Key Words
IEEE 802.11, wireless spoofing, cracking WEP, forged
Deauthentication, rogue/ Trojan access points, session hijacking, war driving.
Abstract
This article describes IEEE 802.11-specific hacking
techniques that attackers have used, and suggests various defensive measures.
We describe sniffing, spoofing and probing in the context of wireless
networks. We describe how SSIDs can be
determined, how a sufficiently large number of frames can be collected so that
WEP can be cracked. We show how easy it
is to cause denial-of-service through jamming and through forged
disassociations and deauthentications.
We also explain three man-in-the-middle attacks using wireless
networks. We give a list of selected open-source
tools. We summarize the activity known
as war driving. We conclude the article
with several recommendations that will help improve security at a wireless
deployment site.
1. Introduction
Wireless networks broadcast their packets using radio
frequency or optical wavelengths. A modern laptop computer can listen
in. Worse, an attacker can manufacture new packets on the fly and persuade
wireless stations to accept his packets as legitimate.
We use the term hacking as described below.
hacker n. [originally, someone who makes
furniture with an axe] 1. A person who enjoys exploring the details
of programmable systems and how to stretch their capabilities, as opposed to
most users, who prefer to learn only the minimum necessary. 2. One
who programs enthusiastically (even obsessively) or who enjoys programming
rather than just theorizing about programming. 3. A person capable
of appreciating hack value. 4. A person who is good at programming
quickly. 5. An expert at a particular program, or one who
frequently does work using it or on it; as in `a Unix hacker'. (Definitions 1
through 5 are correlated, and people who fit them congregate.) 6. An
expert or enthusiast of any kind. One might be an astronomy hacker, for
example. 7. One who enjoys the intellectual challenge of creatively
overcoming or circumventing limitations. 8. [deprecated] A
malicious meddler who tries to discover sensitive information by poking around.
Hence `password hacker', `network hacker'. The correct term for this sense is cracker.
From The Jargon
Dictionary http://info.astrian.net/jargon/
This article describes IEEE 802.11-specific hacking
techniques that attackers have used, and suggests various defensive measures.
It is not an overview of security features proposed in WPA or IEEE
802.11i. We do not consider legal
implications, or the intent behind such hacking, whether malevolent or
benevolent. The article’s focus is in
describing techniques, methods, analyses
and uses in ways unintended by the designers of IEEE 802.11.
2. Wireless LAN Overview
In this section, we give a brief overview of wireless LAN
(WLAN) while emphasizing the features that help an attacker. We assume
that the reader is familiar with the TCP/IP suite (see, e.g., [Mateti 2003]).
IEEE 802.11 refers to a family of specifications (www.ieee802.org/11/) developed by the
IEEE for over-the-air interface between a wireless client and an AP or
between two wireless clients. To be called 802.11 devices, they must
conform to the Medium Access Control (MAC) and Physical Layer specifications. The
IEEE 802.11 standard covers the Physical (Layer 1) and Data Link (Layer 2)
layers of the OSI Model. In this
article, we are mainly concerned with the MAC layer and not the variations of
the physical layer known as 802.11a/b/g.
2.1 Stations and Access Points
A wireless network interface card (adapter) is a device,
called a station, providing the network physical layer over a radio
link to another station. An access
point (AP) is a station that provides frame distribution service to stations
associated with it. The AP itself is typically connected by wire to
a LAN.
The station and AP each contain a network interface
that has a Media Access Control (MAC) address, just as wired network cards do.
This address is a world-wide-unique 48-bit number, assigned to it at the time
of manufacture. The 48-bit address is often represented as a string of six
octets separated by colons (e.g., 00:02:2D:17:B9:E8) or hyphens
(e.g., 00-02-2D-17-B9-E8)
. While the MAC address as
assigned by the manufacturer is printed on the device, the address can be
changed in software.
Each AP has a 0 to 32 byte long Service Set Identifier
(SSID) that is also commonly called a network name. The SSID is used to
segment the airwaves for usage. If two wireless networks are physically close,
the SSIDs label the respective networks, and allow the components of one
network to ignore those of the other. SSIDs can also be mapped to virtual LANs;
thus, some APs support multiple SSIDs. Unlike
fully qualified host names (e.g., gamma.cs.wright.edu), SSIDs are not
registered, and it is possible that two unrelated networks use the same
SSID.
2.2 Channels
The stations
communicate with each other using radio frequencies between 2.4 GHz and 2.5
GHz. Neighboring channels are only 5 MHz apart. Two wireless networks using neighboring
channels may interfere with each other.
2.3 WEP
Wired Equivalent
Privacy (WEP) is a shared-secret key encryption system used to
encrypt packets transmitted between a station and an AP. The WEP
algorithm is intended to protect wireless communication from eavesdropping. A
secondary function of WEP is to prevent unauthorized access to a wireless
network. WEP encrypts the payload of data packets. Management and
control frames are always transmitted in the clear. WEP uses the RC4 encryption algorithm.
The shared-secret key is either 40 or 104 bits long. The key is
chosen by the system administrator. This key must be shared among
all the stations and the AP using mechanisms that are not specified in the IEEE
802.11.
