airbase-ng - multi-purpose tool aimed at attacking clients as opposed
to the Access Point (AP) itself
airbase-ng [options] <interface name>
airbase-ng is multi-purpose tool aimed at attacking clients as opposed
to the Access Point (AP) itself. Since it is so versatile and flexible,
summarizing it is a challenge. Here are some of the feature highlights:
- Implements the Caffe Latte WEP client attack
- Implements the Hirte WEP client attack
- Ability to cause the WPA/WPA2 handshake to be captured
- Ability to act as an ad-hoc Access Point
- Ability to act as a full Access Point
- Ability to filter by SSID or client MAC addresses
- Ability to manipulate and resend packets
- Ability to encrypt sent packets and decrypt received packets
The main idea is of the implementation is that it should encourage
clients to associate with the fake AP, not prevent them from accessing
the real AP.
A tap interface (atX) is created when airbase-ng is run. This can be
used to receive decrypted packets or to send encrypted packets.
As real clients will most probably send probe requests for
common/configured networks, these frames are important for binding a
client to our softAP. In this case, the AP will respond to any probe
request with a proper probe response, which tells the client to
authenticate to the airbase-ng BSSID. That being said, this mode could
possibly disrupt the correct functionality of many APs on the same
Shows the help screen.
If the BSSID is not explicitly specified by using "-a <BSSID>",
then the current MAC of the specified interface is used.
Also capture and process from this interface in addition to the
-w <WEP key>
If WEP should be used as encryption, then the parameter "-w <WEP
key>" sets the en-/decryption key. This is sufficient to let
airbase-ng set all the appropriate flags by itself. If the
softAP operates with WEP encryption, the client can choose to
use open system authentication or shared key authentication.
Both authentication methods are supported by airbase-ng. But to
get a keystream, the user can try to force the client to use
shared key authentication. "-s" forces a shared key auth and "-S
<len>" sets the challenge length.
This is the source MAC for the man-in-the-middle attack. The
"-M" must also be specified.
If this option is not specified, it defaults to "-f allow". This
means the various client MAC filters (-d and -D) define which
clients to accept.
By using the "-f disallow" option, this reverses selection and
causes airbase to ignore the clients specified by the filters.
This sets the beacon WEP flag. Remember that clients will
normally only connect to APs which are the same as themselves.
Meaning WEP to WEP, open to open.
The "auto" option is to allow airbase-ng to automatically set
the flag based on context of the other options specified. For
example, if you set a WEP key with -w, then the beacon flag
would be set to WEP.
One other use of "auto" is to deal with clients which can
automatically adjust their connection type. However, these are
few and far between.
In practice, it is best to set the value to the type of clients
you are dealing with.
-q This surpresses printing any statistics or status information.
-v This prints additional messages and details to assist in
-M This option is not implemented yet. It is a man-in-the-middle
attack between specified clients and BSSIDs.
This causes airbase-ng to act as an ad-hoc client instead of a
normal Access Point.
In ad-hoc mode airbase-ng also sends beacons, but doesn’t need
any authentication/association. It can be activated by using
"-A". The soft AP will adjust all flags needed to simulate a
station in ad-hoc mode automatically and generate a random MAC,
which is used as CELL MAC instead of the BSSID. This can be
overwritten by the "-a <BSSID>" tag. The interface MAC will then
be used as source mac, which can be changed with "-h
The parameter "-Y" enables the "external processing" Mode. This
creates a second interface "atX", which is used to
replay/modify/drop or inject packets at will. This interface
must also be brought up with ifconfig and an external tool is
needed to create a loop on that interface.
The packet structure is rather simple: the ethernet header (14
bytes) is ignored and right after that follows the complete
ieee80211 frame the same way it is going to be processed by
airbase-ng (for incoming packets) or before the packets will be
sent out of the wireless card (outgoing packets). This mode
intercepts all data packets and loops them through an external
application, which decides what happens with them. The MAC and
IP of the second tap interface doesn’t matter, as real ethernet
frames on this interface are dropped dropped anyway.
There are 3 arguments for "-Y": "in", "out" and "both", which
specify the direction of frames to loop through the external
application. Obviously "in" redirects only incoming (through the
wireless NIC) frames, while outgoing frames aren’t touched.
