Wireless mice and keyboards commonly communicate using proprietary protocols operating in the 2.4GHz ISM band. In contrast to Bluetooth, there is no industry standard to follow, leaving each vendor to implement their own security scheme.
Wireless mice and keyboards work by transmitting radio frequency packets to a USB dongle plugged into a user’s computer. When a user presses a key on their keyboard or moves their mouse, information describing the actions are sent wirelessly to the USB dongle.
The dongle listens for radio frequency packets sent by the mouse or keyboard, and notifies the computer whenever the user moves their mouse or types on their keyboard.
In order to prevent eavesdropping, most vendors encrypt the data being transmitted by wireless keyboards. The dongle knows the encryption key being used by the keyboard, so it is able to decrypt the data and see what key was pressed. Without knowing the encryption key, an attacker is unable to decrypt the data, so they are unable to see what is being typed.
Conversely, none of the mice that were tested encrypt their wireless communications. This means that there is no authentication mechanism, and the dongle is unable to distinguish between packets transmitted by a mouse, and those transmitted by an attacker. As a result, an attacker is able to pretend to be a mouse and transmit their own movement/click packets to a dongle.
Problems in the way the dongles process received packets make it possible for an attacker to transmit specially crafted packets which generate keypresses instead of mouse movement/clicks.
Nordic Semiconductor makes the popular nRF24L series of transceivers used in most of the devices vulnerable to MouseJack. The nRF24L transceivers provide a mechanism to wirelessly transmit data between two devices, but the functionality that turns mouse clicks and keypresses into bytes sent over the air is implemented by each vendor. From a research standpoint, this means that the same tools and procedures can be used to evaluate products from different vendors.
Software Defined Radio Decoder
The nRF24L transceivers support multiple data rates, address lengths, packet formats, and checksums. To accommodate this, the initial research was performed using a USRP B210 Software Defined Radio, coupled with a custom GNU Radio block designed to decode all of the possible packet configurations. This proved fruitful, but there were drawbacks to using an SDR.
None of the tested devices employ frequency hopping in the traditional sense, but they all change channels to avoid interference from other 2.4GHz devices (Bluetooth, Wi-Fi, etc). The channel hopping is generally unpredictable, and Software Defined Radios are slower to retune than the nRF24L radios. This makes it difficult for an SDR based decoder to observe all of the transmitted packets.
When a mouse transmits a movement packet to a dongle, the dongle replies with an acknowledgement packet telling the mouse that the movement packet was received. The mouse waits for a short period before determining that the packet it transmitted was lost, which can be as short as 250 microseconds. Due to USB latency and processing overhead, the SDR-based decoder is unable to transmit ACKs within the narrow timeout window, so two way communication between an SDR and dongle/mouse is not a viable option.
Enter the NES Controller
The SDR decoder made it possible to figure out the formats of the data being transmitted over the air, but reliable two way communication would be necessary to start evaluating the devices for potential weaknesses.
Parallel to the MouseJack research, an Arduino/nRF24L-based NES controller was being built as part of a Burning Man project. The nRF24L was chosen because they are inexpensive and easy to use, but the connection with the MouseJack project quickly became apparent and it was decided to build an NES controller that could spoof wireless mice.
The nRF24L chips do not officially support packet sniffing, but Travis Goodspeed documented a pseudo-promiscuous mode in 2011 which makes it possible to sniff a subset of packets being transmitted by other devices. This enables the NES controller to passively identify wireless mice and keyboards without the need for an SDR.
The NES controller proved to be an excellent platform for learning about the behavior of mouse communication protocols. As opposed to passively collecting data, the NES controller translates d-pad arrows into mouse movement packets, and A/B buttons into left and right clicks. In order to achieve a smooth user experience, it was necessary to create a model of the packet timing and specific behavior expected by the dongle.
The concept worked well, and the NES controller was presented at ToorCon in 2015, which demonstrated the viability of controlling previously unseen wireless mice at will.
