Not All ISPs Equally Secure Home Users
An Empirical Study Comparing Wi-Fi Security Provided by UK ISPs
Z. Cliffe Schreuders and Adil M. Bhat
School of Computing, Creative Technologies and Engineering, Leeds Metropolitan University,
Caedmon Hall, Headingley Campus, Leeds, U.K.
Keywords: Wardriving, Wi-Fi Security, ISPs.
Abstract: A majority of home users rely on their Internet service providers (ISPs) to provide them with wireless
equipment that is secure, and assume that they are appropriately protected from threats such as
piggybacking and eavesdropping. In this paper we present the results of an empirical study comparing the
security provided to home users by their ISPs. Passive wireless data collection was used to gather
information on 7,847 unique wireless access points within Leeds, UK. Non-parametric inferential statistical
analysis was used to compare the security provided by the corresponding ISPs, as identified via the SSID
naming used by ISPs in the UK. The ISPs identified included BT, O2, Orange, Plus Net, Sky, TalkTalk, and
Virgin Media. Statistically significant differences in the security of the networks were found between ISPs,
which we contend can in part be explained by their upgrade policies. These results are contrasted with the
security configuration provided by three of the largest ISPs to new customers. For example, BT (the largest
ISP in the UK) was found to have a greater number of access points configured with the cryptographically
broken Wireless Equivalent Privacy (WEP) encryption method in use, compared to most of the other large
ISPs, and this is in contrast to the favourable security configuration of the routers that are provided to new
customers. The paper concludes with recommendations for when ISPs provide Wi-Fi enabled routers to
home users.
1 INTRODUCTION
Many ISPs make a point of advertising that the Wi-
Fi networks of the routers that they provide to home
users are secure. Furthermore, many home users lack
the expertise to configure their own access points,
and assume that the router provided by their ISP is
adequately secure. While many routers provide
relatively secure configurations at the time of
purchase, over time weaknesses are discovered in
protocols that were previously considered secure.
Encryption methods such as Wireless Equivalent
Privacy (WEP) have long been considered
cryptographically broken (Borisov et al., 2001);
(Fluhrer et al., 2001). WEP was introduced by IEEE
in 1999 to provide data confidentiality and integrity
for 802.11 wireless networks, with the intent that the
security was “equivalent” to wired networks.
However, design flaws were quickly discovered –
such as the small initialisation vector (IV) value,
which results in susceptibility to a Fluhrer, Mantin
and Shamir attack (Fluhrer et al., 2001). Other flaws
in the WEP algorithm make the attacks more
efficient, and even enable real-time decryption
(Bittau et al., 2006); (Stubblefield et al., 2004);
(Tews et al., 2007). These flaws effectively render
WEP networks unsuitable for most security
purposes.
Wi-Fi Protected Access (WPA) provides a more
secure alternative for protecting Wi-Fi networks.
However, weaknesses have been found, such as
weaknesses in the pre-shared key mode with TKIP,
which can result in decryption and injection of
packets (Tews and Beck, 2009). As a consequence
WPA has been deprecated in favour of WPA2.
WPA2 is currently considered the most secure of the
common encryption options for securing 802.11
networks.
Wi-Fi Protected Setup (WPS) is a key
distribution method built into many modern network
devices, which reveals the network key (regardless
of encryption method used) when a client specifies
the correct 8 digit PIN. On some devices, this
request can be made wirelessly. It has recently been
568
Schreuders Z. and Bhat A..
Not All ISPs Equally Secure Home Users - An Empirical Study Comparing Wi-Fi Security Provided by UK ISPs.
DOI: 10.5220/0004600405680573
In Proceedings of the 10th International Conference on Security and Cryptography (SECRYPT-2013), pages 568-573
ISBN: 978-989-8565-73-0
Copyright
c
2013 SCITEPRESS (Science and Technology Publications, Lda.)
reported that in many cases this technology is
enabled by default and is vulnerable to (reduced
search space) online brute-force attacks (Stefan
Viehböck, 2011). This is an example of an attack
that many current routers are vulnerable to and, as
subsequently discussed, the problem is analogous to
the upgrade to more secure encryption methods. It is
common practice in the UK for ISPs to provide
“free” routers with wireless access points (APs) to
customers, included with their subscription. BT is
currently the largest broadband provider in the UK,
with reportedly over 6 million subscribers. Virgin
Media, TalkTalk, and Sky are the next largest, with
approximately 4 million broadband customers each.
Other ISPs that provide Wi-Fi APs include O2,
Orange, and Plusnet. The security of the home
routers provided by these ISPs are explored
throughout this paper.
