Szczesniak, Emma Victoria2025-05-202025-05-202025-05-19vt_gsexam:43593https://hdl.handle.net/10919/133139Neural interface systems such as myoelectric-computer interfaces (MCIs) and brain-computer interfaces (BCIs) assist patients with motor impairments due to injury or neurodegenerative conditions. Neural interface research has focused on device accuracy and usability while neglecting to comprehensively assess security risks. These devices store substantial personal data that can lead to exploitation if compromised. Attacks can override user intent, having major implications on a user's physical safety and psychological well-being. As neural interfaces become more prevalent, understanding and addressing their vulnerabilities is imperative to ensure user safety and data privacy. This study aimed to identify distinct frequency characteristics between upper limb motor tasks and examine data transmission frequencies to assess potential vulnerabilities in MCI systems. The HackRF One identified three distinct frequencies involved in the frequency hopping pattern during signal transmission, which allows attackers to intercept EMG data. Surface electromyography (sEMG) data produced by wrist flexion and extension motor tasks were analyzed in the frequency domain and showed statistically significant differences in frequency metrics. Distinguishing frequency metrics enable manipulation of motor commands in MCI systems by sending false signals at specific frequencies. This work provides insight into vulnerabilities in the signal transmission stage of neural interfaces to encourage developers to safeguard against potential attacks and to inform consumers of the security risks associated with these devices and their impact on user safety and protection of neural data.ETDenIn CopyrightMyoelectric-Computer InterfaceBiomedical Device SecurityAdversarial AttackSignal Transmission VulnerabilitiesThesis