2.4 Infrastructure and Ad Hoc Modes
A
wireless network operates in one of two modes. In the ad hoc mode,
each station is a peer to the other stations and communicates directly with
other stations within the network. No AP is involved. All stations
can send Beacon and Probe frames. The ad hoc mode stations form an Independent
Basic Service Set (IBSS).
A station
in the infrastructure mode communicates only with an AP. Basic Service
Set (BSS) is a set of stations that are logically associated with each other
and controlled by a single AP. Together they operate as a fully connected
wireless network. The BSSID is a 48-bit
number of the same format as a MAC address. This field uniquely identifies each
BSS. The value of this field is the MAC address of the AP.
2.5 Frames
Both the station and AP radiate and gather 802.11 frames as
needed. The format of frames is illustrated below. Most of the frames
contain IP packets. The other frames are for the management and control
of the wireless connection.
Figure 1 An IEEE 802.11 Frame
There are three classes of frames. The management
frames establish and maintain communications. These are of Association
request, Association response, Reassociation request, Reassociation response,
Probe request, Probe response, Beacon, Announcement traffic indication message,
Disassociation, Authentication, Deauthentication types. The SSID is part
of several of the management frames. Management messages are always sent in the
clear, even when link encryption (WEP or WPA) is used, so the SSID is visible
to anyone who can intercept these frames.
The control frames help in the delivery of data.
The data frames encapsulate the OSI Network Layer
packets. These contain the source and
destination MAC address, the BSSID, and the TCP/IP datagram. The payload part of the datagram is
WEP-encrypted.
2.6 Authentication
Authentication is the process of proving identity of a
station to another station or AP. In the open system authentication, all stations
are authenticated without any checking. A station A sends an
Authentication management frame that contains the identity of A, to station
B. Station B replies with a frame that indicates recognition, addressed
to A. In the closed network architecture, the stations must know the
SSID of the AP in order to connect to the AP. The shared key
authentication uses a standard challenge and response along with a shared
secret key.
Figure 2: States and Services
2.7 Association
Data can be exchanged between the station and AP only after a
station is associated with an AP in the infrastructure mode or with another station
in the ad hoc mode. All the APs transmit Beacon frames a few times each
second that contain the SSID, time, capabilities, supported rates, and other information.
Stations can chose to associate with an AP based on the signal strength
etc. of each AP. Stations can have a null SSID that is considered
to match all SSIDs.
The association is a two-step process. A station that is
currently unauthenticated and unassociated listens for Beacon frames. The station
selects a BSS to join. The station and the AP mutually authenticate themselves
by exchanging Authentication management frames. The client is now
authenticated, but unassociated. In the
second step, the station sends an Association Request frame, to which the AP
responds with an Association Response frame that includes an Association ID to
the station. The station is now authenticated and associated.
A station can be authenticated with several APs at the same
time, but associated with at most one AP at any time. Association implies
authentication. There is no state where a station is associated but not
authenticated.
3. Wireless Network Sniffing
Sniffing is eavesdropping on the network. A (packet) sniffer is a program that
intercepts and decodes network traffic broadcast through a medium. Sniffing is the act by a machine S of making
copies of a network packet sent by machine A intended to be received by machine
B. Such sniffing, strictly speaking, is not a TCP/IP problem, but it is
enabled by the choice of broadcast media, Ethernet and 802.11, as the physical
and data link layers.
Sniffing has long been a reconnaissance technique used in
wired networks. Attackers sniff the frames necessary to enable the exploits
described in later sections. Sniffing is the underlying technique used in
tools that monitor the health of a network.
Sniffing can also help find the easy kill as in scanning for open access
points that allow anyone to connect, or capturing the passwords used in a
connection session that does not even use WEP, or in telnet, rlogin and ftp
connections.
It is easier to sniff wireless networks than wired ones. It
is easy to sniff the wireless traffic of a building by setting shop in a car
parked in a lot as far away as a mile, or while driving around the block. In a
wired network, the attacker must find a way to install a sniffer on one or more
of the hosts in the targeted subnet. Depending on the equipment used in a
LAN, a sniffer needs to be run either on the victim machine whose traffic is of
interest or on some other host in the same subnet as the victim. An
attacker at large on the Internet has other techniques that make it possible to
install a sniffer remotely on the victim machine.
3.1 Passive Scanning
Scanning is the act of sniffing by tuning to various radio
channels of the devices. A passive network scanner instructs the
wireless card to listen to each channel for a few messages. This does not
reveal the presence of the scanner.
An attacker can passively scan without transmitting at
all. Several modes of a station permit this. There is a mode called RF
monitor mode that allows every frame appearing on a channel to be
copied as the radio of the station tunes to various channels. This is
analogous to placing a wired Ethernet card in promiscuous mode. This mode is
not enabled by default. Some wireless cards on the market today have
disabled this feature in the default firmware. One can buy wireless cards
whose firmware and corresponding driver software together permit reading of all
raw 802.11 frames. A station in monitor mode can capture
packets without associating with an AP or ad-hoc network. The so-called promiscuous
mode allows the capture of all wireless packets of an associated network. In
this mode, packets cannot be read until authentication and association are
completed.