"out" does the opposite, it only loops outgoing packets and
"both" sends all both directions through the second tap
There is a small and simple example application to replay all
frames on the second interface. The tool is called "replay.py"
and is located in "./test". It’s written in python, but the
language doesn’t matter. It uses pcapy to read the frames and
scapy to possibly alter/show and reinject the frames. The tool
as it is, simply replays all frames and prints a short summary
of the received frames. The variable "packet" contains the
complete ieee80211 packet, which can easily be dissected and
modified using scapy.
This can be compared to ettercap filters, but is more powerful,
as a real programming language can be used to build complex
logic for filtering and packet customization. The downside on
using python is, that it adds a delay of around 100ms and the
cpu utilizations is rather large on a high speed network, but
its perfect for a demonstration with only a few lines of code.
This is used to specify the channel on which to run the Access
This causes the Access Point to hide the SSID and to not
broadcast the value.
-s When specfiied, this forces shared key authentication for all
The soft AP will send an "authentication method unsupported"
rejection to any open system authentication request if "-s" is
-S It sets the shared key challenge length, which can be anything
from 16 to 1480. The default is 128 bytes. It is the number of
bytes used in the random challenge. Since one tag can contain a
maximum size of 255 bytes, any value above 255 creates several
challenge tags until all specified bytes are written. Many
clients ignore values different than 128 bytes so this option
may not always work.
Airbase-ng also contains the new caffe-latte attack, which is
also implemented in aireplay-ng as attack "-6". It can be used
with "-L" or "caffe-latte". This attack specifically works
against clients, as it waits for a broadcast arp request, which
happens to be a gratuitous arp. See this for an explaination of
what a gratuitous arp is. It then flips a few bits in the sender
MAC and IP, corrects the ICV (crc32) value and sends it back to
the client, where it came from. The point why this attack works
in practice is, that at least windows sends gratuitous arps
after a connection on layer 2 is established and a static ip is
set, or dhcp fails and windows assigned an IP out of
"-x <pps>" sets the number of packets per second to send when
performing the caffe-latte attack. At the moment, this attack
doesn’t stop, it continuously sends arp requests. Airodump-ng is
needed to capture the replys.
This attack listens for an ARP request or IP packet from the
client. Once one is received, a small amount of PRGA is
extracted and then used to create an ARP request packet targeted
to the client. This ARP request is actually made of up of
multiple packet fragments such that when received, the client
This attack works especially well against ad-hoc networks. As
well it can be used against softAP clients and normal AP
This sets the number of packets per second that packets will be
sent (default: 100).
-y When using this option, the fake AP will not respond to
broadcast probes. A broadcast probe is where the the specific AP
is not identified uniquely. Typically, most APs will respond
with probe responses to a broadcast probe. This flag will
prevent this happening. It will only respond when the specific
AP is uniquely requested.
-0 This enables all WPA/WPA2/WEP Tags to be enabled in the beacons
sent. It cannot be specified when also using -z or -Z.
This specifies the WPA beacon tags. The valid values are:
1=WEP40 2=TKIP 3=WRAP 4=CCMP 5=WEP104.
same as -z, but for WPA2
This specifies the valid EAPOL types. The valid values are:
1=MD5 2=SHA1 3=auto
This option causes airbase-ng to write all sent and received
packets to a pcap file on disk. This is the file prefix (like
-P This causes the fake access point to respond to all probes
regardless of the ESSIDs specified.
This sets the time in milliseconds between each beacon.
The wildcard ESSIDs will also be beaconed this number of
seconds. A good typical value to use is "-C 60" (require -P).
--bssid <MAC>, -b <MAC>
BSSID to filter/use.
--bssids <file>, -B <file>
Read a list of BSSIDs out of that file.
--client <MAC>, -d <MAC>
MAC of client to accept.
--clients <file>, -D <file>
Read a list of client’s MACs out of that file
--essid <ESSID>, -e <ESSID>
Specify a single ESSID. For SSID containing special characters,
--essids <file>, -E <file>
read a list of ESSIDs out of that file.
This manual page was written by Thomas d’Otreppe. Permission is
granted to copy, distribute and/or modify this document under the terms
of the GNU General Public License, Version 2 or any later version
published by the Free Software Foundation On Debian systems, the
complete text of the GNU General Public License can be found in