Conference PaperPDF Available

Ambient Interference Effects in Wi-Fi Networks


Abstract and Figures

This paper presents a measurement study of interference from six common devices that use the same 2.4 GHz ISM band as the IEEE 802.11 protocol. Using both controlled experiments and production environment measurements, we quantify the impact of these devices on the performance of 802.11 Wi-Fi networks. In our controlled experiments, we characterize the interference properties of these devices, as well as measure and discuss implications of interference on data, video, and voice traffic. Finally, we use measurements from a campus network to understand the impact of interference on the operational performance of the network. Overall, we find that the campus network is exposed to a large variety of non-Wi-Fi devices, and that these devices can have a significant impact on the interference level in the network. KeywordsWi-Fi-Interference-Spectrogram-Duty Cycle-Data-Video-Voice
Content may be subject to copyright.
Ambient Interference Effects in Wi-Fi Networks
Aniket Mahanti1, Niklas Carlsson1, Carey Williamson1, and Martin Arlitt1,2
1Department of Computer Science, University of Calgary, Calgary, Canada
2Sustainable IT Ecosystem Lab, HP Labs, Palo Alto, U.S.A.
Abstract. This paper presents a measurement study of interference from six
common devices that use the same 2.4 GHz ISM band as the IEEE 802.11 pro-
tocol. Using both controlled experiments and production environment measure-
ments, we quantify the impact of these devices on the performance of 802.11
Wi-Fi networks. In our controlled experiments, we characterize the interference
properties of these devices, as well as measure and discuss implications of inter-
ference on data, video, and voice traffic. Finally, we use measurements from a
campus network to understand the impact of interference on the operational per-
formance of the network. Overall, we find that the campus network is exposed to
a large variety of non-Wi-Fi devices, and that these devices can have a significant
impact on the interference level in the network.
Keywords: Wi-Fi, Interference, Spectrogram, Duty Cycle, Data, Video, Voice.
1 Introduction
Wireless Fidelity (Wi-Fi) networks allow users with wireless-capable devices to access
the Internet without being physically tied to a specific location. Today’s Wi-Fi networks
are not only being used for Web surfing, but also for viewing live video content (e.g.,
live sports and news events) and voice communication (e.g., voice over IP or VoIP).
Many of these services require high quality of service and reliability, which can be
degraded by wireless interference.
Wi-Fi networks employ the IEEE 802.11 protocol, which uses the unlicensed 2.4
GHz Industrial, Scientific, and Medical (ISM) Radio Frequency (RF) band [6]. Since
the ISM band is unlicensed, it is available for use by multiple devices (both Wi-Fi
and non-Wi-Fi), inherently causing interference for one another. The 802.11 protocol is
considered to be a polite protocol in that an 802.11 device will transmit only if it senses
that the RF channel is free. Non-Wi-Fi devices such as microwave ovens are oblivious
to this protocol. These devices transmit regardless of whether the channel is free or not.
This makes the interference problem challenging in Wi-Fi networks.
Network practitioners often deploy Wi-Fi networks without knowledge of the ambi-
ent usage of the ISM band by non-Wi-Fi devices. For example, trace capture programs,
IFIP, (2010). This is the author’s version of the work. It is posted here by permission of IFIP
for your personal use. Not for redistribution. IFIP Networking 2010, Chennai, India.
employed for site surveys used to select RF channels for problem-free functioning of
Wi-Fi networks, only recognize devices using the 802.11 protocol. Since this process
concentrates on the link layer and up, any activity from non-Wi-Fi devices is ignored.
Unfortunately, since several non-Wi-Fi devices may be operating on the same channel
as a Wi-Fi network, and cause severe interference, it is difficult to identify interference
sources using this trace capture technique.
For a more complete picture of the activity on the 2.4 GHz ISM band, one needs
to consider the physical layer. In this paper, we use an off-the-shelf wireless spectrum
analyzer to understand how non-Wi-Fi devices impact the functioning of Wi-Fi net-
works. Using controlled experiments, we characterize the interference properties of six
non-802.11 devices. Five are unintentional interferers: a microwave oven, two cord-
less phones (one analog and one digital), an analog wireless camera, and a Bluetooth
headset. We also evaluate one intentional interferer, a wireless jammer, for comparison
purposes. In addition to capturing the basic characteristics of these devices, we mea-
sure, quantify, and discuss implications of their interference on data, video, and voice
traffic. Finally, using passive measurements from an operational campus network, we
try to understand the impact of interference on the network.
Our results show that among the unintentional interferers, microwave ovens, analog
cordless phones, and wireless cameras have the most adverse impact on Wi-Fi networks.
Because of its wideband interference, microwave signals affect several Wi-Fi channels
at close range, however, their impact is still felt at longer distances. Analog cordless
phones and wireless cameras are continuous narrowband interferers that completely
obliterate Wi-Fi service on any channels they are using. Digital cordless phones and
Bluetooth headsets have minimal impact on Wi-Fi because of their frequency hopping
nature. From our production network measurements, we find that the campus network
is exposed to a large variety of non-Wi-Fi devices, and that these devices can have a
significant impact on the interference level in the network. For example, during certain
times of the day, almost 80% of a channel may be occupied by interferers, and it is
common to see some interference device active (in the background) almost all the time.
The rest of the paper is organized as follows. Section 2 discusses the interferers we
studied and our experimental setup. Section 3 describes the organization of the Wi-Fi
channels and characterizes the physical layer properties of the six interferers. Section
4 studies the impact of non-802.11 interference on data, video, and voice traffic over
Wi-Fi. Section 5 studies the channel utilization of Wi-Fi and various interferers in a
production network. Section 6 describes related work. Section 7 concludes the paper.
2 Methodology
2.1 Interferers
Microwave Oven: We used a Panasonic NNS615W microwave oven. This device had a
maximum output power of 1,200 W. During our experiments, the oven was operating at
maximum power and had a container of food inside the oven.
