• Ken: “6 or 8, plus a similar number of Bluetooth and ZigBee devices”


RF Exposure from a WLAN

• Under all but one* plausible scenario, the RF exposure from a WLAN is far below major international limits as well as national limits• In the U.S., FCC; in Canada, Safety Code 6.• This applies to both the client card in the laptop and the access point

• Health agencies have not expressed concern about possible health hazards from such exposures (e.g., WHO 2006).

*Using a body-worn mobile phone as a hotspot (“tethering”)


So why spend months reviewing safety studies?

RF Exposure from a WLAN

• Under all but one* plausible scenario, the RF exposure from a WLAN is far below major international limits as well as national limits• In the U.S., FCC; in Canada, Safety Code 6.• This applies to both the client card in the laptop and the access point

• Health agencies have not expressed concern about possible health hazards from such exposures (e.g., WHO 2006)








Methods

• A systematic search of the literature (updated in May-2013)• General scientific databases (e.g., Medline, ISI Web of Science)• Specialized databases (e.g., EMBASE, the IEEE ICES group database)• FCC equipment authorization database

• What I will talk about• Engineering (e.g., frequency, modulation, peak power limits)• Exposure limits• Actual exposures• Exposures compared to other sources• Studies of biological effects of WiFi• Recommendations for future studies


Frequency and Modulation

• Not all WiFi devices are the same• Different versions of the IEEE 802.11 standards• Country-to-country variations in regulations• Peak power varies widely between devices



• Not all WiFi devices are the same• Different versions of the IEEE 802.11 standards• Country-to-country variations in regulations• Peak power varies widely between devices

• The differences are relatively minor in terms of potential health effects• But we are talking about a widely-variable class

of devices, not a single device



• Frequency• Most devices work in the unlicensed (ISM) band at 2400-2483.5 MHz

-2412-2462 MHz in US-This is on top of the standard MW oven frequency

• Also in this band-Bluetooth, ZigBee and some cordless phones- Industrial heating and medical appliances

• Some devices also work in ISM bands near 5000 MHz

• Transmission is in RF pulses up to ~10 msec long• Pulses are modulated by a number of different digital techniques

-DSSS, FHSS or OFDM

• Data rate• From 1-2 Mbit/sec in the original (legacy) version of IEEE 802.11

to 600 Mbit/sec in IEEE 802.11n


Peak Power Output

• Subject to national regulation (not part of the IEEE standard)• In the US this is the FCC (in 47CFR15.247)

-Max power is 1 watt for low-gain devices-Even lower for high-gain devices

• Canadian regs (SC6) are similar to those in the US• European Union regs are substantially stricter


Peak Power Output

• Subject to national regulation (not part of the IEEE standard)• In the US this is the FCC (in 47CFR15.247)

-Max power is 1 watt for low-gain devices-Even lower for high-gain devices

• Canadian regs are similar to those in the US• European Union regs are substantially stricter

• For comparison, the US regulation for MW oven leakage is 1 watt• But a MW oven has a full duty cycle when on, whereas WiFi devices have

a very low duty cycle.• So the comparison may be a bit misleading


0.001

0.003

0.01

0.03

0.1

0.3

1.0

0 100 200 300 400 500 600

Peak Output Power (W)

FCC Grants

Peak Power Output of FCC-Approved WiFi Devices(Foster, IEEE Microwave Mag, 2013)

2450 MHz5300, 5800 MHz


Peak Power Output

• Most device applications are for peak power output below 0.1 W

• The low power is probably a result of two factors• Need to conform to European as well as North American regulations• No need for higher power

-WiFi devices do not have adaptive power control (unlike cell phones)


Peak Power Output


• The low power is probably a result of two factors• Need to conform of European as well as North American regulations• No need for higher


• Warning ONE: this is the number of FCC grants• It may, or may not, reflect what is actually out there

-Some grants could have resulted in millions of devices sold


Peak Power Output


• The low power is probably a result of two factors• Need to conform of European as well as North American regulations• No need for higher


• Warning ONE: this is the number of FCC grants• It may or may not reflect what is actually out there

-Some grants could have resulted in millions of devices sold

• Warning TWO: This is just for WiFi, it does not include things like:• Bluetooth-enabled trackpads and key boards• Bluetooth or ZigBee devices for remote control of thermostats• Wireless intercoms, doorbells and thermostats