2 AIMS
This study aimed to identify whether various ISPs
provide different levels of Wi-Fi security to their
home customers, and aimed to identify any
discrepancies between the protection provided. This
research question was evaluated for a relatively large
population of wireless home networks, and in order
to explore the change over time, the default
configuration of routers recently provided by three
of the largest ISPs were also compared in terms of
the level of security provided to home users.
3 METHODS
Data collection was conducted in two stages:
1. Wardriving to collect information on wireless
networks in Leeds, UK as it pertains to the
security provided by ISPs
2. Manual investigation of routers recently
provided by BT, Virgin Media, and Sky
3.1 Wireless Data Collection
Wardriving was conducted using the following
equipment:
9dBi omnidirectional antenna
802.11b/g/n USB adaptor
GPS USB dongle
Laptop running Backtrack 5 R2 with Kismet
Wireless data collection was performed towards the
end of 2012. The antenna was mounted to the roof of
a car, which was driven around high-density
residential areas of Leeds, UK. Specifically
wardriving (wireless data collection from a motor
vehicle) was conducted in select streets in these
areas: Hyde Park, Woodhouse, Headingely, Armley,
Bramley, Beeston, Roundhay, Harehills,
Chapeltown, Hunslet, Kirkstall, and Horsforth.
These areas were chosen as they were expected to
have a high density of home Wi-Fi networks.
Kismet was the software used to log details of
wireless networks. Kismet is a passive network
detection tool. It cycles through Wi-Fi channels
listening to information that is publicly broadcast by
networks, and records information from packets
indicating the presence of access points. Kismet was
configured not to log traffic content (which arguably
would have further ethical ramifications), but to
record high-level details of existing networks, such
as the SSID and security protocol in use.
Ethics approval was granted by the governing
university. Legal precedent seems to suggest that
piggybacking (that is, actually using someone else's
network) without permission is illegal in the UK.
However, the information collected for this study is
broadcast publicly (so no unauthorised access
occurs) and is also routinely collected and stored by
many consumers and businesses; for example, for
location-based services.
After data collection was complete, the data was
exported from Kismet for analysis.
For each network, as determined by the security
flags recorded, a simple security rating score was
assigned, as illustrated in Table 1. This enables a
mean score to be calculated to give an approximate
view of differences, and defines an ordinal scale for
non-parametric data analysis.
Table 1: Security score calculation.
Encryption method Security rating
None 0
WEP 1
WPA 2
WPA2 3
As a passive tool, Kismet records the properties
of each network corresponding to a MAC address. It
is possible for an SSID to appear multiple times due
to multiple devices connecting to the same network.
When this occurred, multiple records where reduced
to one unique record for the network by keeping the
NotAllISPsEquallySecureHomeUsers-AnEmpiricalStudyComparingWi-FiSecurityProvidedbyUKISPs
569
record with the highest level of security: for the case
where a WPA and WPA2 connection were both
found on the same access point. This approach was
taken in order to ensure that each access point was
depicted once and security levels were not under
reported.
The ISP of each connection was established
based on the SSID in use. In the UK, ISPs
consistently name the access points provided to their
customers: for example, an SSID of “SKY84946”
indicates the router was provided to a Sky home
user, while an SSID of “virginmedia8395730” was
provided by Virgin Media. Using this technique, the
ISPs associated with networks were identified.
Open hotspots were pre-filtered out of the data
set based on SSID. Other networks included
manually configured routers, and hidden SSIDs.
These networks were categorised as “other”, and
were excluded from inferential analysis; however,
descriptive statistics were produced.
Finally, inferential statistics were applied to
investigate correlations between ISP and security.
3.2 Manual Router Investigation
Although wardriving provides information about an
extensive number of networks, manually
investigating routers can provide more depth into
details such as the strength of network keys and
default settings. For this reason, routers from three
of the largest ISPs, which had been provided to
home users within the last year, were analysed in
terms of their security properties. Analysis was
based on a convenience sample of routers from BT,
Virgin Media, and Sky. This information illustrated
notable differences in the approach of the various
ISPs and, as discussed herein, was in contrast with
the results based on wireless data collection.
4 RESULTS
4.1 Analysis of Wireless Data
Just over 10,000 networks were recorded during
wardriving. After filtering of hotspots and removal
of duplicate SSIDs 7,847 networks remained. Of
these, 5,158 were identified as being associated with
a specific ISP. The ISPs identified were: BT, O2,
Orange, Plus Net, Sky, TalkTalk, and Virgin Media.
