Assessment of the possibilities of using behavioral biometrics: analysis of computer mouse movements to protect remote administration sessions

The purpose of this work is to substantiate the possibilities of using mouse dynamics as a method of behavioral biometrics for the tasks of continuous authentication of system administrators in remote access conditions. The research focuses on studying the features of mixed (discrete-continuous) data transmission channels and the specifics of using GUI interfaces used in modern administration scenarios. The paper considers formal models for processing behavioral signs, suggests an approach to integrating asynchronous and fragmentary signals, and performs a comparative analysis of biometric methods based on stability criteria, the possibilities of application in background modes, and the possibility of integration without additional equipment. Particular attention is paid to the architectural requirements for Continuous Authentication Systems (CAS), including assessing the adaptability of models and determining their resistance to data flow fragmentation. The analysis results confirm that mouse dynamics has balanced characteristics for passive biometric authentication.: It is actively used in software and hardware platforms with a graphical interface, does not require specialized sensors, and provides good identification quality with a low level of interference. It is shown that this type of biometric authentication can be effectively applied in conditions of an unstable channel, while meeting the requirements for synchronization, aggregation and profile adaptation. The proposed recommendations on the architecture of CAS systems are focused on real-world application in the IT infrastructure without compromising performance and user experience.