Dr Julian Murphy
Dr Julian Murphy is a Visiting Researcher in the School of Engineering and Digital Arts.
Conference or workshop item
Murphy, J., Howells, G. and McDonald-Maier, K. (2018). On Quaternary 1-of-4 ID Generator Circuits. In: IEEE EST 2018. IEEE. Available at: http://dx.doi.org/10.1109/AHS.2018.8541477.A quaternary 1-of-4 ID generator circuit is
presented. It exploits the properties of quaternary metastability
to provide stable n-bit IDs tolerant to the effects
of nanoscale process scaling and temperature variations,
which is achieved by increasing the margin between
threshold voltage and metastability voltage via quaternary
metastability. A 128-bit ID generator hardware implementation
and electrical evaluation yields an average of 26.6%
uniqueness of ID bits and 100% stability rate over a 10?C
to 40?C temperature range.
Murphy, J., Howells, G. and McDonald-Maier, K. (2017). Multi-factor Authentication using Accelerometers for the Internet-of-Things. In: Seventh IEEE International Conference on Emerging Security Technologies. IEEE, pp. 103-107. Available at: https://doi.org/10.1109/EST.2017.8090407.Embedded and mobile devices forming part of the Internet-of-Things (IoT) need new authentication technologies and techniques. This requirement is due to the increase in effort and time attackers will use to compromise a device, often remote, based on the possibility of a significant monetary return. This paper proposes exploiting a device’s accelerometers in-built functionality to implement multi-factor authentication. An experimental embedded system designed to emulate a typical mobile device is used to implement the ideas and investigated as proof-of-concept.
Murphy, J., Howells, G. and McDonald-Maier, K. (2019). A Machine Learning Method For Sensor Authentication Using Hidden Markov Models. In: Eighth IEEE International Conference on Emerging Security Technologies. IEEE. Available at: http://dx.doi.org/10.1109/EST.2019.8806200.A machine learning method for sensor based authentication is presented. It exploits hidden markov models to generate stable and synthetic probability density functions from variant sensor data. The principle, and novelty, of the new method are presented in detail together with a statistical evaluation. The results show a marked improvement in stability through the use of hidden markov models.