Initial descriptive analysis is presented in Table
2. The table shows the number of wireless networks
of each ISP, and the number and proportion of each
encryption method found in use on access points
provided by the ISP.
Table 2: ISPs and use of encryption methods.
None WEP WPA WPA2 Total
BT
8
(1%)
116
(13.8%)
12
(1.4%)
706
(83.8%)
842
O2
0
(0%)
112
(38.6%)
0
(0%)
178
(61.4%)
290
Orange
0
(0%)
4
(9.1%)
0
(0%)
40
(90.9%)
44
Other
120
(4.5%)
277
(10.3%)
719
(26.7%)
1573
(58.5%)
2689
Plusnet
1
(0.8%)
1
(0.8%)
0
(0%)
116
(98.3%)
118
Sky
1
(0.1%)
23
(1.7%)
398
(29.3%)
936
(68.9%)
1358
Talk-
Talk
0
(0%)
6
(1.2%)
4
(0.8%)
510
(98.1%)
520
Virgin
Media
2
(0.1%)
9
(0.4%)
11
(0.5%)
1964
(98.9%)
1986
The mean of the scores for each ISP is illustrated
in Figure 1, and are as follows: BT=2.68, O2=2.23,
Orange=2.82, Other=2.39, Plusnet=2.96, Sky=2.67,
TalkTalk=2.97, and Virgin Media=2.98. Although
this view of the data is an approximation (being
means of ordinal data), it illustrates that differences
appear to exist between ISPs.
Non-parametric statistical tests were applied to
investigate whether the ISP has a significant effect
of the level of security of home users.
A Kruskal-Wallis H test was conducted to
compare the effect of ISP on Wi-Fi security rating.
There was a statistically significant effect of the ISP
(H(6)=827.211, p < 0.001). It can be concluded that
there is a difference in wireless security correlated
with the ISPs.
Post-hoc analysis was conducted using the
Mann-Whitney U test with Bonferonni adjustment
applied. In each pairwise comparison between three
of the largest ISPs (BT, Virgin Media, and Sky) the
differences were found to be statistically significant
(p < 0.001). That is, all the three ISPs were
significantly different from each other in terms of
wireless security. Pairwise comparison was
performed between each ISP, and the other
significant results include:
O2 was found to provide statistically significant
lower levels of security when compared with all
the other ISPs. WEP usage was of the highest
proportion, compared to other ISPs.
BT was found to provide significantly different
levels of security when compared to all the ISPs,
except Orange (likely due to the small sample of
Orange networks). BT had notably higher
SECRYPT2013-InternationalConferenceonSecurityandCryptography
570
Figure 1: Mean security rating scores by ISP.
proportions of WEP networks compared to
others, except in comparison to O2. Means and
proportions are previously reported.
Sky was also found to provide statistically
different security in comparison with all the
other ISPs, except Orange. Although Sky had a
lower WEP proportion than some others, WPA
(rather than WPA2) usage was higher than the
other ISPs.
Virgin Media, TalkTalk, and Plusnet were
statistically different to O2, BT, and Sky, as
mentioned above. Comparison with other ISPs
did not indicate significant difference of the
security.
Orange (with a small sample size) was
significantly different only to O2.
4.2 Analysis of Routers
Recent routers provided to UK customers of BT,
Virgin Media, and Sky were investigated. A
summary of the findings is presented in Table 3.
5 DISCUSSION
It is commendable that, across all the ISP provided
routers that were identified, WPA2 adoption was
quite high. Also, amongst the ISPs studied in greater
detail, new customers were typically provided with
access points which by default are configured for a
WPA/WPA2 mixed mode, which is generally
considered to provide security adequate for home
use.
However, in the population studied, there was
still a disturbing number of WEP networks in use on
routers provided by ISPs, and even a number of
networks with no encryption in use. Many of the
three largest ISPs' (BT, Virgin Media, and Sky)
routers included home networks with no encryption.
These networks are very few in number and
represent less than 1% of the total analysed
networks. However, if this is representative of the
millions of home Wi-Fi networks, this could be
considered to represent a large total number of users.
One explanation for home routers configured this
way is deliberate configuration by end users to open
their networks to others.
Table 3: Standard router configurations for BT, Virgin
Media and Sky.