An example sniffer is Kismet (http://www.kismetwireless.net).
An example wireless card that permits RF monitor modes is Cisco Aironet AIR-PCM342.
3.2 Detection of SSID
The attacker can discover the SSID of a network usually by
passive scanning because the SSID occurs in the following frame types: Beacon,
Probe Requests, Probe Responses, Association Requests, and Reassociation
Requests. Recall that management frames are always in the clear, even when WEP
is enabled.
On a number of APs, it is possible to configure so that the
SSID transmitted in the Beacon frames is masked, or even turn off Beacons
altogether. The SSID shown in the Beacon frames is set to null in the
hope of making the WLAN invisible unless a client already knows the correct
SSID. In such a case, a station wishing to join a WLAN begins the
association process by sending Probe Requests since it could not detect any APs
via Beacons that match its SSID.
If the Beacons are not turned off, and the SSID in them is
not set to null, an attacker obtains the SSID included in the Beacon frame by
passive scanning.
When the Beacon displays a null SSID, there are two possibilities.
Eventually, an Associate Request may appear from a legitimate station that
already has a correct SSID. To such a request, there will be an Associate
Response frame from the AP. Both frames will contain the SSID in the
clear, and the attacker sniffs these. If the station wishes to join any
available AP, it sends Probe Requests on all channels, and listens for Probe
Responses that contain the SSIDs of the APs. The station considers all
Probe Responses, just as it would have with the non-empty SSID Beacon frames,
to select an AP. Normal association then begins. The attacker waits to
sniff these Probe Responses and extract the SSIDs.
If Beacon transmission is disabled, the attacker has
two choices. The attacker can keep sniffing waiting for a voluntary
Associate Request to appear from a legitimate station that already has a
correct SSID and sniff the SSID as described above. The attacker can also
chose to actively probe by injecting frames that he constructs, and then sniffs
the response as described in a later section.
When the above methods fail, SSID discovery is done by active
scanning (see Section 5).
3.3 Collecting the MAC Addresses
The attacker gathers legitimate MAC addresses for use later
in constructing spoofed frames. The source and destination MAC addresses are
always in the clear in all the frames. There are two reasons why an
attacker would collect MAC addresses of stations and APs participating in a
wireless network. (1) The attacker wishes to use these values in spoofed
frames so that his station or AP is not identified. (2) The targeted AP may be
controlling access by filtering out frames with MAC addresses that were not
registered.
3.4 Collecting the Frames for Cracking WEP
The goal of an attacker is to discover the WEP shared-secret
key. Often, the shared key can be discovered by guesswork based on a
certain amount of social engineering regarding the administrator who configures
the wireless LAN and all its users. Some client software stores the WEP
keys in the operating system registry or initialization scripts. In the
following, we assume that the attacker was unsuccessful in obtaining the key in
this manner. The attacker then employs systematic procedures in cracking
the WEP. For this purpose, a large number (millions) of frames need to be
collected because of the way WEP works.
The wireless device generates on the fly an Initialization
Vector (IV) of 24-bits. Adding these bits to the shared-secret key of
either 40 or 104 bits, we often speak of 64-, or 128-bit encryption. WEP
generates a pseudo-random key stream from the shared secret key and the IV. The
CRC-32 checksum of the plain text, known as the Integrity Check (IC) field, is
appended to the data to be sent. It is then exclusive-ORed with the
pseudo-random key stream to produce the cipher text. The IV is
appended in the clear to the cipher text and transmitted. The receiver extracts
the IV, uses the secret key to re-generate the random key stream, and
exclusive-ORs the received cipher text to yield the original plaintext.
Certain cards are so simplistic that they start their IV as
0 and increment it by 1 for each frame, resetting in between for some events. Even the better cards generate weak IVs from
which the first few bytes of the shared key can be computed after statistical
analyses. Some implementations generate
fewer mathematically weak vectors than others do.
The attacker sniffs a large number of frames from a single
BSS. These frames all use the same key. The mathematics behind the
systematic computation of the secret shared key from a collection of cipher
text extracted from these frames is described elsewhere in this
volume. What is needed however is a collection of frames that were
encrypted using “mathematically-weak” IVs. The number of encrypted frames that
were mathematically weak is a small percentage of all frames. In a
collection of a million frames, there may only be a hundred mathematically weak
frames. It is conceivable that the collection may take a few hours to
several days depending on how busy the WLAN is.
Given a sufficient number of mathematically weak frames, the
systematic computation that exposes the bytes of the secret key is
intensive. However, an attacker can employ powerful computers. On
an average PC, this may take a few seconds to hours. The storage of the
large numbers of frames is in the several hundred-mega bytes to a few giga
bytes range.
An example of a WEP cracking tool is AirSnort ( http://airsnort.shmoo.com ).
3.5 Detection of the Sniffers
Detecting the presence of a wireless sniffer, who remains
radio-silent, through network security measures is virtually impossible.
Once the attacker begins probing (i.e., by injecting packets), the presence and
the coordinates of the wireless device can be detected.
