PROJECT 5

Bilinear Mechanisms for Physical Layer Security in the Internet of Things

Project Leader: Urbashi Mitra

Website: https://viterbi.usc.edu/directory/faculty/Mitra/Urbashi

Abstract: The Internet of Things (IoT) has introduced a wide gamut of interconnected physical objects interacting with online services. While many IoT devices propose novel services to simplify and enhance the end user’s quality of life (e.g. smart speakers, health monitoring devices, wearables, . . . ), this is often achieved by collecting and transmitting personal user information over wireless channels. The IoT context raises new security and privacy challenges. In particular, data streams carrying confidential information on the user are exposed to the risk of eavesdropping. Thus, developing new edge security protocols to protect confidential data is critical. Our focus is on developing new information theoretic and cryptographic approaches for secure and privacy preserving learning and inference by securing the wireless signals themselves. We will exploit properties of communication over wireless channels and theoretical properties of the mapping of data to the signals received. This mapping falls within the class of bilinear functions and we, and others, have been investigating the hardness of inverting bilinear functions. We will develop novel physical layer communication protocols that leverage the relative hardness of bilinear inverse problem to enhance privacy in wireless networks. Our problem framework has two important features that we will address directly, the large number of connected devices and computationally pragmatic algorithms.