2019 Projects
Communication and Security in IoT
Advisor: Dr. Kim
The Internet of Things (IoT) consists of any device that can be connected to the Internet. These devices can communicate with each other and gather data which can be sent to different networks or analyzed to help someone or solve other problems. With its rapid rise in society, the need for security across machine-to-machine (M2M) communication becomes even more necessary. One communication protocol designed for M2M communication is Message Queuing Telemetry Transport (MQTT). MQTT is a publishing/subscribing communication protocol that uses broker-based communication to create a connection between devices. Broker-based communication uses a message broker to communicate between the publisher and subscriber(s). In this project, we will not only implement MQTT, but we will also add Transport Layer Security (TLS), which provides end-to-end privacy and data integrity over a network. This security measure uses keys to encrypt and decrypt messages over the network, which ensures that attackers will not be able to intercept any of these messages. TLS currently supports Advanced Encryption Standard (AES) 128 with Secure Hash Algorithms (SHA) 256 and AES 256 with SHA 384 in JDK 11. We intend to increase this to AES 384 with SHA 512 so larger messages can be safely sent over the network.
Lecture Notes |
ForensicExaminer for Multimedia of Things (MoT)
Advisor: Dr. Malik
Fake audio detection is expected to become an important research area in the field of multimedia of things (MoT). This project aims to perform vulnerability analyses of voice-driven interfaces in MoT and propose a countermeasure to detect spoofing attacks-- namely replay and voice cloning attacks. We will model replay and speech cloning attacks distortions, and then compare the performance of various machine learning algorithms in terms of accuracy of classification and efficiency. More specifically, using a controlled data set prepared by SMILES Lab, we will explore the suitable algorithms and minimum number of features required to differentiate between first order replay attacks, second order replay attacks, and the original audio in automated speaker verification (ASV)-capable devices. Additionally we will clone the audio using SMILES Lab crowd-sourced data set and compare the cloned audio’s effects on ASV-capable devices with that of replay attacks. The outcome of the project will be a novel, lightweight, and efficient classifier tool that will be able to differentiate between the original, first order replay, second order replay, and cloned audio samples.
Lecture Notes |
Asymmetric Lightweight Centralized Group Key Management Protocol for VANETs
Advisor: Dr. Malik
A Vehicular ad hoc network (VANET) is a dynamic model designed to provide communication between vehicles, with the goal of assisting vehicles with several problems including traffic management and providing safe transportation. In VANET, multicast groups are formed when group of entities (vehicles, RSU) have same application requirements in certain vicinity. These multicast groups rely on paramount security involving the confidential sharing of cryptographic group keys. The group membership frequently change and therefore group keys require updates. Earlier proposed asymmetric lightweight multicast scalable (ALMS) group key management protocol, proposed by members of SMILES lab effectively integrates Chinese remainder theorem, prime factorization, discrete logarithm and noise parameter. The tasks of this project include a) to test effectiveness of ALMS protocol in hierarchical settings, b) and extend the ALMS (e-ALMS) for hierarchical setting to further improve ALMS in terms of scalability. Next, we will identify appropriate network simulator (such as ns3 , OPNET) to compare extended-ALMS with existing protocols and Native-ALMS.
Lecture Notes |
Lightweight Authentication in IoT Applications designed using Information Centric Networking
Advisor: Dr. Sen
The Internet of Things (IoT) is a field with applications based in healthcare, industries, and smart cities and is largely attributed to the service of sharing information between billions of heterogeneous devices that can be both resource-constrained or high-powered. A new model called Information Centric Networking (ICN) has been proposed for the IoT as an alternative to the current IP-routing method. The difference with the IP-routing method is that ICN does not have IP addresses, so instead of focusing on identifiers on the source and receiver, it has a primary focus on the data packets being sent. These data packets can be cached within different nodes in the network and can thus be retrieved by name, since they may not be contained at the source node and thus does not require a source identifier. However, since ICN is message/data packet centered, there is a need to authenticate the packet of data while also retaining its integrity. Cryptography is then a natural choice since it allows both authentication of the packet and the ability to retain its integrity. However, since the IoT contains many resource-constrained nodes where packets are able to be sent/received from, there is a need for this encryption/authentication scheme to be lightweight and optimized in terms of energy and resource consumption.
Lecture Notes |
Improving FlowDroid's Taint Analysis Using String Information
Advisor: Dr. Lu
Android offers a popular platform for the creation and use of mobile applications, but with this popularity has come the risk of endemic security issues. Through malicious or careless design, Android apps may leak a user’s private information to untrusted sources by exploiting broad security permissions. Methods of static analysis for Android applications, such as the taint analysis provided by the open source FlowDroid project, can identify the flow of confidential information to unsafe locations as a way of testing a program’s security. However, the abstraction techniques required by static analysis results in a loss of precision, increasing the number of false positives. By refining the string analysis methods within the preexisting FlowDroid tool, we hope to increase the precision of its security analysis.
Lecture Notes |
Trajectory Privacy in Autonomous Vehicles Communications and Applications
Advisor: Dr. Fu
Vehicular ad-hoc networks (VANETs) allow vehicles to communicate with infrastructure as well as with one another, which allows for navigation of autonomous vehicles (AVs) and improved safety and traffic management. However, these networks are susceptible to passive attackers who aim to ascertain and misuse vehicles’ location data through eavesdropping. In order to ensure location privacy, it is necessary to utilize macroscopic location privacy without sacrificing safety and quality of service for the user (in this case, the passenger of the AV). We aim to use the pre-defined k-anonymity protocol in a simulated environment to test optimal mix-zone locations, the incorporation of dummy trajectories, and alternate/redirected paths, in order to maximize privacy throughout the AV’s entire trajectory.
Lecture Notes |