Incident Power Density

• FCC, Canadian and European Union regulations are based on power output

• But safety standards are based on the power absorbed by the body• Because health effects are presumed to depend on absorbed

power

• Specifically, safety standards are based on:• Specific absorption rate (SAR in W/kg)

-Fundamental, but hard to measure• Incident power density (W/m2)

-Related to SAR, but easier to measure


Exposure Limits

• ICNIRP guidelines have been adapted by most countries• US and Canadian guidelines are based on a similar IEEE standard

• For general public exposure (at 2450 MHz):• 10 W/m2 averaged over 30 minutes

• For general public exposure (at 5800 MHz):• 10 W/m2 averaged over 6 (ICNIRP) to 30 (FCC) minutes

• For antennas close to the body a more relevant limit is local SAR• 2 W/kg in any 10 g of tissue


Calculating Incident Power Density

• For a low-gain WiFi device operating at 0.1 W• 0.33 W/m2 at 20 cm• 0.013 W/m2 at 1 meter

• Compare this to the 10 W/m2 standard• At 20 cm we have a 30X margin

• Compare to the 100 W/m2 needed to get consistent bioeffects• At 20 cm we have a 300X margin


Real-World Exposures

• Because of the nature of the emissions (low duty cycle) specialized (and expensive) equipment is needed.• Most (all?) units on the consumer market can give misleading results

• A survey of laptops in diverse environments (Foster, 2007)• 0.000001-0.00001 W/m2 at 1 meter from a WiFi-enabled laptop• Duty cycles typically below 1%• Signal fall-off is inverse square

• Survey of WiFi access points (Lunca et al., 2012)• Peak of 0.054 W/m2 at 1 meter from a 2450 MHz access point

• The exposure standard is 10 W/m2


More Real-World Exposures(Peymann et al., 2011; Khalid et al., 2011)

• Access points and client cards in UK schools

• Mean peak power density• 0.005 W/m2 at 1 meter (for 2450 MHz)• 0.002 W/m2 at 1.5 meters (for 5800 MHz)

• Duty cycle in busy classroom• 0.02 to 0.91% for the laptops• 1-12% for the access points

• Peak localized SAR in torso of child at 34 cm• 0.000080 W/kg vs 2 W/kg safety standard• 25,000X safety margin


What About Exposures from Multiple Devices?(Malone & Malone, 2009; Fang and Malone, 2010)

• The public worries about being in a room with multiple WiFi users.

• But as the usage approaches the max capacity of the access point:• Total radiated power approaches that of a single node

-Because of collision-avoidance protocols


What About Exposures from Multiple Devices?(Malone & Malone, 2009; Fang and Malone, 2010)

• The public worries about being in a room with multiple WiFi users.

• But as the usage approaches the max capacity of the access point:• Total radiated power approaches that of a single node

-Because of collision-avoidance protocols

• One can contrive scenarios where the above limiting behavior fails• But total transmitted power will not exceed about 100 mW

-Similar to a single mobile phone-With most exposure sources at some distance from an individual


Power Density and WiFi

Proven biological effects

Safety standards

Theoretical peak (at 20 cm) from an 0.1 W device

Measured average 100 cm from access point

100 W/m2

10 W/m2

1 W/m2

0.1 W/m2

0.01 W/m2

0.001 W/m2

0.0001 W/m2

0.00001 W/m2

Measured average in UK school rooms

Measured 100 cm from a laptop

Power Density


What About SAR?(Findlay & Dimbylow, 2010)

• Power density may be easy to measure, • But health effects probably would be proportional to

absorbed dose (i.e., SAR)

• Worst case (WiFi antenna 3 cm from child’s face)• Peak is 0.82 W/kg (during an RF pulse)• With 1% duty cycle this means less than 0.01 W/kg• Compared to ICNIRP standard of 2.0 W/kg


Exposure Summary

• Peak power density from WiFi devices is well-characterized• Test data submitted to FCC• Independent direct measurements

• Power density (averaged over times specified by exposure standards) is a small fraction of the peak power density

• Thus:• RF exposure of an individual from a WLAN (under all but one

realistic conditions) is a tiny fraction of the safety limits (IEEE/FCC/ICNIRP).


The Exception: Tethering


The Exception: Tethering

• If the phone is body-worn and facing out:• Exposure standards for mobile phones may be exceeded

• Why?• Testing of the phone assumes that the antenna (typically on

the back of the phone) is not that close to the body.

• Solutions• Turn the phone around• Don’t wear the phone while using it as a tether• Put something between you and the phone

-Thickness needed: maybe 20 cm??