Analog Wireless Video Camera: We used a Lorex SG8840F analog wireless video cam-
era operating at 2.4 GHz. These cameras provide long-range surveillance using analog
signals, typically use directional antennas, and can have a range up to 1.5 km.
8 meters
Server Access
5, 10, 15, 20,
25, 30 meters
(a) Data and Video Experiment (b) Voice Experiment
Fig. 1. Experimental setup
Analog Cordless Phone: We used a Vtech GZ2456 analog cordless phone operating at
2.4 GHz. The phone consists of a base and a handset, and as with other analog phones,
works by converting voice into electronic pulses and sending them on a designated
channel between the base and handset.
Digital Cordless Phone: We used a Uniden DCT648 digital phone (base and handset)
operating at 2.4 GHz. This particular phone uses Digital Spread Spectrum (DSS) to
change channels frequently for enhanced voice quality and security.
Bluetooth Headset: We used a Plantronics Pulsar 590A headset. These are short range
devices and have a maximum operating range of 10 m.
Wireless Jammer: Wireless jammers may be used for RF Denial of Service attacks. The
power levels of these jammers vary from 1 mW to 30 mW.
2.2 Experimental Setup
We conducted two sets of measurement experiments to study the physical layer charac-
teristics of the above-mentioned interferers and their impact on Wi-Fi networks.
The first set of experiments examines the physical-layer characteristics of these in-
terferers in an isolated environment. We used an off-the-shelf spectrum analyzer called
AirMagnet Spectrum Analyzer3for this purpose. The spectrum analyzer is a hard-
ware/software unit that consists of a radio for detecting RF energy in the 2.4 GHz
ISM band, and a software engine that performs Fast Fourier Transforms (FFTs). The
spectrum analyzer uses these FFTs to classify known interferers. The physical layer
measurements were taken in an interference-neutral environment such that we only
captured RF energy from the specific device under study.
The second set of experiments quantifies the impact of interferers on the perfor-
mance of a Wi-Fi network. We measured the performance degradation of the network
22 MHz 5 MHz
1 2 3 4 5 6 7 8 910 1311 12 14
Channel Centre Frequency (GHz)
Fig. 2. Structure of Wi-Fi Channels in the 2.4 GHz band
in the presence of the interferers for three types of traffic workloads: data, video, and
voice. Our Wi-Fi network consisted of a single D-Link 2100 access point (AP) running
in IEEE 802.11g mode. We chose the 802.11g standard because it offers higher trans-
mission rates than 802.11b and is a popular choice for deployment in enterprise, aca-
demic, home, and hot spot networks. Additionally, most modern laptops are equipped
with 802.11g capable Network Interface Cards (NICs).
We used the setup shown in Figure 1(a) for data and video experiments. We con-
nected a server workstation using an Ethernet cable. We placed four Lenovo T61 laptop
clients with built-in 802.11abg NICs at a distance of 8 m from the AP. We placed a
single interferer at successive distances of 5, 10, 15, 20, 25, and 30 m from the AP to
record the performance degradation of the interferer on the network. For each interferer,
we repeated this scenario four times to record our readings.
The experimental setup for the voice experiment is shown in Figure 1(b). The AP
was placed in the centre and was connected to a measurement workstation using an
Ethernet cable. We placed two pairs of laptops facing each other diagonally at a distance
of 8 m from the AP. We configured Ekiga4VoIP software on the laptops to allow direct
IP calling without using a SIP server. VoIP client 1 was communicating with VoIP client
2 and VoIP client 3 was communicating with VoIP client 4. Interferer placement was
the same as in the case of the data and video experiments.
3 Physical-Layer Characteristics
3.1 Channel Structure in Wi-Fi
Wi-Fi channels are organized into 14 overlapping channels each having a spectral band-
width of 22 MHz. Figure 2 shows a graphical representation of the Wi-Fi channels in
the 2.4 GHz band. The figure shows the centre frequencies of each Wi-Fi channel. Ad-
jacent channels are separated by 5 MHz, except for channel 14 whose centre frequency
is separated from channel 13 by 12 MHz. A single channel can handle 50 or more
simultaneous users [6]. Usage of Wi-Fi channels are governed by national regulatory
(a) Microwave Oven (b) Analog Wireless Video Camera
(c) Analog Cordless Phone (d) Digital Cordless Phone
(e) Bluetooth Headset (f) Wireless Jammer
Fig. 3. Spectrograms for six non-Wi-Fi devices
agencies of the respective countries. In North America, only the first 11 channels are
available for use. In the rest of the world the first 13 channels are available for use.
Japan allows the use of channel 14 as well, however, it is only available to 802.11b
using Direct Sequence Spread Spectrum (DSSS) modulation. We restrict our attention
to the 11 channels in use in North America.
To avoid interference, wireless radios are expected to operate on non-overlapping
channels, i.e., channels separated by at least 22 MHz. For example, if two APs are
operating on the same channel in a wireless cell, then their signals will interfere with
each other. The same applies to any other radiating device, such as a microwave oven or
cordless phone. From Figure 2, we observe that the following channel combinations do
not overlap with each other: {1,6,11},{2,7},{3,8},{4,9}, and {5,10}. Channels 1, 6,
and 11 are the most commonly used non-overlapping channels in Wi-Fi deployments.
(a) Microwave Oven (b) Analog Wireless Video Camera
(c) Analog Cordless Phone (d) Digital Cordless Phone
(e) Bluetooth Headset (f) Wireless Jammer
Fig. 4. Duty cycle analysis for six non-Wi-Fi devices
3.2 Metrics
We use spectrograms and duty cycles to characterize interferers at the physical layer.
The spectrogram is a representation of the RF power levels over time in the spectrum.
Each vertical line in the spectrogram shows the RF power as a function of frequency
measured over a time interval of 1 second. Spectrograms offer a temporal perspective
of RF power in the frequency domain. The duty cycle measures the RF power in the
spectrum. In this work, duty cycle is calculated by measuring the percentage of time the
RF signal is 20 dBm above the noise floor. Duty cycle is an indicator of the impact of
RF power on network performance. We next describe the physical-layer characteristic
of each interferer in isolation.