4. Wireless Spoofing
There are well-known attack techniques known as spoofing in both
wired and wireless networks. The attacker constructs frames by filling
selected fields that contain addresses or identifiers with legitimate looking
but non-existent values, or with values that belong to others. The
attacker would have collected these legitimate values through sniffing.
4.1 MAC Address Spoofing
The attacker generally desires to be hidden. But the
probing activity injects frames that are observable by system
administrators. The attacker fills the Sender MAC Address field of the
injected frames with a spoofed value so that his equipment is not identified.
Typical APs control access by permitting only those stations
with known MAC addresses. Either the attacker has to compromise a
computer system that has a station, or he spoofs with legitimate MAC addresses
in frames that he manufactures. MAC addresses are assigned at the time of
manufacture, but setting the MAC address of a wireless card or AP to an
arbitrary chosen value is a simple matter of invoking an appropriate software
tool that engages in a dialog with the user and accepts values. Such
tools are routinely included when a station or AP is purchased. The
attacker, however, changes the MAC address programmatically, sends several
frames with that address, and repeats this with another MAC address. In a
period of a second, this can happen several thousand times.
When an AP is not filtering MAC addresses, there is no need
for the attacker to use legitimate MAC addresses. However, in
certain attacks, the attacker needs to have a large number of MAC addresses
than he could collect by sniffing. Random MAC addresses are
generated. However, not every random sequence of six bytes is a MAC
address. The IEEE assigns globally the first three bytes, and the
manufacturer chooses the last three bytes. The officially assigned
numbers are publicly available. The attacker generates a random MAC
address by selecting an IEEE-assigned three bytes appended with an additional
three random bytes.
4.2 IP spoofing
Replacing the true IP address of the sender (or, in
rare cases, the destination) with a different address is known as IP
spoofing. This is a necessary operation in many attacks.
The IP layer of the OS simply trusts that the source
address, as it appears in an IP packet is valid. It assumes that the
packet it received indeed was sent by the host officially assigned that source
address. Because the IP layer of the OS normally adds these IP addresses to a data
packet, a spoofer must circumvent the IP layer and talk directly to the raw
network device. Note that the attacker’s machine cannot simply be
assigned the IP address of another host X using
ifconfig
or a similar configuration tool. Other hosts, as
well as X, will discover (through ARP, for example) that there are two machines
with the same IP address.
IP spoofing is an integral part of many attacks. For
example, an attacker can silence a host A from sending further packets to B by
sending a spoofed packet announcing a window size of zero to A as though it
originated from B.
4.3 Frame Spoofing
The attacker will inject frames that are valid by 802.11
specifications, but whose content is carefully spoofed as described above.
Frames themselves are not authenticated in 802.11
networks. So when a frame has a spoofed source address, it cannot be
detected unless the address is wholly bogus. If the frame to
be spoofed is a management or control frame, there is no encryption to deal
with. If it is a data frame, perhaps as part of an on-going MITM attack,
the data payload must be properly encrypted.
Construction of the byte stream that constitutes a spoofed
frame is a programming matter once the attacker has gathered the needed
information through sniffing and probing. There are software
libraries that ease this task. Examples of such libraries are
libpcap
(sourceforge.net/projects/libpcap/),
libnet
(libnet.sourceforge.net/), libdnet
(libdnet. sourceforge.net/)
and libradiate
(www.packetfactory.net/projects/libradiate/
).
The difficulty here is not in the construction of the
contents of the frame, but in getting, it radiated (transmitted) by the station
or an AP. This requires control over the firmware and driver of the
wireless card that may sanitize certain fields of a frame. Therefore, the
attacker selects his equipment carefully. Currently, there are
off-the-shelf wireless cards that can be manipulated. In addition, the
construction of special purpose wireless cards is within the reach of a
resourceful attacker.
5. Wireless Network Probing
Even though the attacker gathers considerable amount of
information regarding a wireless network through sniffing, without revealing
his wireless presence at all, there are pieces that may still be missing.
The attacker then sends artificially constructed packets to a target that
trigger useful responses. This activity is known as probing or active
scanning.
The target may discover that it is being probed, it might
even be a honey pot (www.honeynet.org/)
target carefully constructed to trap the attacker. The attacker would try
to minimize this risk.
5.1 Detection of SSID
Detection of SSID is often possible by simply sniffing
Beacon frames as describe in a previous section.
If Beacon transmission is disabled, and the
attacker does not wish to patiently wait for a voluntary Associate Request to
appear from a legitimate station that already has a correct SSID, or Probe
Requests from legitimate stations, he will resort to probing by injecting a
Probe Request frame that contains a spoofed source MAC address. The Probe
Response frame from the APs will contain, in the clear, the SSID and other
information similar to that in the Beacon frames were they enabled. The
attacker sniffs these Probe Responses and extracts the SSIDs.
Some models of APs have an option to disable responding to Probe
Requests that do not contain the correct SSID. In this case, the attacker
determines a station associated with the AP, and sends the station a forged
Disassociation frame where the source MAC address is set to that of the
AP. The station will send a
Reassociation Request that exposes the SSID.