RF Exposures from Other Sources(Joseph 2010, 2013, Viel 2009 a,b)

Power Density

Cell p

hone

use

by o

ther

s

DECT

Cordle

ss p

hone

100 W/m2

10 W/m2

1 W/m2

0.1 W/m2

0.01 W/m2

0.001 W/m2


Safety standards

FM B

road

cast

TV B

road

cast

Cell p

hone

tower

s


WiFi Exposures Compared to Other Sources(Joseph 2010, 2013, Viel 2009 a,b)

Power Density

Cell p

hone

use

by o

ther

s

DECT

Cordle

ss p

hone

100 W/m2

10 W/m2

1 W/m2

0.1 W/m2

0.01 W/m2

0.001 W/m2


Safety standards

FM B

road

cast

TV B

road

cast

Cell p

hone

tower

s

WiF

i


Interference with Medical Devices

• Extensive studies in connection with cell phones (e.g., Carranza, 2011)• WiFi issues should be similar

• Rare instances of potentially harmful interference• Require handset very close to the device• Unlikely scenario for WiFi

• Solution• Require minimum distances

-As is usually specified by the manufacturer of the medical device


Biological Studies of WiFi

• Until the mobile phone controversy erupted in the early ’90’s, most RF biology studies focused on the ISM bands at 915 and 2450 MHz• More recently studies have focused on the frequencies and modulations used

by mobile phones.

• As a result there is a massive literature on biological effects at the frequencies used by WiFI• But few of these studies have used the specific waveforms used by WiFi.



• Until the mobile phone controversy erupted in the early ’90’s, most RF biology studies focused on the ISM bands at 915 and 2450 MHz• More recently studies have focused on the frequencies and modulations used

by mobile phones.

• As a result there is a massive literature on biological effects at the frequencies used by WiFI• But few of these studies have used the specific waveforms used by WiFi.

• There is no known biological , or biophysical, basis for expecting that WiFi signals would produce effects different from other low-level RF exposure.• But we still looked for WiFi-specific biological studies



• The initial literature search focused on studies that:• Were peer-reviewed• Had well-characterized exposure systems and dosimetry

• This yielded only 7 studies



• The initial literature search focused on studies that:• Were peer-reviewed• Had well-characterized exposure systems and dosimetry

• This yielded only 7 studies

• We then expanded the search to studies:• that were not peer-reviewed, or• which had apparent technical deficiencies

- typically, lacked well-defined exposure systems and dosimetry

• This yielded 6 additional studies


Poorly-Defined Exposure Systems

• Several studies used laptops or access points as exposure systems with no dosimetry• Exposure levels are unknown• Thermal effects are possible• Exposure and analysis blinding was usually absent



• We further reviewed the studies looking for:• Blinding of exposure and analysis (7/13)• Formal evaluation of possible thermal effects (4/13)• Presence of positive controls (3/13)• Presence of sham-irradiated controls (8/13)

• All of the above are generally considered to be required for such studies.


Peer-Reviewed Studies with Well-Defined Exposure Systems

• All of these studies were of WiFi exposure of fetal or neonatal rats• fertility, pregnancy outcome, fetal and neonatal development,

immune system and brain development• dose (SAR) from 0.08 to 4.0 W/kg

- all well above real-world WiFi exposures levels• reasonably long exposure durations

-1-2 hrs per day for 10-50 days.


Peer-Reviewed Studies with Well-Defined Exposure Systems

• All of these studies were of WiFi exposure of fetal or neonatal rats• fertility, pregnancy outcome, fetal and neonatal development,

immune system and brain development• dose (SAR) from 0.08 to 4.0 W/kg

- all well above real-world WiFi exposures levels• reasonably long exposure durations

-1-2 hrs per day for 10-50 days.