3.3 Measurement Results
Figure 3 shows the spectrograms for the interferers. The X-axis represents the time
period of the measurements. The Y-axis tic marks represent the centre frequencies of
the even numbered Wi-Fi channels; i.e., channels 2, 4, 6, 8, 10, 12, and 14. The colour
contour lines represent the power levels of the signal, where red indicates the strongest
and blue the weakest power levels. When interpreting these graphs, it is important to
note that each device may use a different range of RF power levels and the same colour
therefore may refer to a different power level in the different graphs. Figure 4 shows the
duty cycle FFT measurements for the interferers. The bottom X-axis tic marks represent
the centre frequencies of the even numbered Wi-Fi channels, while the top X-axis tic
marks show the channel numbers corresponding to those centre frequencies.
Microwave Oven: Figure 3(a) shows that the microwave oven affects about half the
available channels (6-12) in the 2.4 GHz band with the highest energy concentrated on
channel 9. Note that the microwave oven is operating at high power levels of -80 to
-60 dBm. RF power levels above -80 dBm are enough to cause interference with Wi-
Fi networks. Figure 4(a) shows that the average duty cycle of the microwave is 50%.
Since the microwave signal sweeps through a wideband of the spectrum and has a high
average duty cycle, it is likely to affect nearby Wi-Fi devices. Microwave ovens are
created with a shield such that all radiation is restricted to the oven cavity, however,
with use over time these ovens can leak some radiation.
Analog Wireless Video Camera: Figure 3(b) illustrates the narrowband continuous
transmitting nature of an analog video camera. The transmit power is similar to that of
the microwave oven, however, in this case this energy is concentrated on a very small
portion of the spectrum (channels 4-8). Figure 4(b) shows that the duty cycle of the
analog video camera reaches 100% indicating that no Wi-Fi device in the vicinity will
be able to operate on channels 4-8. Because of its continuous transmission nature, this
device can cause prolonged periods of service disruption.
Analog Cordless Phone: The analog phone concentrates most of its energy on chan-
nel 3 in Figure 3(c). The figure also highlights the narrowband fixed-frequency trans-
mission nature of the analog phone. Figure 4(c) shows that the analog phone has duty
cycle as high as 85%. The high duty cycle of the analog phone indicates that it will
severely impact Wi-Fi operation. Analog phones are quickly becoming out-dated, how-
ever, they are still available for purchase and are widely used around the world.
Digital Cordless Phone: The frequency hopping feature of the digital phone is il-
lustrated in Figure 3(d). This phone utilizes DSS, a newer technology than the one used
by analog phones. The phone continuously changes channels, only staying on a portion
of the spectrum for a small amount of time, reducing its interference. In Figure 4(d)
we observe that the maximum duty cycle is 4.5%, indicating that this device may not
severely interfere with Wi-Fi devices.5
Bluetooth Headset: The Frequency Hopping Spread Spectrum (FHSS) nature of
Bluetooth is highlighted in Figure 3(e). The Bluetooth device hops across all the chan-
nels. Although the energy emitted by the Bluetooth device may appear high, its duty
5More recently, Digital Enhanced Cordless Telecommunications (DECT) phones have become
popular. These phones operate on the 1.9 and 5.8 GHz band and do not interfere with 802.11
cycle values are much lower. Furthermore, FHSS technology is limited to 2 Mbps and
does not consume much of the available bandwidth. Figure 4(e) shows that the maxi-
mum duty cycle attained is 3.5%, which may not affect Wi-Fi devices seriously.
Wireless Jammer: Figure 3(f) shows the wideband characteristics of a wireless jam-
mer. The jammer emits signals on all channels at high power levels in quick succession.
Figure 4(f) shows the duty cycle varies between 10% and 60%. Wi-Fi devices use Clear
Channel Assessment to sense when the channel is clear for transmission. The wideband
jammer ensures that the RF medium is never clear, thus preventing Wi-Fi devices from
functioning properly.
4 Impact of Interferers on Wi-Fi Traffic
In this section, we study the impact of interferers on Wi-Fi traffic. In particular, we
study the impact of interferers on data traffic, video traffic, and voice traffic. The first
experiment for each workload was conducted in an interference-neutral environment.
Next, we repeated the experiments where the Wi-Fi link was subjected to interference
from one specific interferer at a certain distance. The distances used were 5, 10, 15,
20, 25, and 30 m. We did not consider the digital phone since its physical layer char-
acteristics are similar to that of Bluetooth. Each experiment was performed four times,
and we report the average percentage difference between the baseline performance (no
interference) and that when the experiment was subjected to interference. For the data
experiment, we used the throughput Quality of Service (QoS) metric. For the video
and voice experiments, we used a Quality of Experience6(QoE) metric called Mean
Opinion Score (MOS).
4.1 Experimental Workload Traffic
Data Traffic: We used the Iperf7tool to measure throughput of the Wi-Fi link. We ran
Iperf in server mode at the server and in client mode at the four wireless client laptops.
Iperf can run throughput tests using TCP or UDP packets. We used the TCP option for
our experiment. For the TCP tests, Iperf requires the user to set an appropriate TCP
window size. If the window size is set too low, the throughput measurements may be
incorrect. We found that a window size of 148 KB was sufficient to properly measure
the throughput of the Wi-Fi link. The workload consisted of creating bidirectional TCP
traffic between the server and the four clients for a period of 3 minutes.
Video Traffic: Video (or Voice) quality can be quantified using subjective methods
such as MOS. For our experiments, we used a hybrid subjective assessment scheme
called Pseudo Subjective Quality Assessment (PSQA) [7, 8]. PSQA uses random neu-
ral networks to automatically calculate MOS of a video or voice sample. We set up the
server workstation to stream a 3-minute video using the VLC8media player. The (nor-
mal quality standard-definition) video had a resolution of 624 ×352, frame rate of 29
frames per second, and was encoded using the Xvid9codec at a bit rate of 656 Kbps.
