Collins College of Professional Studies Student Research Archive

The IAEM Region #2 newsletter (Fourth Quarter 2023) contains contributions from (2) of our Homeland Security, Doctor of Professional Studies students: Shayla Clarke & Michael Etzel. 

Click here to read more about their contributions.

Personalized Outsourced Privacy-Preserving Database Updates for Crowd-Sensed Dynamic Spectrum Access L Truong, E Troja, N Yadav, SAC Bukhari, M Aliasgari 21st IEEE Mediterranean Electrotechnical Conference (MELECON)

The wireless networks people use everyday depends on Radio Frequency (RF) technology. Radio frequencies are a finite resource that is managed by the Federal Communications Commission (FCC). The explosive popularity of the Internet of Things (IoT) wireless devices, like Google Home stations and household appliances, have resulted in a severe spectrum shortage. Dynamic Spectrum Access (DSA) through Cognitive Radios (CRs), which can detect and utilize which channels are not currently occupied, have been successful in mitigating the spectrum shortage problem. Such methods allow a user to dynamically share the spectrum with other (secondary) user. The paper focuses on the crowd-sensing approach of DSA architectures. Crowd-sensing is a method of collecting information through different types of mobile devices. For instance, Google Maps utilizes crowd-sensing to display the amount of traffic in a certain area. In this approach, secondary users issue location-based queries to a crowd-sensing server. The crowd-sensing server maintains an updated database through query outsourcing and replies to the secondary user request. Our paper proposes a method to save crowd-sensed resources by allowing the server to update its geospatial records based on a priority queue while maintaining privacy. This means that the server can still be updated by remaining oblivious to the location of the crowd-sensing worker secondary user. The simulation described in the paper shows how much space is unused within a 100m by 100m area. The problem here is the higher the number of users, the lower the percentage of populated tiles. This highlights the need for a personalized crowd-sensed approach to give higher concentrated areas priority to query responses. The personalized scheme is secure and efficient, as shown in the paper.

Machine Learning Architecture for Signature-Based IoT Intrusion Detection in Smart Energy Grids N Yadav, L Truong, E Troja, M Aliasgari 21st IEEE Mediterranean Electrotechnical Conference (MELECON)

With the overwhelming presence of Internet of Things (IoT) devices comes a new gateway for malicious threat agents. Self-propagating malware called botnets are capable of automatically infecting IoT devices, which include routers, IP cameras, smart home appliances, and much more. Modern frameworks such as Network Intrusion Detection Systems (NIDS) and Network Intrusion Prevention Systems (NIPS) are designed to detect and prevent these attacks. These architectures can be signature-based or anomaly-based. Signature-based systems offer higher detection rates at the cost of tedious manual work and are incapable of learning through network traffic. For an attack to be detected, its signature must match what is known in the internal lookup database of the system. On the other hand, anomaly-based systems mitigate the shortcomings of signature-based systems at the cost of high false-positives rates. The paper proposes an approach using supervised and unsupervised machine learning techniques to automate rule generation for the signature-based approach. For our research, we extract data from the packet capture (PCAP) files of benign traffic and Man-in-the-Middle (MITM) attacks using a program called CICFlowMeter. We then conduct statistical significance tests on what we calculate to be the most significant attack features according to the CICFlowMeter output. In future work our architecture’s Rule Generator and Data Aggregator and Labeler will be fully realized to dynamically generate Suricata rules. The use of deep learning approaches will also be explored and become necessary in a real world scenario with high volume network traffic.

Dr. Erald Troja, coordinator of the Cybersecurity program and the Cybersecurity Center of Academic Excellence, along with Computer Science Associate Professor Dr. Nikhil Yadav and fourth year Cyber Security Systems student Laura M. Truong ’22CCPS, had their joint paper, “Personalized Outsourced Privacy-Preserving Database Updates for Crowd-Sensed Dynamic Spectrum Access” accepted for publication in the highly rated IEEE MELECON22 interdisciplinary international flagship conference. 

Click here to learn more their research as well as other papers published from CCPS Faculty in the Department of Computer Science, Math and Science.

Computer Science graduate student Christopher Singh, along with Professor Nikhil Yadav's research paper was accepted to the HICSS (The Hawaii International Conference on System Sciences 2022). This is a premier conference in Information Systems, the Sciences, and IT management. The project was a 6-7 month internal research effort where they looked at building a machine learning infrastructure to study mental health in NY during COVID-19’s first wave. 

Faizan Ahmed, a student in The Lesley H. and Willliam L. Collins College of Professional Studies, along with Fazel Keshtkar, Ph.D., Associate Professor, Division of Computer Science, Mathematics, and Science, and Syed Ahmad Chan Bukhari, Ph.D., Assistant Professor, Division of Computer Science, Mathematics, and Science, and Director, B.S. in Healthcare Informatics, recently published a research article, “A Deep Learning Approach for COVID-19 & Viral Pneumonia Screening with X-ray Images.”

Click here to read more on their findings!