• None of the more than 150 endpoints assessed in these studies showed statistically-significant effects.• The lack of teratogenic effects is consistent with the vast majority of

previous studies of low-level RF exposure


Other Studies

• A wide range of endpoints• human performance on memory and attention tasks (n=1)• effects on EEG (n=2)• effects on human sperm in vitro (n=2)• testicular oxidative stress (n=1)• gene expression in embryonic stem cells (n=1)

• All studies reported effects for one or more endpoints


Other Studies

• A wide range of endpoints• human performance on memory and attention tasks (n=1)• effects on EEG (n=2)• effects on human sperm in vitro (n=2)• testicular oxidative stress (n=1)• gene expression in embryonic stem cells (n=1)

• All studies reported effects for one or more endpoints

• Problems:• Dose (unknown in 4/6)• Control of thermal effects (unknown in 3/6)• Blinding of exposure and analysis (not mentioned in 5/6)• Sham-exposed controls (absent in 5/6)


Effects on Human Performance and EEG

• Several of these studies reported effects of WiFi-like exposures on human performance or EEG • Some previous studies have also reported effects of low-level RF

exposures (other than WiFi) on brain activity or EEG -but these effects have generally been small and difficult to independently confirm

• The possible biological significance of the reported effects (e.g., small changes in amplitude of the EEG or P300-evoked response) is unclear in any event.

• Moreover, interpreting small changes in the EEG is complicated by potential artifacts associated with interaction between the RF field and EEG electrodes and their sensitive amplifiers.


WiFi and “Electrosensitivity”

• Related are reports in the mass media and on the Internet that WiFi exposure causes subjective symptoms in humans• for example: headaches, fatigue, skin sensations• the effect popularly called “electrohypersensitivity” or “electrical

hypersensitivity” (EHS).

• No reports were identified in the peer-reviewed literature that specifically examined EHS and WiFi exposures. • Challenge studies (in which hypersensitive individuals are exposed

to RF energy) have shown that such individuals are unable to identify exposure at levels that are not thermally-significant


WiFi and “Electrosensitivity”

• Related are reports in the mass media and on the Internet that WiFi exposure causes subjective symptoms in humans• for example: headaches, fatigue, skin sensations• the effect popularly called “electrohypersensitivity” or “electrical

hypersensitivity” (EHS).

• No reports were identified in the peer-reviewed literature that specifically examined EHS and WiFi exposures. • Challenge studies (in which hypersensitive individuals are exposed

to RF energy) have shown that such individuals are unable to identify exposure at levels that are not thermally-significant

• In addition the possibility must be considered that the media and Internet reports themselves may lead people to experience these types of subjective symptoms (Witthöft et al. 2013).


WiFi Effects on Sperm

• The three studies showing effects of on sperm were done under very different conditions, with poorly-characterized exposure systems, and with doses that ranged from uncertain to unknown. • The one study of WiFi effects on male fertility done with a well-

characterized exposure system found no effects (Poulletier de Gannes et al. 2013).

• The literature on effects of low-level RF exposure (other than with WiFi) on the testes is diverse and inconsistent• and does not make a convincing case for effects of RF radiation at

levels that do not cause significant temperature increases

• Before these studies can be used in risk assessment they need to be repeated with:• better study design, including blinding and sham-exposed controls• better exposure assessment.


Why the Focus on Fertility and Fetal Effects?

• No biological (or biophysical) reasons to expect that WiFi exposure would affect these endpoints.• Perhaps responding to calls by the WHO for studies of

reproductive effects of RF exposure.

• Perhaps also driven by assumption that WiFi-enabled laptops are actually used in laps• Such use is not that common

- In part because of the uncomfortable level of heatThere are reports burns and dermatitis from such use

• And since the antenna in a laptop (not a tablet) is typically in the upper case (behind the screen) and vertically oriented- Little actual testicular exposure occurs

• Ovaries and fetuses in humans are not shallow enough to get appreciable WiFi exposure


Biological Studies of the Different ISM Bands

• All of the biological studies using WiFi exposures we found used the 2450 MHz ISM band

-none used the 5200 or 5800 MHz bands.

• The major biophysically-relevant difference between these frequencies is a shorter penetration depth at the higher frequency• 0.4 (5800 MHz) vs. 1.1 cm (2450 MHz) in muscle (Gabriel et al. 1996).

• There are no other clear biophysical (or biological) reasons to expect the biological effects of WiFi radiation to be different for the two bands


Summary of the Biological Studies

• Several studies reported diverse biological effects of WiFi radiation. • As a group the studies are weak, variously lacking:

- blinding and sham controls- adequate dosimetry- adequately characterized exposure systems- control for possible heating artifacts.

• The reported the effects are generally small, and uncertain in their health significance. • With the possible exception of the EEG studies, all of the studies that

reported effects of WiFi exposure are inconsistent with similar studies using non-thermal RF exposures of other waveforms.


Summary of the Biological Studies

• Several studies reported diverse biological effects of WiFi radiation. • As a group the studies are weak, variously lacking:

- blinding and sham controls- adequate dosimetry- adequately characterized exposure systems- control for possible heating artifacts.