BT Virgin Media SKY
Encryption
options
WEP, WPA,
WPA2,
WPA/WPA2
(mixed mode)
WEP, WPA,
WPA2,
WPA/WPA2
(mixed
mode),
WPA/WPA2-
Enterprise
(802.1X)
WEP, WPA,
WPA2,
WPA/WPA2
(mixed
mode), WPA
Enterprise
(802.1X)
Default
encryption
WPA/ WPA2
Mixed mode
WPA/WPA2
Mixed mode
WPA/WPA2
Mixed mode
Authentication
option available
PSK PSK PSK
WPS-PIN Yes Yes Yes
Default network
key
Alphanumeric
Alphabetical
(Lower case
only)
Alphabetical
(Uppercase
only)
Default network
key length
10-digit 8-digit 8-digit
Key size 62
10
26
8
26
8
Default router
administration
password
Alphanumeric “changeme” “sky”
Of the routers provided by ISPs, over 5% (271)
NotAllISPsEquallySecureHomeUsers-AnEmpiricalStudyComparingWi-FiSecurityProvidedbyUKISPs
571
were configured to use WEP. The distribution
clearly varies between ISPs, with some ISPs with
substantially higher proportions of WEP networks
than others.
Data analysis showed that the ISP has a
significant effect on the level of security on their
users' home networks. Which raises the question:
should ISPs be considered responsible for the Wi-Fi
security of their customers? It could be argued that
ISPs have a “duty of care” when they provide
routers with wireless access points, given that ISPs
often advertise that they provide secure networks,
and the fact that many home users are unlikely to
reconfigure the routers provided to them.
In many cases ISPs were found to have
significantly different distributions of encryption
employed by their routers. Some ISPs appear to have
kept their users' security more up-to-date than
others. This may be of interest to consumers who
could be determining their own future security status
when choosing an ISP.
Of the largest ISPs (BT, Virgin Media, TalkTalk,
and Sky), BT was found to have the highest
proportion of routers configured to use WEP
encryption, at 13.8%. If the sample in this study was
representative of the population at large, BT's
customer base could have upwards of 800,000 home
users using WEP in the UK. This could be
considered cause for concern, as these networks
would be vulnerable to easy attack, and could be
targeted for eavesdropping, piggybacking, and
various other threats.
This is in contrast to the findings from the
analysis of recently provided routers, which
illustrates that BT does seem to provide
comparatively well configured routers to new users.
BT home networks using the “BT Home Hub”
1.0 and 1.5 were all using WEP encryption. These
networks were identifiable based on their SSID,
starting with “BTHomeHub-” followed by an
alphanumeric pseudo-random string: for example,
“BTHomeHub-7AFC”. Other BT networks, using
newer routers (SSID names starting with
“BTHomeHub2-”, “BTHomeHub3-” and “BTHub3-
”) were found to be using WPA2.
This suggests that many of the insecurely
configured routers were installed some time ago, and
have yet to be updated with more modern encryption
methods. We contend that this is likely due to the
upgrade policies of the ISPs.
The recent vulnerabilities discovered in WPS
authentication, as described in the introduction,
illustrates the ongoing importance of ISP response
time to Wi-Fi security threats, and applies to many
of the routers provided by ISPs. The way that ISPs
react to these issues is expected to have a significant
effect on the ongoing security posture of their
customers.
6 RECOMMENDATIONS
Based on these findings, when ISPs provide routers
to their customers, the following recommendations
are offered:
ISPs should make an effort to track which model
of router and the version of firmware their
customers are using, and automatically push
updates and upgrades to users
ISPs should proactively upgrade their users'
routers (software and/or hardware) as soon as
possible after critical security vulnerabilities are
discovered and fixes are available – or at least
contact customers and educate them on risks and
provide update options
When fixes are not yet available (as may
currently be the case with some WPS
configurations), ISPs should inform customers of
the threats they face and estimate time until
solutions will be available
WEP should be removed as an option from new
routers, or at least trigger obvious and
informative alerts to users if they choose to
switch to this non-default configuration
Routers should come securely pre-configured,
including adequately pseudo-random network
keys and web administration passwords (rather
than default passwords such as Sky's “sky” or
Virgin Media's “changeme”)
Unless it has been demonstrated to be
significantly less usable, as far as is practical,
larger key spaces should be employed for pre-
configured passwords: for example, by using
alphanumeric rather than alphabetical passwords
Other ISP specific recommendations:
BT should consider ways to improve the
security posture of customers currently using
BT Home Hub version 1 or 1.5
O2 should investigate and act upon their
customers' routers employing WEP
Users should be educated regarding maintaining
the security of their networks
SECRYPT2013-InternationalConferenceonSecurityandCryptography
572
7 RELATED RESEARCH
Various small scale wardriving efforts have been
conducted in Leeds, UK: such as a report from 2004
of 66 networks, all reported to be using either WEP
or no encryption (Dlaverty, 2004).