Despite being a marked improvement over the SDR decoder, the NES controller was not without problems. Running off of battery power made it impractical to use amplified transceivers, limiting the practical range to around 10 meters.
Crazyradio PA Dongles
The Crazyflie is an open source drone which is controlled with an amplified nRF24L-based USB dongle called the Crazyradio PA. This is equivalent to an amplified version of the USB dongles commonly used with wireless mice and keyboards. Modifying the Crazyradio PA firmware to include support for pseudo-promiscuous mode made it possible to distill the packet sniffing and injection functionality down to a minimal amount of Python code.
The Crazyradio PA dongles made it possible to implement an efficient and effective fuzzer. Mouse and keyboard USB dongles communicate user actions to the operating system in the form of USB HID packets, which can be sniffed by enabling the usbmon kernel module on Linux.
The implemented fuzzer takes advantage of this by transmitting RF packets to a mouse/keyboard dongle attached to the same computer, and monitoring USB traffic for generated USB HID packets. Anytime mouse movement or keypresses are sent to the operating system, the recently transmitted RF packets are recorded for analysis. Fuzzing variants of observed packet formats and behaviors yielded the best results.
1. Keystroke injection, spoofing a mouse
When processing received RF packets, some dongles do not verify that the type of packet received matches the type of device that transmitted it. Under normal circumstances, a mouse will only transmit movement/clicks to the dongle, and a keyboard will only transmit keypresses. If the dongle does not verify that the packet type and transmitting device type match, it is possible for an attacker to pretend to be a mouse, but transmit a keypress packet. The dongle does not expect packets coming from a mouse to be encrypted, so it accepts the keypress packet, allowing the attacker to type arbitrary commands on the victim’s computer.
2. Keystroke injection, spoofing a keyboard
Most of the tested keyboards encrypt data before transmitting it wirelessly to the dongle, but not all of the dongles require that encryption is used. This makes it possible for an attacker to pretend to be a keyboard, and transmit unencrypted keyboard packets to the dongle. This bypasses the encryption normally used by the keyboard, and allows an attacker to type arbitrary commands on the victim’s computer.
3. Forced pairing
Before a wireless keyboard or mouse leaves the factory, it is paired with a dongle. This means that it knows the wireless address of the dongle, and in the case of a keyboard, the secret encryption key.
Some vendors include the ability to pair new devices with a dongle, or pair an existing keyboard or mouse with a new dongle. If a dongle is lost, for example, this means that the user only needs to purchase a new dongle, instead of an entirely new set.
To prevent unauthorized devices from pairing with a dongle, the dongle will only accept new devices when it has been placed into a special “pairing mode” by the user, which lasts for 30-60 seconds.
It is possible to bypass this pairing mode on some dongles and pair a new device without any user interaction. In the case where a victim only has a mouse, but is using a dongle vulnerable to keystroke injection by spoofing a keyboard, an attacker can pair a fake keyboard with the dongle, and use it to type arbitrary commands on the victim’s computer.
Anatomy of an Attack
Carrying out a MouseJack attack does not require specialized or expensive equipment, and can be done with a $15 USB dongle.
First, the attacker identifies a target wireless mouse or keyboard by listening for RF packets transmitted when a user is moving/clicking the mouse or typing on the keyboard.
If necessary, the attacker now force-pairs a fake keyboard with the victim’s dongle.
Finally, the attacker transmits keypress packets to type a series of commands into the victim’s computer. This can include downloading a virus or rootkit, transferring files off of the victim’s computer, or anything else the attacker could do if they were physically typing on the computer’s keyboard.
There are two basic types of nRF24L chips used by keyboards, mice, and dongles: one-time-programmable, and flash memory. One-time-programmable devices cannot be updated once they leave the factory, but flash memory devices can.
For non-updateable devices, which represent the majority of those tested, there is no mechanism to secure a vulnerable device short of unplugging the USB dongle from the computer.
For devices with updated firmware available from the manufacturer, it is recommended to install the update before continuing to use the affected mouse or keyboard.
