This REU Site project at Montclair State University recruits 10 undergraduate students and nurtures them with immersive research experiences in the broad areas of Cybersecurity. Participants will closely work with MSU faculty mentors on cutting-edge research projects, enrich cybersecurity knowledge, increase their creative capacity and confidence in cybersecurity R&D, cultivate a strong desire to apply for advanced degrees in the related fields.
The REU Site aims to provide advanced research experiences to participating students in three focus areas of Cybersecurity:
Each selected REU student may choose one project, working alone or in groups, mentored by 1-2 faculty members. The list of projects for 2021 summer program are highlighted below.
Due to the requisite nature of user privacy and recommendations in OSNs, there is a strong need to design privacy-enhanced OSNs. This project will explore private friend recommendations as it is considered to be the first service in OSNs for identifying new friends to a given target user based on different metrics such as profile similarity and geographical locations. Students will develop a privacy-preserving friend recommendation protocol for social networks in outsourced environments by analyzing the trade-offs among various performance factors, such as accuracy, security, and efficiency.
More and more corporations choose to outsource their data analytics tasks to third-party cloud-based service providers (e.g., Databricks). Even though outsourcing provides a cost-efficient solution, it raises serious security concerns. There are multiple reasons for the server to provide wrong results, either due to software bugs, or the intention to save computational power by terminating the training program early. Therefore, it is necessary to verify the data analytics results returned by the server before making strategic decisions based on them. This project focuses on designing efficient authentication protocols for mainstream machine learning algorithms, such as random forest and deep learning.
The dependency on online social networks (OSN’s) and their ability to rapidly spread information provides an opportunity for social bots to control and manipulate the information we read. The aim of this project is to investigate methods to detect bots (software programs) and misinformation. How much information do you read online is created by bots? How do you know if the information you read online is “fake news?” What algorithms or techniques can help us identify bots? These are some questions we will investigate in this project.
Although Big data processing frameworks have been widely adopted, the security issues in such large-scale systems have not been well studied yet. This project will investigate a potential attack from a compromised internal node against the overall system performance. A set of execution features will be designed for our machine learning models including the fine gridded progress of different types of tasks on each server of the cluster. The abnormal detection will be combined with the speculation strategy of the native Hadoop system to furthermore block the attacked server.
Many security vulnerabilities found in software systems can be traced back to the mistakes committed by the programmers during the software development process. This project will study programmers’ mistakes from the Cognitive Psychology perspective of human error. Cognitive Psychologists have studied how people make errors while performing different types of tasks. This project will study programmer behavior in light of some of the prominent human error theories.
Software security is a major concern of the developers who intend to deliver a reliable software. the goal of this research is to make the current software metrics more security-centric by identifying threshold values for the metrics based on which a file can be interpreted as a vulnerable file. This research will assist the developers in secure software development through the security assessment of their code using the values of the metrics.
Robots are widely employed in strict and complex hybrid assembly tasks involved in smart manufacturing. The security and efficiency of the human-robot collaboration system have a significant impact on the task quality. we will develop an effective and novel teaching-learning-collaboration (TLC) framework for the robots to learn from the human demonstrations to and collaborate with the human actively in collaborative tasks through natural language.