Some other independent work has also analysed
and critiqued the security of the routers provided by
UK ISPs to their customers. For example, the
security of the BT Home Hub router has received
extensive criticism, not only for the choice of
encryption and key length (as discussed herein), but
also for insufficient entropy of pseudo-random
passwords, vulnerabilities in the web interface, and
open ports for management services (Adrian Pastor,
2007). Problems have also been discovered with
Sky's pseudo-random passwords (in this case with
their older Netgear v2 DG934g routers), passwords
can be deduced based on the (public) MAC address
(John Leyden, 2008). These routers are also
vulnerable to an attack that can determine the ADSL
password, when the username is known
(NewsreadeR, 2008).
As far as we are aware, this is the first empirical
study to investigate correlations between security
and ISPs, and how ISPs differ from each other in
terms of the types of security provided to their users.
8 CONCLUSIONS
Analysis of data collected via wardriving in Leeds,
UK, has shown a statistically significant effect on
Wi-Fi security by ISPs, and significant differences
between many individual ISPs. A number of
networks were found to be using WEP, despite this
being known to be a cryptographically broken
encryption method, and these routers were provided
by identifiable ISPs, who are in a position to be able
to keep track of out-of-date routers. We contend that
this highlights the importance of router upgrades,
and have provided a number of recommendations for
ISPs, router manufacturers, and home users that
apply when ISPs provide routers with wireless
access points to customers.
The question of duty of care was raised: should
ISPs be considered responsible for the Wi-Fi
security of their customers when they provide
routers with wireless access points, given that ISPs
often advertise that they provide secure networks,
and many home users are unlikely to reconfigure the
routers provided to them?
REFERENCES
Adrian Pastor, 2007. BT home flub: pwnin the BT Home
Hub [WWW Document]. GNUCITIZEN. URL
http://www.gnucitizen.org/blog/bt-home-flub-pwnin-
the-bt-home-hub/ (accessed 10.28.12).
Bittau, A., Handley, M., Lackey, J., 2006. The final nail in
WEP’s coffin, in: Proceedings of the 2006 IEEE
Symposium on Security and Privacy, SP ’06. IEEE
Computer Society, Washington, DC, USA, pp. 386–
400.
Borisov, N., Goldberg, I., Wagner, D., 2001. Intercepting
mobile communications: the insecurity of 802.11, in:
Proceedings of the 7th Annual International
Conference on Mobile Computing and Networking,
MobiCom ’01. ACM, New York, NY, USA, pp. 180–
189.
Dlaverty, 2004. Open all hours - Wardriving in Leeds,
West Yorkshire, England [WWW Document].
Openxtra. URL http://www.openxtra.co.uk/articles/
wardriving-leeds (accessed 10.28.12).
Fluhrer, S., Mantin, I., Shamir, A., 2001. Weaknesses in
the key scheduling algorithm of RC4, in: Vaudenay,
S., Youssef, A. (Eds.), Selected Areas in
Cryptography, Lecture Notes in Computer Science.
Springer Berlin / Heidelberg, pp. 1–24.
John Leyden, 2008. Sky Broadband puts the fault into
default Wi-Fi security: Users in guess-able random
keys quandary [WWW Document]. The Register.
URL http://www.theregister.co.uk/2008/02/21/sky_
broadband_wi_fi_keys_unpicked/ (accessed 10.28.12).
NewsreadeR, 2008. Is your router secure? [WWW
Document]. Sky User. URL http://www.skyuser.co.uk/
skyinfo/783.html (accessed 10.28.12).
Stefan Viehböck, 2011. Brute forcing Wi-Fi Protected
Setup: When poor design meets poor implementation
[WWW Document]. URL http://packetstorm.foofus.
com/papers/wireless/viehboeck_wps.pdf
Stubblefield, A., Ioannidis, J., Rubin, A.D., 2004. A key
recovery attack on the 802.11b wired equivalent
privacy protocol (WEP). ACM Trans. Inf. Syst. Secur.
7, 319–332.
Tews, E., Beck, M., 2009. Practical attacks against WEP
and WPA, in: Proceedings of the Second ACM
Conference on Wireless Network Security, WiSec ’09.
ACM, New York, NY, USA, pp. 79–86.
Tews, E., Weinmann, R.-P., Pyshkin, A., 2007. Breaking
104 Bit WEP in less than 60 seconds, in: Proceedings
of the 8th International Conference on Information
Security Applications, WISA’07. Springer-Verlag,
Berlin, Heidelberg, pp. 188–202.
NotAllISPsEquallySecureHomeUsers-AnEmpiricalStudyComparingWi-FiSecurityProvidedbyUKISPs
573