-  Goodspeed, Travis. (February 7, 2011). Promiscuity is the nRF24L01+’s Duty. Retrieved from http://travisgoodspeed.blogspot.com/2011/02/promiscuity-is-nrf24l01s-dut…
-  Newlin, Marc. (October 24, 2015). Hacking Wireless Mice with an NES Controller. Presented at ToorCon 17, San Diego, CA
-  Bitcraze AB. (2016). Crazyflie 2.0. Retrieved from https://www.bitcraze.io/crazyflie-2/
-  Bitcraze AB. (2016). Crazyradio PA. Retrieved from https://www.bitcraze.io/crazyradio-pa/
- – “KeyKeriki v2.0 – 2.4GHz”, Thorsten Schröder & Max Moser, CanSecWest 2010.
- – “Promiscuity is the nRF24L01+’s Duty”, Travis Goodspeed, 2011.
- – “KeySweeper”, Samy Kamkar, 2015.
How does MouseJack build upon previous work in this area?
MouseJack takes a fresh look at the undiscovered insecurities still present in non-Bluetooth wireless mice and keyboards, and presents a new class of vulnerabilities that affects multiple vendors and models. The novel aspects of these vulnerabilties relate to:
- 1. ‘Keystroke Injection’: A new attack vector for injecting malicious keystrokes into a victim’s computer via a wireless mouse dongle, so a victim is not required to be using a wireless keyboard (previously, injection had only been into victim’s keyboard dongle).
- 2. ‘Force Pairing’: The abilitiy to pair an attacker’s MouseJack transmitter with a dongle that has not been instructed to accept new pairings. This is akin to accepting an intruder as a legitimate member to a trusted party.
- 3. ‘Fake Mouse’: Tricking the host computer (and the victim, via the manufacturer’s software user interface) into believing that a legitimate mouse has been paired, even though the device will actually act as a malicious keyboard and be used to inject keystrokes. One would have expected that a device presenting itself as mouse would only be able to have mouse movement/click transmissions accepted by the target paired dongle, and all other traffic (e.g. keystrokes) would be rejected.
- 4. ‘Denial-Of-Service’: Mice and keyboards will stop operating after receiving a malicious payload, and only resume normal operation after they are power cycled.
Keyboard encryption has been cracked before (e.g. KeyKeriki). What is the contribution of this research?
Previous efforts, such as KeyKeriki, and later implementations such as KeySweeper, have been highly successful at showing the weakness in keyboard encryption. That research focused on the vulnerability present in a weak encryption scheme implemented in Microsoft keyboards, and showed very effectively how keystrokes could be sniffed, the encryption key could be calculated, and malicious keystrokes could be injected. Since that time, manufacturers have attempted to implement better encryption schemes, such as AES which is present in newer Microsoft keyboards (mouse traffic remains unencrypted). MouseJack shows that, despite stronger encryption being used, it is possible to circumvent the entire scheme by creating specially-crafted keystroke frames. This is a fundamental flaw in the dongle firmware, separate from whichever encryption scheme is used, and allows an attacker to act as if encryption was not implemented.
You released advisories that affect many vendors. Did you take the time to engage with them?
We notified all vendors, and acted according to our Vulnerability Disclosure Policy. We were proactive in our communication with all parties, and have repeatedly contacted vendors to request acknowledgement of the advisories and progress updates. We had multiple teleconferences to ensure vendors could reproduce the attack, and coordinate actions following public disclosure. In particular instances, we provided Proof-of-Concept code and hardware to enable reproduction of the attack, and invited representatives on-site to give them direct access to our research team and tools.
The public repository only contains sniffer/device discovery code. Where is the Proof-of-Concept exploit code?
We disclosed full, technical information to all affected vendors, and only those vendors, so they would be able to reproduce the attacks in an effort to expedite the release of patched firmware. The open source code we released is only to enable the public to find vulnerable devices and take appropriate steps to remediate the MouseJack threat.