5.2 Detection of APs and stations
Every AP is a station, so SSIDs, MAC addresses are gathered
as described above.
Certain bits in the frames identify that the frame is from
an AP. If we assume that WEP is either disabled or cracked, the attacker
can also gather the IP addresses of the AP and the stations.
5.3 Detection of Probing
Detection of probing is possible. The frames that an
attacker injects can also be heard by the intrusion detection systems (IDS) of
hardened wireless LAN. There is GPS-enabled equipment that can identify
the physical coordinates of a wireless device through which the probe frames
are being transmitted.
6. AP Weaknesses
APs have weaknesses that are both due to design mistakes and
user interfaces that promote weak passwords, etc. It has been
demonstrated by many publicly conducted war-driving efforts (www.worldwidewardrive.org) in
major cities around the world that a large majority of the deployed APs are
poorly configured, most with WEP disabled, and configuration defaults, as set
up the manufacturer, untouched.
6.1 Configuration
The default WEP keys used are often too trivial. Different
APs use different techniques to convert the user’s key board input into a bit
vector. Usually 5 or 13 ASCII printable characters are directly mapped by
concatenating their ASCII 8-bit codes into a 40-bit or 104-bit WEP key. A
stronger key can be constructed from an input of 26 hexadecimal digits. It is
possible to form an even stronger104 bit WEP key by truncating the MD5 hash of
an arbitrary length pass phrase.
6.2 Defeating MAC Filtering
Typical APs permit access to only those stations with known
MAC addresses. This is easily defeated by the attacker who spoofs his
frames with a MAC address that is registered with the AP from among the ones
that he collected through sniffing. That a MAC address is registered can
be detected by observing the frames from the AP to the stations.
6.3 Rogue AP
Access points that are installed without proper
authorization and verification that overall security policy is obeyed are
called rogue APs. These are installed and used by valid
users. Such APs are configured poorly, and attackers will find them.
6.4 Trojan AP
An attacker sets up an AP so that the targeted station
receives a stronger signal from it than what it receives from a legitimate
AP. If WEP is enabled, the attacker would have already cracked it.
A legitimate user selects the Trojan AP because of the stronger signal,
authenticates and associates. The Trojan AP is connected to a system that
collects the IP traffic for later analyses. It then transmits all the
frames to a legitimate AP so that the victim user does not recognize the
on-going MITM attack. The attacker can steal the users password, network
access, compromise the user’s system to give himself root access. This
attack is called the Evil Twin Attack.
It is easy to build a Trojan AP because an AP is a computer
system optimized for its intended application. A general purpose PC with
a wireless card can be turned into a capable AP. An example of such
software is HostAP (http://hostap.epitest.fi/
). Such a Trojaned AP would be formidable.
6.5 Equipment Flaws
A search on www.securityfocus.com
with “access point vulnerabilities” will show that numerous flaws in equipment
from well-known manufacturers are known. For example, one such AP crashes
when a frame is sent to it that has the spoofed source MAC address of
itself. Another AP features an embedded TFTP (Trivial File Transfer
Protocol) server. By requesting a file named
config.img
via TFTP, an attacker receives the binary image of the AP configuration. The
image includes the administrator’s password required by the HTTP user
interface, the WEP encryption keys, MAC address, and SSID. Yet another AP
returns the WEP keys, MAC filter list, administrator’s password when sent a UDP
packet to port 27155 containing the string “gstsearch
”.
It is not clear how these flaws were discovered. The
following is a likely procedure. Most manufacturers design their
equipment so that its firmware can be flashed with a new and improved one in
the field. The firmware images are downloaded from the manufacturers’ web
site. The CPU used in the APs can be easily recognized, and the firmware
can be systematically disassembled revealing the flaws at the assembly language
level.
Comprehensive lists of such equipment flaws are likely
circulating among the attackers.
7. Denial of Service
A denial of service (DoS) occurs when a system is
not providing services to authorized clients because of resource exhaustion by
unauthorized clients. In wireless networks, DoS attacks are difficult to
prevent, difficult to stop an on-going attack and the victim and its clients
may not even detect the attacks. The duration of such DoS may range from
milliseconds to hours. A DoS attack against an individual station enables
session hijacking.
7.1 Jamming the Air Waves
A number of consumer appliances such as microwave ovens,
baby monitors, and cordless phones operate on the unregulated 2.4GHz radio
frequency. An attacker can unleash large amounts of noise using these devices
and jam the airwaves so that the signal to noise drops so low, that the
wireless LAN ceases to function. The only solution to this is RF proofing
the surrounding environment.
7.2 Flooding with Associations
The AP inserts the data supplied by the station in the
Association Request into a table called the association table that the
AP maintains in its memory. The IEEE 802.11 specifies a maximum value of
2007 concurrent associations to an AP. The actual size of this table
varies among different models of APs. When this table overflows, the AP
would refuse further clients.
Having cracked WEP, an attacker authenticates several
non-existing stations using legitimate-looking but randomly generated MAC
addresses. The attacker then sends a flood of spoofed associate requests
so that the association table overflows.
