Source-linked AI summary

MAVSec: Securing the MAVLink Protocol for Ardupilot/PX4 Unmanned Aerial Systems

Azza Allouch, Omar Cheikhrouhou, Anis Koubaa, Mohamed Khalgui, Tarek Abbes

arXiv:1905.00265v2cs.CRcs.RO

TL;DR

MAVLink communications lack security protections, leaving UAV–GCS messages exposed to attacks. The paper proposes MAVSec, implementing encryption mechanisms in MAVLink and evaluating their performance. Its experimental study reports that ChaCha20 can maintain message confidentiality without affecting MAVLink performance.

  • Problem

    MAVLink lacks confidentiality and authentication, leaving UAV–GCS communications vulnerable to interception and other attacks.

    Source: §IV · p. 3
  • Method

    The paper proposes MAVSec, adding cryptographic mechanisms to MAVLink and evaluating their implementation and performance.

    Source: §I · p. 2
  • Results

    ChaCha20 can secure MAVLink while maintaining message confidentiality without affecting its performance.

    Source: §VII · p. 7
  • Takeaways & Limitations

    The authors conclude that ChaCha20 can be applied to secure MAVLink communications.

    Source: §VII · p. 7

Abstract

from arXiv · show

The MAVLink is a lightweight communication protocol between Unmanned Aerial Vehicles (UAVs) and ground control stations (GCSs). It defines a set of bi-directional messages exchanged between a UAV (aka drone) and a ground station. The messages carry out information about the UAV's states and control commands sent from the ground station. However, the MAVLink protocol is not secure and has several vulnerabilities to different attacks that result in critical threats and safety concerns. Very few studies provided solutions to this problem. In this paper, we discuss the security vulnerabilities of the MAVLink protocol and propose MAVSec, a security-integrated mechanism for MAVLink that leverages the use of encryption algorithms to ensure the protection of exchanged MAVLink messages between UAVs and GCSs. To validate MAVSec, we implemented it in Ardupilot and evaluated the performance of different encryption algorithms (i.e. AES-CBC, AES-CTR, RC4, and ChaCha20) in terms of memory usage and CPU consumption. The experimental results show that ChaCha20 has a better performance and is more efficient than other encryption algorithms. Integrating ChaCha20 into MAVLink can guarantee its messages confidentiality, without affecting its performance, while occupying less memory and CPU consumption, thus, preserving memory and saving the battery for the resource-constrained drone.

I. INTRODUCTION

MAVLink is widely used for bidirectional UAV–GCS communication, but its lack of security mechanisms exposes the link to attacks. The paper proposes MAVSec, implements cryptographic mechanisms in Ardupilot, and evaluates their performance.

I. INTRODUCTION

  • MAVLink supports bidirectional exchange: the GCS sends commands and controls, while the UAV returns telemetry and status information.The protocol is used by autopilot systems including Ardupilot and PX4.Source: §I · p. 1
  • MAVLink lacks security mechanisms and encryption, leaving UAV–GCS communications vulnerable to spoofing, message forging, and denial of service.The paper notes that the unencrypted channel is subject to attacks.Source: §I · p. 1
  • The authors describe prior MAVLink security solutions as few, with examples using Caesar cipher or RC5 and limitations in key security or authentication.They also note that a secure MAVLink version was under discussion but had not yet been developed.Source: §I · pp. 1–2
  • The paper identifies MAVLink security threats and implements its proposed mechanisms in Ardupilot to demonstrate feasibility.It evaluates the mechanisms’ memory usage and CPU consumption.Source: §I · p. 2
  • MAVSec adds cryptographic mechanisms to protect the confidentiality of messages exchanged between UAVs and ground control stations.The proposed algorithms are RC4, AES-CBC, AES-CTR, and ChaCha20.Source: §I · p. 2

II. RELATED WORKS

Earlier UAV security work includes hardware and software approaches, but the cited studies report performance costs, limited testing, or weaknesses in their security mechanisms. The paper positions its implementation and evaluation as addressing some of these gaps.

II. RELATED WORKS

  • A hardware AES-CBC-MAC solution encrypts and authenticates commands and payloads, but added hardware weight negatively affects performance and power consumption.The solution uses an FPGA module connected to the drone.Source: §II · p. 2
  • A Raspberry Pi encrypted-channel approach aims to regain UAV control during an attack, but adds delay and CPU usage and was not tested on real drone communications.Source: §II · p. 2
  • Software proposals using Caesar cipher and RC5 are described as vulnerable or lacking explicit performance analysis and testing details.The paper contrasts these studies with its evaluation of four cryptographic methods.Source: §II · p. 2
  • Marty’s vulnerability analysis proposed cryptographic techniques and an evaluation methodology, but did not implement them or demonstrate feasibility.The paper implements encryption mechanisms in Ardupilot and evaluates their performance.Source: §II · p. 2

III. MAVLINK SYSTEM ARCHITECTURE

MAVLink is a lightweight, header-only protocol for bidirectional communication between GCSs and UAVs. Its messages carry commands, control, and state information, with packet fields supporting message identification, integrity checking, and authentication.