(a) Data Traffic (b) Video Traffic
(c) Voice Traffic
Fig. 5. Impact of interference on different workloads
The VLC players on the four clients were configured to receive this video stream. We
collected the MOS measurements using PSQA from the client-side.
Voice Traffic: We recorded a 3-minute VoIP conversation between two people. We
next separated the voices of the two people and created two different audio files. Recall
our voice experiment setup had two pairs of communicating laptops. After establishing
a direct connection via the Wi-Fi link for a pair of laptops, we played one audio file on
one end and the second audio file on the other end. The two audio files were synchro-
nized such that at no time the two audio files played a human voice simultaneously. The
(telephone quality) audio files were encoded using the FLAC10 codec at a sampling rate
of 16 kHz and a bit rate of 56 Kbps. We used a workstation connected to the AP to
collect the MOS measurements using PSQA of the voice communication.
4.2 Experimental Results
Figures 5(a), (b) and (c) show the degradation in data throughput, video quality, and
voice quality, respectively, in the presence of interference.
As the diverse nature of these workloads require different quality measures, direct
comparisons of the degradation of the different services require some care. Voice traf-
fic, which typically uses smaller packets, handles interference the best (with the metrics
used here). As expected, we consistently notice close to 100% degradation for our ex-
periments using the wireless jammer. This is consistent up to 20 m, beyond which we
observe a slight decline in its impact. In the following, we discuss the impact of the
unintentional interferers on the different workloads.
Data Traffic: The Bluetooth headset reduced the throughput by 20% at close dis-
tances. The degradation may be surprising, since Bluetooth devices have low duty cycle
and are designed to accommodate Wi-Fi devices; however, it is less serious compared
to other interferers. For example, the microwave oven resulted in zero throughput at
close distance. While its impact declined gradually as it was moved away from the Wi-
Fi network, it caused a 25% degradation in throughput even at a distance of 25 m. The
analog phone and video camera are both continuous transmitters. Hence, their impact
on throughput is similar. Their interference is significant at close distance, however,
there is a sharp decrease at distances beyond 20 m.
Video Traffic: Although Bluetooth had some impact on data traffic, there was min-
imal impact on video traffic. The microwave oven at close distance severely disrupted
the video stream, while at longer distances it reduced the video quality by only 10%.
The analog camera and analog phone had similar impacts on the video stream. Even at
longer distances, they reduced the video quality by 50%.
Voice Traffic: Bluetooth had minor impact at short distance and no impact at longer
distances. The microwave oven caused approximately 75% degradation at close range,
and this plummeted as the distance increased. The analog phone and video camera had
severe impact at close range, but this consistently decreased with longer distances. They
still caused about 30% degradation at a distance of 30 m. The impact of interference is
comparatively lower for voice traffic.
5 Interference in Campus Network
We used the Spectrum Analyzer to study the channel utilization of our campus network.
We took physical-layer measurements for 8 hours during a weekday (Tuesday, January
20, 2009). The passive measurements were taken at a popular location in the campus
frequented by students and faculty members. We report our results using the channel
utilization metric. Channel utilization is the percentage of time a transmission is present
from a known RF source in a given channel. It helps us understand how much of the
channel is available for use at a given time.
Figure 6(a) shows the channel utilization of the three channels used in our campus
network. The figures record the channel utilization by all devices (Wi-Fi and Interferers)
transmitting in the 2.4 GHz ISM band.11 We observe that channels 1 and 11 had heavier
usage in comparison to channel 6. In case of channels 1 and 6, utilization peaked near
60%, while for channel 11 it was over 90%. After close inspection, we found that most
of the spikes were caused by interferers.
On channel 11, the spikes are the most alarming. We focus on three spikes and try
to understand the interferers involved. These spikes are observed at times 12:45, 15:38,
and 17:17. We found that each of these spikes was mainly caused due to microwave
11 The Spectrum Analyzer cannot separate the Wi-Fi and non-Wi-Fi channel utilization over time,
but it did allow to us to investigate the utilization at particular points in time.
(a) Channel Utilization (b) Interferers on Channel 11
(c) Active Time of Interferers
Fig. 6. Passive measurements from campus Wi-Fi network
ovens and cordless phones operating simultaneously in the area of the Wi-Fi network.
For the first spike, we also noticed some other fixed-frequency interferers, whereas the
second and third spike were highly influenced by Bluetooth communication. In total,
the interferers consumed 80% of the channel during the first spike. The second and third
spike used much less of the channel.
Figure 6(b) shows the average number of interferers observed on channel 11 as a
function of time. We observed 100s of interference occurrences involving non-Wi-Fi
devices on channel 11. These include 22 instances of Bluetooth device pairs commu-
nicating (6 unique piconets), 3 instances of analog cordless phones (base and handset),
71 instances of digital cordless phones (30 bidirectional base and handset instances and
41 cases where only the base station was observed), as well as 46 instances involving
other unclassified interferers.
Overall, we found that microwave ovens and cordless phones were surprisingly ac-
tive throughout the trace period, while Bluetooth piconet instances were temporally
clustered. Furthermore, we note that most active times for the various interferers were
relatively short, but with a few long instances. For example, the average active period
for microwave ovens (2:12), Bluetooth (3:05), digital cordless phones (7:32) and other
interferers (0:42) were relatively short compared to the corresponding devices’ longest
observed active periods (16:37, 22:27, 2:28:01, 5:02).12
Figure 6(c) shows the cumulative active time (i.e., the time the interferers were
transmitting) of interferers over the trace period. Microwave ovens were active 31% of
12 The active period length is reported in hours:minutes:seconds.
the total trace period and had an average (and maximum) duty cycle of 9% (40%). The
average signal strength was approximately -80 dBm, which is enough to cause some
interference. Bluetooth devices were active 12.5% of the time and had an average (low)
duty cycle of roughly 5%. With digital base stations always transmitting (at a duty cycle
<7%) when not connected with the handset, it was not surprising that we found that
cordless phones were cumulatively active for more than the entire trace period. We
note, however, that the average (maximum) duty cycle increased to 15% (20%) when
the base and handset were connected. Other interferers were observed active only 6%
of the total trace period, but had the highest average duty cycle (15%), maximum duty
cycle (60%), and signal strength (-43 dBm) of all the interferers observed in the campus
Wi-Fi network.
