The Detection Of Malicious Activities Within A Robotic Swarm

Ian Sargeant

Research output: ThesisDoctoral Thesis

396 Downloads (Pure)


Much research has been undertaken in the field of swarm robotics and its potential applications. However, the majority of the research work assumes a benign operational environment for the robotic swarm. This work considers the effect of malicious attackers attempting to either prevent a swarm from achieving its goal or influence the swarm in such a way, so as to make it less efficient.

This work provides details the characteristics of a swarm and provides a taxonomy for swarms. While the threat types for a swarm are no different to the threats types for traditional information systems, we show that there are differences with the attack types and attack methods, due to the unique characteristics of a swarm when compared to a traditional information system.

The work then provides a generic model for swarms, their entities and their operational environment. The generic model is used in conjunction with a taxonomy of threats, to enable the research to develop simulations for the swarms, to investigate whether swarms could be maliciously manipulated, with the results of the simulations demonstrating that the swarms could be manipulated.

The research progressed to researching and implementing Intrusion Detection Systems (IDS), tailored for the swarm. This was based on research around current IDS methods and it was determined that signature based IDS could be utilised but anomaly based IDS were not suitable for use in swarms.

Simulations were undertaken that utilised IDS functionality within scenarios that had previously been attacked and the IDS functionality was shown to be successful in the majority of instances. That is, the swarms were either more efficient at achieving their goal, when compared to being attacked without the IDS functionality, or the swarms were simply able to complete their tasks, when previously they had not.
Original languageEnglish
Awarding Institution
  • Royal Holloway, University of London
  • Tomlinson, Allan, Supervisor
Award date1 Mar 2019
Publication statusUnpublished - 2019


  • robotic swarm
  • robotic swarms
  • security
  • attacks
  • taxonomy
  • Modelling and Simulation
  • intrusion detection system

Cite this