Enabling MAC filtering in the AP will prevent this attack.
7.3 Forged Dissociation
The attacker sends a spoofed Disassociation frame where the
source MAC address is set to that of the AP. The station is still authenticated
but needs only to reassociate and sends Reassociation Requests to the AP.
The AP may send a Reassociation Response accepting the station and the station
can then resume sending data. To prevent Reassociation, the attacker continues
to send Disassociation frames for a desired period.
7.4 Forged Deauthentication
The attacker monitors all raw frames collecting the source
and destination MAC addresses to verify that they are among the targeted
victims. When a data or Association Response frame is observed, the
attacker sends a spoofed Deauthentication frame where the source MAC address is
spoofed to that of the AP. The station is now unassociated and
unauthenticated, and needs to reconnect. To prevent a reconnection, the
attacker continues to send Deauthentication frames for a desired period. The
attacker may even rate limit the Deauthentication frames to avoid overloading
an already congested network.
The mischievous packets of Disassociation and
Deauthentication are sent directly to the client, so these will not be logged
by the AP or IDS, and neither MAC filtering nor WEP protection will prevent it.
7.5 Power Saving
Power conservation is important for typical station laptops,
so they frequently enter an 802.11 state called Doze. An attacker can
steal packets intended for a station while the station is in the Doze state.
The 802.11 protocol requires a station to inform the AP
through a successful frame exchange that it wishes to enter the Doze state from
the Active state.
Periodically the station awakens and sends a PS-Poll frame
to the AP. The AP will transmit in response the packets that were buffered for
the station while it was dozing. This polling frame can be spoofed by an
attacker causing the AP to send the collected packets and flush its
internal buffers. An attacker can repeat these polling messages so that when
the legitimate station periodically awakens and polls, AP will inform that
there are no pending packets.
8. Man-in-the-Middle Attacks
Man-in-the-middle (MITM) attack
refers to the situation where an attacker on host X inserts X between all communications
between hosts B and C, and neither B nor C is aware of the presence of X.
All messages sent by B do reach C but via X, and vice versa. The attacker
can merely observe the communication or modify it before sending it out.
An MITM attack can break connections that are otherwise secure. At the
TCP level, SSH and VPN, e.g., are prone to this attack.
8.1 Wireless MITM
Assume
that station B was authenticated with C, a legitimate AP. Attacker X is a laptop with two wireless
cards. Through one card, he will present
X as an AP. Attacker X sends
Deauthentication frames to B using the C’s MAC address as the source, and the
BSSID he has collected. B gets
deauthenticated and begins a scan for an AP and may find X on a channel
different from C. There is a race
condition between X and C. If B
associates with X, the MITM attack succeeded.
X will re-transmit the frames it receives from B to C, and the frames it
receives from C to B after suitable modifications.
The
package of tools called AirJack (http://802.11ninja.net/airjack/) includes a program called
monkey_jack
that automates the MITM attack. This is
programmed well so that the odds of it winning in the race condition mentioned
above are improved.8.2 ARP Poisoning
ARP cache poisoning is an old problem in wired
networks. Wired networks have deployed
mitigating techniques. But, the ARP
poisoning technique is re-enabled in the presence of APs that are connected to
a switch/hub along with other wired clients.
ARP is used to determine the MAC address of a device whose
IP address is known. The translation is performed with a table
look-up. The ARP cache accumulates as the host continues to
network. If the ARP cache does not have an entry for an IP address, the
outgoing IP packet is queued, and an ARP Request packet that effectively
requests “If your IP address matches this target IP address, then please let me
know what your Ethernet address is” is broadcast. The host with the target IP
is expected to respond with an ARP Reply, which contains the MAC address of the
host. Once the table is updated because
of receiving this response, all the queued IP packets can now be sent. The
entries in the table expire after a set time in order to account for possible
hardware address changes for the same IP address. This change may have
happened, e.g., due to the NIC being replaced.
Unfortunately, the ARP does not provide for any verification
that the responses are from valid hosts or that it is receiving a spurious
response as if it has sent an ARP Request. ARP poisoning is an attack
technique exploiting this lack of verification.
It corrupts the ARP cache that the OS maintains with wrong MAC addresses
for some IP addresses. An attacker accomplishes this by sending an ARP Reply
packet that is deliberately constructed with a “wrong” MAC address. The
ARP is a stateless protocol. Thus, a machine receiving an ARP Reply
cannot determine if the response is due to a request it sent or not.
ARP poisoning is one of the techniques that enables the
man-in-the-middle attack. An attacker on machine X inserts himself between two
hosts B and C by (i) poisoning B so that C’s IP address is associated with X’s
MAC address, (ii) poisoning C so that B’s address is associated with X’s MAC
address, and (iii) relaying the packets X receives.
The ARP poison attack is applicable to all hosts in a
subnet. Most APs act as transparent MAC layer bridges, and so all stations
associated with it are vulnerable. If an access point is connected directly to
a hub or a switch without an intervening router/firewall, then all hosts
connected to that hub or switch are susceptible also. Note that recent devices
aimed at the home consumer market combine a network switch with may be four or
five ports, an AP, a router and a DSL/cable modem connecting to the Internet at
large. Internally, the AP is connected to the switch. As a result, an attacker on a wireless station
can become a MITM between two wired hosts, one wired one wireless, or both
wireless hosts.