III. MAVLINK SYSTEM ARCHITECTURE

  • MAVLink is an open-source, lightweight, header-only protocol used mainly for bidirectional GCS–UAV communication.MAVLink 2.0 is backward compatible with version 1.0 and is the current recommended version.Source: §III · p. 2
  • GCS-to-UAV messages carry commands and control, while UAV-to-GCS messages carry state information such as position, heartbeat, and system status.The heartbeat is mandatory and generally sent every second to report that the drone is active and connected.Source: §III · pp. 2–3
  • Each MAVLink message has a header and payload; the header contains message information, and the checksum is intended to verify integrity during transmission.Source: §III · p. 3
  • MAVLink packet lengths range from 11 bytes minimum to 297 bytes maximum with full payload and signature, and the payload length varies with the transmitted parameters.The signature field allows message authentication and checks that the message comes from a trusted source.Source: §III · p. 3

IV. SECURITY THREATS

MAVLink’s lack of confidentiality, authentication, and integrity protections exposes UAV–GCS communications to interception, modification, and interruption. The section describes how these threats affect messages and control links.

IV. SECURITY THREATS

  • MAVLink provides no confidentiality or authentication, allowing anyone with an appropriate transmitter to inject commands or impersonate a drone.Unencrypted message streams can also be intercepted and eavesdropped.Source: §IV · p. 3
  • The threats are framed in terms of confidentiality, integrity, and availability: interception, modification, and interruption, respectively.Source: §IV · p. 3
  • Interception lets unauthorized users read messages, including GCS commands to UAVs and data sent back; authentication and encryption are proposed to mitigate this risk.Source: §IV · p. 3
  • Modification can tamper with messages, and MAVLink’s lack of integrity protection might let an attacker hijack a UAV from its GCS.Source: §IV · p. 3
  • Interruption blocks MAVLink control signals, potentially leaving the aircraft in a lost-link state; strong authentication can help mitigate unavailability risk.The paper says Table I lists threats and corresponding mitigation techniques.Source: §IV · p. 3

V. MAVSEC: SECURITY OF THE MAVLINK PROTOCOL

MAVSec adds cryptographic mechanisms to MAVLink to mitigate confidentiality vulnerabilities, using symmetric encryption intended to suit resource-constrained drones. It describes four encryption algorithms and their integration into MAVLink messages.

V. MAVSEC: SECURITY OF THE MAVLINK PROTOCOL

  • MAVSec adds cryptographic mechanisms to MAVLink to mitigate vulnerabilities in confidentiality.The approach is presented as an enhanced MAVLink version.Source: §V · p. 3

A. Background on the encryption mechanisms.

  • The paper evaluates AES-CTR, AES-CBC, RC4, and ChaCha20; symmetric encryption is described as fast and energy-efficient for low-resource drones.The paper distinguishes stream ciphers from block ciphers and selects AES-CBC and AES-CTR modes.Source: §V · pp. 3–4, A. Background on the encryption… · p. 4

1) AES Counter Mode (CTR) :

  • CTR encrypts counter blocks and XORs the resulting keystream with plaintext; encryption and decryption can run in parallel without padding.The paper presents CTR as a flexible way to use a block cipher as a stream cipher for MAVLink's limited buffering.Source: 1) AES Counter Mode (CTR) · p. 4

2) AES Cipher Block Chaining Mode (CBC) :

4) ChaCha20 :

  • ChaCha20 uses a 256-bit key, 96-bit nonce, and 32-bit initial block counter to encrypt arbitrary-length plaintext without padding.It generates a keystream from the key, nonce, and block counter, then XORs that stream with plaintext.Source: 4) ChaCha20 · p. 5

B. MAVSec: Integration of encryption mechanisms into MAVLink

  • MAVSec integrates encryption and decryption on both the autopilot and GCS, encrypting only payloads so message identifiers remain readable.The UAV encrypts payloads with a session key; the checksum is computed after encryption, checked by the GCS, and the payload is then decrypted.Source: B. MAVSec: Integration of encryption… · p. 6

VI. EXPERIMENTAL VALIDATION

The evaluation used an Ardupilot Software-In-The-Loop simulated drone and compared unsecured MAVLink with four encryption algorithms on packet transmission, memory, and CPU use. ChaCha20 performed better overall than the alternatives, with memory use indistinguishable from unsecured MAVLink and CPU use close to it.

A. Simulation environment settings

  • The testbed used an Ardupilot SITL simulated drone connected to QGroundControl, and each algorithm was run ten times.The setup ran ArduPilot 3 on a Linux virtual machine with a UAV copter; encryption was integrated into both the autopilot and ground station to exchange and decode the cipher stream.Source: A. Simulation environment settings · p. 6, B. Performance evaluation · p. 6

B. Performance evaluation

  • Fig. 3 compares packet transmission: ChaCha20 and CTR sent more packets than CBC and RC4, while unsecured MAVLink sent only negligibly more than ChaCha20.The paper attributes CBC's lower packet count to its processing time, key chaining, lack of parallel encryption, and padding for plaintext sizes that are not block-size multiples.Source: B. Performance evaluation · pp. 6–7, Figure 3 · p. 7
  • ChaCha20 performed better overall: its memory use matched unsecured MAVLink, and its CPU utilization was close to the unsecured version.RC4 had the highest memory and CPU use; its sequential KSA and PRGA require more registers, while ChaCha20 uses CPU-friendly ARX instructions.Source: B. Performance evaluation · p. 7, Figure 4 · p. 7, Figure 5 · p. 7

VII. CONCLUSION

The authors identify MAVLink security threats, propose cryptographic solutions, and report that ChaCha20 preserves message confidentiality without affecting performance.

VII. CONCLUSION

  • The authors conclude that ChaCha20 can preserve MAVLink message confidentiality without affecting performance.They implemented encryption algorithms in MAVLink source code and identify testing in a real scenario as future work.Source: §VII · p. 7
Loading 1905.00265v2…