The exact channel usage of these observed non-Wi-Fi devices may differ from that
of the particular devices studied in our controlled experiments, however, the general
characteristics are similar. With a small number of heterogeneous non-Wi-Fi devices
active at each point in time, it is not surprising that we do not find strong correlation
between the average number of active devices and the peaks observed above (e.g., com-
paring Figures 6(a) and 6(b)). Instead, individual device characteristics and the distance
of devices from the network are likely to have the most impact. Referring back to the
impact of device and distance (Section 4), it is clear that a few non-Wi-Fi devices can
have a huge effect on the performance of the Wi-Fi traffic, especially if located close
to the network. Furthermore, while our measurements were taken (in a student/faculty
lounge) away from the campus food court and office areas, the interference from mi-
crowave ovens and cordless phones would likely be even more significant in such areas.
We leave comparisons of the non-Wi-Fi interference in different areas as future work.
6 Related Work
Interference in wireless networks has been investigated mostly in the context of coex-
isting Wi-Fi networks, Bluetooth networks, and microwave ovens.
Taher et al. [14] used laboratory measurements to develop an analytical model of
microwave oven signals. Karhima et al. [12] performed measurements on an ad-hoc
wireless LAN under narrowband and wideband jamming. They found that in case of
wideband jamming, 802.11g can offer higher transmission rates than 802.11b, when the
packet error rate is the same. They also found that 802.11g was more prone to complete
jamming, while 802.11b could still operate at lower transmission rates due to its DSSS
modulation scheme. Golmie et al. [9] explored the mutual impact of interference on a
closed loop environment consisting of Wi-Fi and Bluetooth networks. They found that
even by sufficiently increasing the transmission power levels of the Wi-Fi network to
that of the Bluetooth network could not reduce packet loss. Our work complements
these studies. We use measurements to characterize a wide range of common non-Wi-
Fi devices (including microwave ovens and Bluetooth) at the physical layer and use
experiments to quantify their impact on different traffic workloads.
Vogeler et al. [15] presented methods for detection and suppression of interference
due to Bluetooth in 802.11g networks. Ho et al. [11] studied the performance impact on
a Wi-Fi network due to Bluetooth and HomeRF devices using simulations and described
design challenges for deployment of 5 GHz wireless networks. Our work is orthogonal
to these papers. We focus on passive measurements to study interference from common
non-Wi-Fi devices.
Gummadi et al. [10] studied the impact of malicious and unintentional interferers
on 802.11 networks. They found that some of these interferers could cause consider-
able performance degradation for commercial 802.11 NICs. Using an SINR model, they
noted that changing 802.11 parameters was not helpful, and proposed a rapid channel
hopping scheme that improved interference tolerance for these NICs. Farpoint Group
studied the effect of interference on general, video, and voice traffic in a Wi-Fi network
from two vantage points: short-range (25 feet) and long-range (50 feet) [3–5]. In con-
trast, we present a comprehensive measurement study of impact of interference from
non-Wi-Fi devices. We also used QoS and QoE metrics to quantify the degradation
of video and voice quality in a Wi-Fi network due to interference. We observed that
operational Wi-Fi networks may be subject to ambient interference effects at anytime.
7 Concluding Remarks
In this paper, we characterized the RF behaviour of non-Wi-Fi devices, analyzed the
impact of interference on data, video, and voice traffic, and examined interference in a
live campus network. Overall, we found that the campus network is exposed to a large
variety of non-Wi-Fi devices, and that these devices can have a significant impact on
the interference level in the network. Our controlled experiments showed that, even at
distances up to 30 m, some of these non-Wi-Fi devices can have a significant negative
impact on data, video, and voice traffic. While microwave ovens, wireless analog video
cameras, and analog cordless phones typically have the most significant negative im-
pact on Wi-Fi networks, the performance degradation due to a digital cordless phone
or a Bluetooth device (which tries to be Wi-Fi friendly) can be noticeable (e.g., 20%
at close distances). With the campus network (and likely many other home and enter-
prise networks) being highly exposed to different types of unintentional interferers, it is
important to find (new) ways to identify and mitigate non-Wi-Fi interference.
Network practitioners often use link-layer and transport-layer statistics to investi-
gate interference in a Wi-Fi network. We note that these alone should not be the means
for troubleshooting a network. We observe that physical-layer characteristics may be
used as primary indicators to identify and mitigate interferers [2]. In general, we be-
lieve that interference can be mitigated by identifying and removing the interfering
device (if possible) or shielding interferers instead. Careful channel selection at APs
may be helpful (e.g., in both our controlled and real world measurements we found
that microwave ovens primarily affected channels 6 and higher). APs may be fitted with
sensors to detect interference and switch channels automatically [1]. Using multi-sector
antennas [13] and controlling data rates to avoid false backoffs can make the network
more interference-resilient (although this is a trade-off since the lower data rates allow
more noise-immune communication). We conclude by noting that there are many differ-
ent non-Wi-Fi devices that may cause interference in Wi-Fi networks, and it is therefore
important to understand and quickly adapt to the devices affecting the performance of
the Wi-Fi channels.
8 Acknowledgements
Financial support for this research was provided by Informatics Circle of Research Ex-
cellence (iCORE) in the Province of Alberta, as well as by Canada’s Natural Sciences
and Engineering Research Council (NSERC). The authors also thank the anonymous
reviewers for their constructive comments on an earlier version of this paper.