The tool called Ettercap ((http://ettercap.sourceforge.net) is
capable of performing ARP poisoning.
8.3 Session Hijacking
Session hijacking occurs in the context of a “user”,
whether human or computer. The user has an on-going connection with
a server. Hijacking is said to occur when an attacker causes the user to
lose his connection, and the attacker assumes his identity and privileges for a
period.
An attacker disables temporarily the user’s system, say by a
DoS attack or a buffer overflow exploit. The attacker then takes the
identity of the user. The attacker now has all the access that the user
has. When he is done, he stops the DoS attack, and lets the user
resume. The user may not detect the interruption if the disruption lasts
no more than a couple of seconds. Such hijacking can be achieved by using
forged Disassociation DoS attack.
Corporate wireless networks are often set up so that the
user is directed to an authentication server when his station attempts a connection
with an AP. After the authentication, the attacker employs the session
hijacking described above using spoofed MAC addresses.
9. War Driving
Equipped with wireless devices and related tools, and
driving around in a vehicle or parking at interesting places with a goal of
discovering easy-to-get-into wireless networks is known as war driving. War-drivers (http://www.wardrive.net/)
define war driving as “The benign act of locating and logging wireless access
points while in motion.” This benign act is of course useful to the
attackers.
9.1 War chalking
War chalking is the practice of marking sidewalks and walls
with special symbols to indicate that wireless access is nearby so that others
do not need to go through the trouble of the same discovery. A search on www.google.com with key words “war driving
maps” will produce a large number of hits.
Yahoo! Maps can show "Wi-fi Hotspots" near an address you
give.
Figure 3: War Chalking Symbols
9.2 Typical Equipment
The typical war driving equipment consists of a laptop
computer system or a PDA with a wireless card, a GPS, and a high-gain
antenna. Typical choice of an operating system is Linux or FreeBSD
where open source sniffers (e.g., Kismet) and WEP crackers (e.g., AirSnort) are
available. Similar tools (e.g., NetStumbler) that run on Windows are
available.
War drivers need to be within the range of an AP or station
located on the target network. The range depends on the transmit
output power of the AP and the card, and the gain of the antenna.
Ordinary access point antennae transmit their signals in all directions.
Often, these signals reach beyond the physical boundaries of the intended work
area, perhaps to adjacent buildings, floors, and parking lots. With the typical
30mW wireless cards intended for laptops, the range is about 300 feet, but
there are in 2004 wireless cards for laptops on the market that have 200mW.
Directional high-gain antennae and an RF-amplifier can dramatically extend the
range.
Figure 4: War Drivers' Equipment
10. Wireless Security Best Practices
This section describes best practices in mitigating the
problems described above.
10.1 Location of the APs
APs should be topologically located outside the perimeter
firewalls. The wireless network segments
should be treated with the same suspicion as that for the public Internet. Additionally, it is important to use
directional antennae and physically locate them in such a way that the
radio-coverage volume is within the control of the corporation or home.
10.2 Proper Configuration
Statistics collected by www.worldwidewardrive.org show a
distressingly large percentage of APs left configured with the defaults.
Before a wireless device is connected to the rest of the
existing network, proper configuration of the wireless device is necessary.
The APs come with a default SSID, such as “Default SSID”, “WLAN”, “Wireless”,
“Compaq”, “intel”, and “linksys”. The default passwords for the administrator
accounts that configure the AP via a web browser or SNMP are well known for all
manufacturers. A proper configuration
should change these to difficult to predict values.
Note that the SSID serves as a simple handle, not as a
password, for a wireless network. Unless the default SSID on the AP and stations
is changed, SSID broadcasts are disabled, MAC address filtering is enabled, WEP
enabled, an attacker can use the wireless LAN resources without even
sniffing.
The configuration via web browsing (HTTP) is provided by a
simplistic web server built into an AP. Often this configuration
interface is provided via both wired connections and wireless
connections. The web server embedded in a typical AP does not contain
secure HTTP, so the password that the administrator submits to the AP can be
sniffed. Web based configuration via wireless connections should be
disabled.
WEP is disabled in some organization because the throughput
is then higher. Enabling WEP encryption makes it necessary for the
attacker intending to WEP-crack to have to sniff a large number of
frames. The higher the number of bits in the encryption the larger the
number of frames that must be collected is. The physical presence in the radio
range of the equipment for long periods increases the odds of his equipment
being detected. WEP should be enabled.
The IEEE 802.11 does not describe an automated way of
distributing the shared-secret keys. In large installations, the manual
distribution of keys every time they are changed is expensive. Nevertheless,
the WEP encryption keys should be changed periodically.
10.3 Secure Protocols
If the WEP is disabled, or after the WEP is cracked, the
attacker can capture all TCP/IP packets by radio-silent sniffing for later
analyses. All the wired network attacks are possible. There are real-time
tools that analyze and interpret the TCP/IP data as they arrive.
