Assessment of stability of algorithms based on trust and reputation model

Abstract

Swarm robotic systems are actively developed and widely studied in the world scientific practice. It is expected that multiagent, distributed approach to creating artificial intelligence of autonomous systems will allow to solve a great number of complex problems in the areas of environmental protection, medicine, cleaning, patrolling, etc. Thereby the research of these systems (design and testing) in terms of information security becomes relevant. The key to a wide practical use of swarm robotic systems is the development of specific guidelines and algorithms for the organization of group actions. This research proposes the use of trust and reputation model for information security of swarm system. Swarm's agents generate trust levels to each other basing on the analysis of situation on the kth iteration step of the algorithm and using their sensor devices. On the calculated trust levels the collective recognition of saboteurs is carried out. To perform experiment software simulator was designed. It allows to vaiy the basic parameters of swarm robotic system (number of agents, number of targets, range of communication, number of saboteurs).