1. Aruba Networks: Advanced RF Management for Wireless Grids. Technical Brief (2007),
Available online at
2. Cisco: 20 Myths of Wi-Fi Interference: Dispel Myths to Gain High-Performing and
Reliable Wireless. White Paper (2007), Available online at
3. Farpoint Group: The Effects of Interference on General WLAN Traffic. Techical
Note FPG 2006-328.3 (January, 2008), Available online at
4. Farpoint Group: The Effects of Interference on Video Over Wi-Fi. Technical Note FPG 2006-
329.3 (January, 2008), Available online at
5. Farpoint Group: The Effects of Interference on VoFi Traffic. Techical Note FPG 2006-330.3
(January, 2008), Available online at
6. Flickenger, R.: Building Wireless Community Networks. O’Reilly (2003)
7. Ghareeb, M., Viho, C.: Performance Evaluations of a QoE-Based Multipath Video Streaming
Mechanism over Video Distribution Network. In: Proc. of International Workshop on Future
of Multimedia Networking (FMN). pp. 236–241. Coimbra, Portugal (June, 2009)
8. Ghareeb, M., Viho, C., Ksentini, A.: An Adaptive Mechanism For Multipath Video Stream-
ing Over Video Distribution Network. In: Proc. of International Conference on Advances in
Multimedia (MMEDIA). pp. 6–11. Colmar, France (July, 2009)
9. Golmie, N., VanDyck, R., Soltanian, A., Tonnerre, A., Rebala, O.: Interference evaluation of
Bluetooth and IEEE 802.11b Systems. Wireless Networks 9(3), 201–211 (May, 2003)
10. Gummadi, R., Wetherall, D., Greenstein, B., Seshan, S.: Understanding and Mitigating the
Impact of RF Interference on 802.11 Networks. In: Proc. of ACM SIGCOMM International
Conference. pp. 385–396. Koyoto, Japan (August, 2007)
11. Ho, M.J., Rawles, M., Vrijkorte, M., Fei, L.: RF Challenges for 2.4 and 5 GHz WLAN
Deployment and Design. In: Proc. of IEEE Wireless Communications and Networking Con-
ference (WCNC). pp. 783–788. Orlando, U.S.A. (March, 2002)
12. Karhima, T., Silvennoinen, A., Hall, M., Haggman, S.: IEEE 802.11 b/g WLAN Tolerance
to Jamming. In: Proc. of Military Communications Conference (MILCOM). pp. 1364–1370.
Monterey, U.S.A. (October, 2004)
13. Lundgren, H., Subramanian, A., Salonidis, T., Carrera, M., Guyadec, P.L.: Interference Miti-
gation in WiFi Networks using Multi-sector Antennas. In: Proc. of ACM international Work-
shop on Experimental Evaluation and Characterization (WiNTECH). pp. 87–88. Beijing,
China (September, 2009)
14. Taher, T., Al-Banna, A., Ucci, D., LoCicero, J.: Characterization of an Unintentional Wi-Fi
Interference Device - The Residential Microwave Oven. In: Proc. of Military Communica-
tions Conference (MILCOM). Washinton D.C., U.S.A. (October, 2006)
15. Vogeler, S., Brotje, L., Kammeyer, K.D., Ruckriem, R., Fechtel, S.: Suppression of Blue-
tooth Interference on OFDM in the 2.4 GHz ISM Band. In: Proc. of International OFDM-
Workshop (InOWo). Hamburg, Germany (September, 2003)
... In order to understand, predict, and manage the effects of external wireless devices on IEEE 802.11 networks, research is needed about their effects on throughput, the primary measure of a network's quality of service [3,6,7]. This requires analyzing the physical (PHY) layer next to the data (MAC) layer of OSI model [22]. ...
... For example, one study found that microwave ovens decreased throughput by 28-49%, depending on proximity to the network (1-3 m), whereas Bluetooth loudspeakers reduced it by 5-9% [22]. Another study even found that microwave ovens could stop network throughput entirely [7]. Interference from microwave ovens can be mitigated using a "cognitive radio" technique [23]. ...
... Radiofrequency noise from a laptop computer can reduce throughput and other aspects of IEEE 802.11 network performance [28]. Analogue wireless video cameras and analogue cordless phones can reduce network throughput by 90-100% because they transmit continuously and therefore interfere with network traffic most of the time [7]. ...
Full-text available
Wireless networks, including IEEE 802.11-based or Wi-Fi networks, are inexpensive and easy to install and therefore serve as useful connectivity alternatives in areas lacking wired-network infrastructure. However, IEEE 802.11 networks may not always provide the seamless connectivity and minimal throughput required for Industry 4.0 communications because of their susceptibility to interference from other devices operating in the unlicensed “Industrial, Scientific, and Medical” frequency band. Here we analyzed how a wireless audio transmitter operating on this band influences the throughput of an IEEE 802.11 b/g/n network under laboratory conditions. Wireless audio transmission reduced mean throughput by 85%, rendering the IEEE 802.11 b/g/n network nearly unusable. Our analysis suggests that in order for IEEE 802.11 wireless networks to support Industrial 4.0 applications, attention should be paid to the physical layer as well as the data or upper layers, and critical services should not transmit on the 2.4 GHz band. These findings may contribute to understanding and managing IEEE 802.11 wireless networks in various Industry 4.0 contexts.
... The retransmission procedure is based on contention window (CW) window exponential increase. CW= [2 j (Wmin+2 j )-1], CWmin = 15 and CWmax = 1023 and the exponential increase is as (15,31, 63 ……1023) [7]. j is the number of retransmission and jmax is the maximum number of retransmission. ...
... From other hand, in term of channel occupancy, LAA and LTE-U have more chance of channel occupancy than Wi-Fi. Hence, the performance of Wi-Fi degrades when coexist with LAA and LTE-U.The study in[15] evaluated the ambient inference effects in 2.4GHz for Wi-Fi networks. They evaluated the impact of six non-Wi-Fi devices (Microwave oven, Bluetooth headset, Analog cordless phone, Digital cordless phone, Analog wireless video camera and Wireless jammer) on Wi-Fi performance. ...