All protocols that send passwords and data in the clear must
be avoided. This includes the rlogin family, telnet, and POP3.
Instead one should use SSH and VPN.
In general, when a wireless segment is involved, one should
use end-to-end encryption at the application level in addition to enabling WEP.
10.4 Wireless IDS
A wireless intrusion detection system (WIDS) is often a
self-contained computer system with specialized hardware and software to detect
anomalous behavior. The underlying
software techniques are the same hacking techniques described above. The special wireless hardware is more capable
than the commodity wireless card, including the RF monitor mode, detection of
interference, and keeping track of signal-to-noise ratios. It also includes GPS equipment so that rogue
clients and APs can be located. A WIDS
includes one or more listening devices that collect MAC addresses, SSIDs, features
enabled on the stations, transmit speeds, current channel, encryption status,
beacon interval, etc. Its computing
engine will be powerful enough that it can dissect frames and WEP-decrypt into
IP and TCP components. These can be fed
into TCP/IP related intrusion detection systems.
Unknown MAC addresses are detected by maintaining a registry
of MAC addresses of known stations and APs.
Frequently, a WIDS can detect spoofed known MAC addresses because the
attacker could not control the firmware of the wireless card to insert the
appropriate sequence numbers into the frame.
10.5 Wireless Auditing
Periodically, every wireless network should be audited. Several audit firms provide this service for
a fee. A security audit begins with a
well-established security policy. A
policy for wireless networks should include a description of the geographical
volume of coverage. The main goal of an
audit is to verify that there are no violations of the policy. To this end, the typical auditor employs the
tools and techniques of an attacker.
10.6 Newer Standards and Protocols
Many improvements in wireless network technology are
proposed through proprietary channels (e.g., Cisco Lightweight Extensible
Authentication Protocol) as well as through the IEEE. The new IEEE 802.11i (ratified in June 2004) enhances
the current 802.11 standard to provide improvements in security. These include Port Based Access Control for
authentication, Temporal Key Integrity Protocol for dynamic changing of
encryption keys, and Wireless Robust Authentication protocol. An interim solution proposed by vendors is
the Wi-Fi Protected Access (WPA), a subset of 802.11i, is only now becoming
available in some products. Time will
tell if these can withstand future attacks.
10.7 Software Tools
Below we describe a collection of cost-free tools that can
be used both as attack tools and as audit tools.
·
AirJack (http://802.11ninja.net/airjack/) is
a collection of wireless card drivers and related programs. It includes a program called
monkey_jack
that automates the MITM
attack. Wlan_jack
is a DoS tool that accepts a target source and
BSSID to send continuous deauthenticate frames to a single client or an entire
network (broadcast address). Essid_jack
sends a disassociate frame to a target client in order to force the client to reassociate
with the network, thereby giving up the network SSID. - AirSnort (www.airsnort.shmoo.com ) can break WEP by passively monitoring transmissions and computing the encryption key when enough packets have been gathered.
- Ethereal (www.ethereal.com ) is a LAN analyzer, including wireless. One can interactively browse the capture data, viewing summary and detail information for all observed wireless traffic.
- FakeAP (ww.blackalchemy.to/project/fakeap) can generate thousands of counterfeit 802.11b access points.
- HostAP (www.hostap.epitest.fi) converts a station that is based on Intersil's Prism2/2.5/3 chipset to function as an access point.
- Kismet (www.kismetwireless.net) is a wireless sniffer and monitor. It passively monitors wireless traffic and dissects frames to identify SSIDs, MAC addresses, channels and connection speeds.
- Netstumbler (www.netstumbler.com) is a wireless access point identifier running on Windows. It listens for SSIDs and sends beacons as probes searching for access points.
- Prismstumbler (prismstumbler.sourceforge.net/) can find wireless networks. It constantly switches channels and monitors frames received.
- The Hacker’s Choice organization (www.thc.org) has LEAP Cracker Tool suite that contains tools to break Cisco LEAP. It also has tools for spoofing authentication challenge-packets from an AP. The WarDrive is a tool for mapping a city for wireless networks with a GPS device.
- StumbVerter (www.sonar-security.com/sv.html) is a tool that reads NetStumbler's collected data files and presents street maps showing the logged WAPs as icons, whose color and shape indicating WEP mode and signal strength.
- Wellenreiter (http://www.wellenreiter.net/) is a WLAN discovery tool. It uses brute force to identify low traffic access points while hiding the real MAC address of the card it uses. It is integrated with GPS.
- WEPcrack (www.wepcrack.sourceforge.net) cracks 802.11 WEP encryption keys using weaknesses of RC4 key scheduling.
11. Conclusion
This article is an introduction to the techniques an
attacker would use on wireless networks. Regardless of the protocols,
wireless networks will remain potentially insecure because an attacker can
listen in without gaining physical access.
In addition, the protocol designs were security-naïve. We have pointed out several existing tools
that implement attack techniques that exploit the weaknesses in the protocol
designs. The integration of wireless
networks into existing networks also has been carelessly done. We pointed out several best practices that
can mitigate the insecurities.