• The reported the effects are generally small, and uncertain in their health significance. • With the possible exception of the EEG studies, all of the studies that

reported effects of WiFi exposure are inconsistent with similar studies using non-thermal RF exposures of other waveforms.

• Despite the weaknesses of the studies, the reported effects have been widely invoked on the Internet to justify claims of hazard


Summary of What We Know

• Engineering aspects of WiFi exposure are complex, but well understood• We have a good (but not perfect) understanding of user exposure




• The biological literature is uneven in quality• The higher-quality studies report no effects• But some effects have been reported

-Technical limitations make these effects difficult to interpret-Artifacts cannot be excluded




• The biological literature is uneven in quality• The higher-quality studies report no effects• But some effects have been reported

-Technical limitations make these effects difficult to interpret-Artifacts cannot be excluded

• The larger body of RF studies (and mechanistic considerations) gives us no basis for anticipating any health effects at all from WiFi exposure.• But the WiFi-specific health and biology literature is scant


How to Respond to Public Concerns?

• Some further research may be warranted• A full-scale effort (as with mobile phones) seems unnecessary• But more exploratory biological studies of a variety of endpoints, won’t

help either




help either

• If further biological studies of WiFi are to be done:• They should be done in vivo

- with endpoints clearly related to human health• Given the lack of any known mechanism in vitro studies are nearly

impossible to interpret




help either

• If further biological studies of WiFi are to be done:• They should be done in vivo

- with endpoints clearly related to human health• Given the lack of any known mechanism in vitro studies are nearly

impossible to interpret

• A useful approach for health agencies would be to carefully monitor the bioeffects literature • If a mechanism for low-level effects were identified, or a biological effect

at low exposure levels were demonstrated• then a useful research program could be planned.


Appropriate Experimental Design

• Models should take into account the limited penetration off the WiFi signal• No point in examining organ systems in rodents that would not be

exposed in humans

• Well-characterized exposure systems and dosimetry are needed• Putting a lap-top computer on top of a mouse cage is not sufficient• Thermal effects must be prevented• This will be expensive


Appropriate Experimental Design

• Models should take into account the limited penetration off the WiFi signal• No point in examining organ systems in rodents that would not be

exposed in humans

• Well-characterized exposure systems and dosimetry are needed• Putting a lap-top computer on top of a mouse cage is not sufficient• Thermal effects must be prevented• This will be expensive

• Good study design is critical• Blinding of exposure and assessment• Positive controls and sham-irradiated controls

• Relevant SAR, duration and duty cycle should be used


The Big Problems with Planning WiFi Studies

• The dearth (complete lack?) of unequivocal biological effects from low-level RF exposures, and lack of a biophysical or biological basis for expecting any such effects• means that we have no basis for recommending study endpoints

• The large parameter space with Wi-Fi (modulation, pulse parameters and frequency); and the rapidly evolving nature of wireless networking • make it impossible to design comprehensive bioeffects studies that would

convince all critics that all bases had been covered.


The Big Problems with Planning WiFi Studies

• The dearth (complete lack?) of unequivocal biological effects from low-level RF exposures, and lack of a biophysical or biological basis for expecting any such effects• means that we have no basis for recommending study endpoints

• The large parameter space with Wi-Fi (modulation, pulse parameters and frequency); and the rapidly evolving nature of wireless networking • make it impossible to design comprehensive bioeffects studies that would

convince all critics that all bases had been covered.

• These problems will arise with other wireless technologies, • and many more are coming in the near future.


A Final Note of Caution

• Wi-Fi and WLANs can raise immediate and urgent safety issues apart from possible RF bioeffects.

• Wireless networks (depending on how they are configured) are more or less susceptible to privacy invasion and hacking.

• The Internet, which many access via WLANs, raises safety issues (particularly with children) that have nothing to do with RF exposure.


A Final Note of Caution

• Wi-Fi and WLANs can raise immediate and urgent safety issues apart from possible RF bioeffects.

• Wireless networks (depending on how they are configured) are more or less susceptible to privacy invasion and hacking.

• The Internet, which many access via WLANs, raises safety issues (particularly with children) that have nothing to do with RF exposure.

• Excessive concern about speculative health hazards from RF exposures to Wi-Fi, without concern for these more immediate potential hazards, is comparable to worrying about health effects of using mobile phones without concern for hazards of texting while driving.

Documents

Wi-Fi and Health: Review of the Current Research