... There are numerous studies related to Wi-Fi interference (Fuxjager, Valério, & Ricciato, 2007;Sui et al., 2016;Kokkinos et al, 2016;Golmie et al, 2003;Taher et al., 2008;Lee et. al, 2017;van Bloem et al., 2012;Mahanti et al, 2010;Soldo Malaric, 2013). Much of the available public literature related to Wi-Fi interference is focused either on co-channel or adjacent channel interference (Fuxjager, Valério, & Ricciato, 2007;Sui et al., 2016;Kokkinos et al, 2016) or on specific devices such as Wi-Fi Bluetooth ( Golmie et al, 2003) and microwave ovens (Taher et al., 2008). ...
... al, 2017) show severe impact of audio/video transmitters which causes significant overall QoS degradation of Wi-Fi communication in contrast to microwave and Bluetooth interference. Authors in ( Mahanti et al, 2010) provide an accurate measurement study of interference from six common devices that use the 2.4 GHz bandwidth in both controlled and uncontrolled campus environments. In (Soldo & Malaric, 2013) a simple user available app tool is used for a quick measurement setup in order to reveal the influence of a limited number of factors on Wi-Fi signal and upload and download data rates. ...
Full-text available
Operating in probably the most crowded radio frequency bandwidth, the 2.4 GHz Wi-Fi is still common in the majority of living environments. Besides having to face problems regarding co-channel and adjacent channel influence, Wi-Fi often coexists with other radio technologies operating in this bandwidth which makes it ideal for interference testing. In this article authors propose the use of freely available mobile applications to measure Wi-Fi signal spreading within a residential home as well as to quantify Wi-Fi performance in the presence of some commonly available interferers. A complete analysis of spectrum use is also provided by using a spectrum analyzer. We have found that Wi-Fi throughput degradation due to Bluetooth technology operated in the immediate vicinity of a mobile client is on average 26,5 % and decreases with distance from the Wi-Fi client. The presence of an active video baby monitor leads to a 7,5 % throughput degradation, but its functionality in heavy Wi-Fi traffic environment is however limited. The use of spectrograms in interference analysis is essential for a better understanding of the measured results.
... A few concerns are raised by the move from sub-GHz frequencies to the widely used unlicensed 2.4 GHz ISM band. The interference properties of wireless local area networks (WLAN, also denoted as Wi-Fi) devices operating in this band were characterized in [18]. In another study, coexistence issues with Wi-Fi and LoRa 2.4 GHz were addressed [19]. ...
Full-text available
Recently, Semtech has released a Long Range (LoRa) chipset which operates at the globally available 2.4 GHz frequency band, on top of the existing sub-GHz, km-range offer, enabling hardware manufacturers to design region-independent chipsets. The SX1280 LoRa module promises an ultra-long communication range while withstanding heavy interference in this widely used band. In this paper, we first provide a mathematical description of the physical layer of LoRa in the 2.4 GHz band. Secondly, we investigate the maximum communication range of this technology in three different scenarios. Free space, indoor and urban path loss models are used to simulate the propagation of the 2.4 GHz LoRa modulated signal at different spreading factors and bandwidths. Additionally, we investigate the corresponding data rates. The results show a maximum range of 133 km in free space, 74 m in an indoor office-like environment and 443 m in an outdoor urban context. While a maximum data rate of 253.91 kbit/s can be achieved, the data rate at the longest possible range in every scenario equals 0.595 kbit/s. Due to the configurable bandwidth and lower data rates, LoRa outperforms other technologies in the 2.4 GHz band in terms of communication range. In addition, both communication and localization applications deployed in private LoRa networks can benefit from the increased bandwidth and localization accuracy of this system when compared to public sub-GHz networks.
... In addition, there are currently well-studied methodologies that estimate and optimize wireless communication network parameters, particularly those applied to Wireless Local Area Network (WLAN) [18][19][20]. However, the general objective of this paper is to evaluate and optimize VHTS systems whose link conditions, architecture, and cost structure are very specific and rather different to other wireless communication systems. ...
Full-text available
The concept of geostationary VHTS (Very High Throughput Satellites) is based on multibeam coverage with intensive frequency and polarization reuse, in addition to the use of larger bandwidths in the feeder links, in order to provide high capacity satellite links at a reduced cost per Gbps in orbit. The dimensioning and design of satellite networks based on VHTS imposes the analysis of multiple trade-offs to achieve an optimal solution in terms of cost, capacity, and the figure of merit of the user terminal. In this paper, we propose a new method for sizing VHTS satellite networks based on an analytical expression of the forward link CINR (Carrier-to-Interference-plus-Noise Ratio) that is used to evaluate the trade-off of different combinations of system parameters. The proposed method considers both technical and commercial requirements as inputs, including the constraints to achieve the optimum solution in terms of the user G/T, the number of beams, and the system cost. The cost model includes both satellite and ground segments. Exemplary results are presented with feeder links using Q/V bands, DVB-S2X and transmission methods based on CCM and VCM (Constant and Variable Coding and Modulation, respectively) in two scenarios with different service areas.
Full-text available
The most reliable approach to assess the destabilizing factors that exist in 802.11 wireless channels is through experimental research and the creation of empirical models and methods based on them. The application of such models and methods makes it possible to predict the efficiency of channels for both fixed and mobile subscribers, during the design stages and during connection to the networks. But for a certain type of task, it is relevant to evaluate the impact of a particular destabilizing factor on the basic parameters of a wireless channel. This paper proposes empirical models to evaluate the impact of any type of energy and information destabilizing factor on the performance of an 802.11x wireless channel. The model, which contains two diagnostic parameters, allows to evaluate the energy and information efficiency of the wireless channel under the influence of destabilizing factors. The reliability of such a model is high and depends on the standard regression models of the diagnostic parameters, but there is a considerable period of obtaining the result. To improve performance, a simplified model based on the correlation of basic diagnostic parameters was proposed. To verify the models obtained, mathematical studies were conducted based on the data obtained for interference interference as a destabilizing factor. The impact of such interference on the energy diagnostic parameter is insignificant, unlike the information, where there is a restriction of the physical channel resource distributed between all created radio circuits on the network and between combined and neighboring interference interference. All of this leads to a significant reduction in the effective data rate for a particular radio. It has been found that a model with two diagnostic parameters can have high reliability, which increases with the expansion of the database of experimental studies, both for existing standards and for future modifications. The effect of power and information interference factors on the main channel parameters is different, so the simplified model will have an error of estimation, which will increase as the number of interference, simultaneously active devices in the network and the load in the interference network channels increase.
