Li QianQian

Curriculum
Computer Science and Innovation for Societal Challenges, XXX series
Grant sponsor
CARIPARO
Supervisor

Mauro Conti
Co-supervisor

Luciano Gamberini
 
 

Project: Human-Computer Interaction: Security Aspects
Full text of the dissertation book can be downloaded from: http://paduaresearch.cab.unipd.it/10691/

Abstract
Along with the rapid development of intelligent information age, users are having a growing interaction with smart devices. Such smart devices are interconnected together in the Internet of Things (IoT). The sensors of IoT devices collect information about users' behaviors from the interaction between users and devices. Since users interact with IoT smart devices for the daily communication and social network activities, such interaction generates a huge amount of network traffic. Hence, users' behaviors are playing an important role in the security of IoT smart devices, and the security aspects of Human-Computer Interaction are becoming significant. In this dissertation, we provide a threefold contribution: (1) we review security challenges of HCI-based authentication, and design a tool to detect deceitful users via keystroke dynamics; (2) we present the impact of users' behaviors on network traffic, and propose a framework to manage such network traffic; (3) we illustrate a proposal for energy-constrained IoT smart devices to be resilient against energy attack and efficient in network communication.
More in detail, in the first part of this thesis, we investigate how users' behaviors impact on the way they interact with a device. Then we review the work related to security challenges of HCI-based authentication on smartphones, and Brain-Computer Interfaces (BCI). Moreover, we design a tool to assess the truthfulness of the information that users input using a computer keyboard. This tool is based on keystroke dynamics and it relies on machine learning technique to achieve this goal. To the best of our knowledge, this is the first work that associates the typing users' behaviors with the production of deceptive personal information. We reached an overall accuracy of 76% in the classification of a single answer as truthful or deceptive.
In the second part of this dissertation, we review the analysis of network traffic, especially related to the interaction between mobile devices and users. Since the interaction generates a huge amount of network traffic, we propose an innovative framework, GolfEngine, to manage and control the impact of users behavior on the network relying on Software Defined Networking (SDN) techniques. GolfEngine provides users a tool to build their security applications and offers Graphical User Interface (GUI) for managing and monitoring the network. In particular, GolfEngine provides the function of checking policy conflicts when users design security applications and the mechanism to check data storage redundancy. GolfEngine not only prevents the malicious inputting policies but also it enforces the security about network management of network traffic. The results of our simulation underline that GolfEngine provides an efficient, secure, and robust performance for managing network traffic via SDN.
In the third and last part of this dissertation, we analyze the security aspects of battery-equipped IoT devices from the energy consumption perspective. Although most of the energy consumption of IoT devices is due to user interaction, there is still a significant amount of energy consumed by point-to-point communication and IoT network management. In this scenario, an adversary may hijack an IoT device and conduct a Denial of Service attack (DoS) that aims to run out batteries of other devices. Therefore, we propose EnergIoT, a novel method based on energetic policies that prevent such attacks and, at the same time, optimizes the communication between users and IoT devices, and extends the lifetime of the network. EnergIoT relies on a hierarchical clustering approach, based on different duty cycle ratios, to maximize network lifetime of energy-constrained smart devices. The results show that EnergIoT enhances the security and improves the network lifetime by 32%, compared to the earlier used approach, without sacrificing the network performance (i.e., end-to-end delay).