Research Interest

My primary research focuses on network and systems security within Cyber-Physical Systems (CPS). Building upon my experience with V2X data offloading in 5G ecosystems and my current work in securing diverse IoT environments, I aim to address the fundamental security paradigms of these critical systems. Specifically, I want to explore the vulnerabilities of networks and systemic flaws in CPS, and how to secure them. I also maintain a strong interest in leveraging machine learning and deep learning to advance cybersecurity measures. Additionally, due to my professional background, I have a keen interest in various software engineering tools and software testing methods.


Research Experience

My research experience started with my undergraduate thesis, where my primary thesis topic was on 5G data offloading for Vehicle-to-everything (V2X) communication. Ideally, vehicles can communicate with each other using 5G base stations (gNBs) or roadside units (RSUs). In our research, we experimented with the optimal probability of distributing the traffic between the gNB and RSU to minimize the packet delay. The modeling was done using M/M/1 queues, and the data offloading considered both communication and computation capacities of the nodes. Network simulation was done using a Python tool called Ciw to validate the model's outcomes. The results from this work were published in IEEE GLOBECOM [2]. 


For my current research, I am working on securing smart environment automation. Smart environments mainly consist of different IoT devices that are interconnected. Recently, there have been numerous works done to make this IoT interaction more seamless by understanding human patterns from different sensor readings and generating automated policies that would trigger based on the user's behaviour. This kind of automation can be easily disrupted by anomalous sensor events or intrusive sensor actuation. My research focuses on two parts: 1. removing anomalous sensor events before generating automated smart policies for smart environments, and 2. detecting and preventing malicious policies once they are generated. The first part of my research was recently published in IEEE CCNC [1], where we have used unsupervised learning methods on resampled data events to detect the anomalous sensor events to improve the automation. I am currently working on the second part, where you secure a generated policy by preventing intrusive traffic and detecting malicious policies.


Apart from my thesis, I have collaborated on different projects from other domains as well. I worked with other peers from one of my courses (Software Quality Assurance) on a project on generating different novel metamorphic test cases for spectral clustering. The work was tested and validated on numerous existing datasets and is currently under review in the Software Testing, Verification, and Reliability journal. I also assisted with analyzing large feedback data using LLMs from an NSF-granted project, which is currently under review at the International Journal of Computer Science Education in Schools. From the six-week workshop done under the NSF ROSE grant, we completed a project on real-time IoT network anomaly detection on lightweight devices. That work is also currently in progress, which we plan to submit in the near future.


Publications

[PDF]