Full-text available
Sectorized antennas provide an attractive solution to increase wireless network capacity through interference mitigation. Despite their increasing popularity, the real-world performance characteristics of such antennas in dense wireless mesh networks are not well understood. We demonstrate our multi-sector antenna prototypes and their performance through video streaming over an indoor wireless network in the presence of interfering nodes. We use our graphical tool to vary the sender, receiver, and interferer antenna configurations and the resulting performance is directly visible in the video quality displayed at the receiver.
Conference Paper
Full-text available
Some devices not used for data communications radiate in the 2.4 GHz wireless-fidelity (Wi-Fi) band, thus causing unintentional interference that degrades the performance of IEEE 802.11 wireless systems. An analytical model for radio emissions from one of the most common unintentional interferers, the residential microwave oven, is developed from laboratory measurements. Simulation of the analytical model results in a power spectral density and spectrogram that are in good agreement with experimental data. An interference mitigation technique is proposed for the microwave oven emission
Conference Paper
Full-text available
We study the impact on 802.11 networks of RF interference from devices such as Zigbee and cordless phones that increasingly crowd the 2.4GHz ISM band, and from devices such as wireless camera jammers and non-compliant 802.11 devices that seek to disrupt 802.11 operation. Our experiments show that commodity 802.11 equipment is surprisingly vulnerable to certain patterns of weak or narrow-band interference. This enables us to disrupt a link with an interfering signal whose power is 1000 times weaker than the victim's 802.11 signals, or to shut down a multiple AP, multiple channel managed network at a location with a single radio interferer. We identify several factors that lead to these vulnerabilities, ranging from MAC layer driver implementation strategies to PHY layer radio frequency implementation strategies. Our results further show that these factors are not overcome by simply changing 802.11 operational parameters (such as CCA threshold, rate and packet size) with the exception of frequency shifts. This leads us to explore rapid channel hopping as a strategy to withstand RF interference. We prototype a channel hopping design using PRISM NICs, and find that it can sustain throughput at levels of RF interference well above that needed to disrupt unmodified links, and at a reasonable cost in terms of switching overheads.
The emergence of several radio technologies, such as Bluetooth and IEEE 802.11, operating in the 2.4 GHz unlicensed ISM frequency band, may lead to signal interference and result in significant performance degradation when devices are colocated in the same environment. The main goal of this paper is to evaluate the effect of mutual interference on the performance of Bluetooth and IEEE 802.11b systems. We develop a simulation framework for modeling interference based on detailed MAC and PHY models. First, we use a simple simulation scenario to highlight the effects of parameters, such as transmission power, offered load, and traffic type. We then turn to more complex scenarios involving multiple Bluetooth piconets and WLAN devices.
Conference Paper
In the objective of improving the video quality as perceived by end-users, multipath video streaming in Video Distribution Network (VDN) comes as a promising solution. In this paper, we present a new adaptive mechanism to maximize the overall video quality at the client. Overlay path selection is dynamically done based on available bandwidth estimation, while the Quality of Experience (QoE) is subjectively measured using Pseudo-Subjective Quality Assessment (PSQA) tool. Simulation results show that our proposed method can automatically adapt to the load variation on the different Internet paths, in a way that it guarantees a best perceived quality by end-users.
Conference Paper
Multipath video streaming over Video Distribution Network comes as a promising solution to overcome the limitations of the classical single path and IP-level video streaming approaches. For this purpose, we have proposed a dynamic multipath video streaming mechanism, based on quality of experience evaluations and available bandwidth estimations. In this paper, we present the results of performance evaluation when applying our video streaming method. Results show the ability of our mechanism to adapt automatically and in real-time to the load variation on different paths in order to maximize the overall quality perceived by end-users.
Abstract, In the presented paper, two new meth-ods for blind detection and suppression of Bluetooth (BT) interference in case of an OFDM-based transmis-sion system according to the novel WLAN standard IEEE 802.11g are proposed. While the rst method searches for abrupt power changes of the received sym-bols by differentiation in frequency direction, the sec-ond approach makes use of the median ltering, which is well known from the digital image processing. After detection, the in uence of the BT interference is sup-pressed by modifying the values of the affected soft-decision demodulated bits at the input of the channel decoder. Especially for the PSK-modulated modes of IEEE 802.11g, this leads to a signi cant improvement of the frame error rate, as we will show in our simula-tion results.
Conference Paper
In this paper, the performance of IEEE 802.11b and 802.11g wireless local area networks (WLANs) performance under barrage jamming is evaluated theoretically and using measurements. Single WLAN ad-hoc link performance is measured when receiver station is jammed. WLAN link vulnerability assessment is based on throughput and packet error rate (PER), which depends on different physical (PHY) layer characteristic. WLAN system is found to be more tolerant to narrow-band jamming than to wideband-band jamming. In 802.11b WLAN, processing gain increases the tolerance to jamming whereas coding and different modulations are used to improve the link performance in 802.11g WLAN. Measured link throughputs are far from theoretical values, especially when high data rate modes are used.
Conference Paper
With their low cost and high-speed data rate capabilities, IEEE 802.11-based wireless local area networks (WLANs) are growing exponentially. This paper discusses the interference impact from the co-channel interference on current 2.4 GHz WLAN application and addresses several RF design challenges for future 5 GHz frequency band operations. Coverage and migration issues from 2.4 GHz to 5 GHz are also included.