Virginia Tech® home

2022 CCI Student Researcher Showcase

Friday, March 18, 2022

Date:  Friday, March 18, 2022

Time: 10 - 11:30 AM

Location: ISERC, 170 New Classroom Building,
Virginia Tech, Blacksburg campus

In Southwest Virginia, hundreds of graduates and undergraduates are working in labs, internships or training programs that allow them to get hands on with cybersecurity or engage in CCI research. These experiences equip students with highly specialized cybersecurity skills, preparing them to meet the growing demand for cybersecurity talent in the commonwealth and beyond. 

The CCI Researcher Showcase is an opportunity for the students to present their ongoing research projects and share their methods and results to date. There will be awards for the top three posters. 

Titles and abstracts

Bruce Barbour, Virginia Tech, Ph.D. program, Aerospace and Ocean Engineering
Richard Gibbons, Virginia Tech, Ph.D. program, Electrical and Computer Engineering
Joseph Long
Virginia Tech, B.S. program, Electrical and Computer Engineering

5G and 6G satellite networks are becoming more relevant in modern society. Rapid growth in the development and deployment of small satellites (SmallSats) have fueled the rising interest of commercial space launches and operations. This surge in satellite technology fundamentally changes the economics of outer space and presents new opportunities in global, high-speed internet connectivity. The Network Testbed for Small Satellites (NeTSat) collaboratively adesses the challenge of space networking constellations through inter-communication and intra-communication studies of internet satellites. The design of the system focuses on cooperatively merging off-the-shelf hardware and computer simulations in a tabletop configuration to replicate a realistic commercial space network, promoting cheaper access of SmallSat communication studies for researchers and academics. Partnering with the Center for Space Science and Engineering (Space@VT), the National Security Institute (NSI), Wireless @ Virginia Tech, and the University of Surrey Centre for Cyber Security, this hardware-in-the-loop network testbed will demonstrate an internet satellite constellation through transatlantic communication as well as uniquely simulate communications between constellations. We will present the system design, applications, current development phases, and future plans for the testbed. Future studies include SmallSat routing and simulation, inter-connected ground station and LEO satellite protocol studies, disruption tolerant networking systems, integration of cyber security and cyber hacking measures, efficient scheduling of uplink/downlink, custom SmallSat technology development testbed, and exploration of 5G/6G technology for LEO satellites and ground stations.

Faculty:  Samantha Parry Kenyon and Jonathan Black

 

Rahul Varma Chintalapati, Virginia Tech, M.S. program, Electrical and Computer Engineering 

We all know how time consuming it is to wait at traffic intersections. We aim to utilize V2V and V2X communications to make the intersections autonomous and reliable and efficient.

Faculty: Jeffery Reed, Nishith Tripathi, Vijay Shah

Biplav Choudhury, Virginia Tech, Ph.D. program, Electrical and Computer Engineering     

The accommodation of a large number of smart grid devices in 5G leaves a smaller share of resource blocks for time- critical applications such as coordination among inverter-based resources (IBR) of microgrids using ultra reliable low latency communication (URLLC). This makes age of information (AoI) an attractive choice for scheduling IBRs in URLLC to fulfill their latency requirements. This paper evaluates the performance of IBR coordination in microgrids under AoI-based and non-AoI- based 5G schedulers in a co-simulation environment created using PSCAD/EMTDC software and Python. The proposed method for IBR coordination is discussed first. Then, the employed 5G network model and scheduling algorithms are presented. The designed co-simulation environment is discussed next. Finally, time-domain simulation case studies are carried out in PSCAD/EMTDC software to evaluate IBR coordination under the employed 5G schedulers. The results show that the AoI based schedulers offer a better performance for the smart grid applications considered.        

Faculty: Jeff Reed and Vijay K. Shah  

Emerson Dove, Virginia Tech, B.S. program, Electrical and Computer Engineering 

Battledrones is an experiential learn program funded by the Commonwealth Cyber Initiative (CCI). The competition is designed to engage students from institutions across Virginia in autonomous systems and the data and algorithms that drive them. Aligned with CCI’s aims of developing and enhancing experiential learning projects at the intersection of cybersecurity, autonomous systems, and data, this project will build multidisciplinary teams across the Node and Commonwealth to compete in battle one racing competitions.        

Faculty: Kevin Schroeder       

Ehsan Fouladi, Virginia Tech, Ph.D. program, Electrical and Computer Engineering   

Recently, 5G has received significant attention worldwide for being highly reliable and its high data rates. In modern power systems, 5G can be deployed for various purposes, like controlling renewable energy resources. Therefore, detailed studies are required to evaluate the impact of communication systems on power system performance. To perform these studies and analyses, we need to combine power system and communication network simulators. As the fundamentals of these simulators are different, smart approaches should be taken to achieve this goal. One of these approaches to combine different simulators is co-simulation, which is a method that enables simulators from various domains to interact with each other. In this study, we first introduce some of these approaches, including co-simulation, and discuss the advantages and disadvantages of each of them. Afterward, we will discuss one example of co-simulation frameworks called HELICS.    

Faculty: Ali Mehrizi-Sani        

Chirag Gupta, Virginia Tech, M.S. program, Computer Science      

Wireless applications (voice, data, video) and emerging technologies (e.g. the Internet of Things, intelligent transportation systems, unmanned aerial systems) are iving increased demand for wireless data capacity; each wireless device adds yet another instance of radio frequency spectrum usage. Over 70 percent of the global population (and an estimated 29.3 billion IoT devices) will have mobile connectivity by 2023, as reported by Cisco Systems. The increasing number of devices that occupy the spectrum has created a demand for more efficient use of radio frequency bands, especially frequencies that are allocated or assigned for one or more specific uses, yet not used continuously in all locations. Working towards spectrum sharing motivates development of systems to ensure trustworthiness of data used as a basis for granting access to shared spectrum. The Virginia Tech SAS combined with CORNET provides a functional and educational suite of hardware devices and software systems to test spectrum data and cognitive radio functions against malicious users, their data, and validate security applications.

Faculty: Carl Dietrich 

Joseph Harrison, Virginia Tech, B.S. program, Statistics / Computational Modeling and Data Analytics
Mary Neyaro, Virginia Tech, B.S. program, Business Information Technology
Jaden Leonard, Virginia Tech, B.S. program, Computer Science 

This project aims at tracking who sells our information and to whom. This is achieved by signing up to a website/organization with a fake identity. We then trace the mentions of that identity across the web as well use our collected email and voicemail database in order to aw a web of information sharing. Currently, this project is still in the design process – setting up the email database and developing the tools to process it, figuring out how to automatically simulate user activity, and developing procedures that will work on a very large datasets.

Faculty:  Alan J. Michaels       

Naru Jai, Virginia Tech, Ph.D. program, Electrical and Computer Engineering

Citizen Broadband Radio Service (CBRS) band is a frequency spectrum starting from 3550 MHz to 3700 MHz. The usage of this bandwidth is currently governed by FCC through a three-tiered architecture: 1) at the highest tier are the incumbent users (e.g., federal, military), 2) at the middle tier are Priority Access License (PAL) holders (e.g., Verizon, AT&T), 3) at the lowest tier are General Authorized Access (GAA) users who are unlicensed users. The operation and management of the band is centrally performed by a cloud-based service called Spectrum Access System (SAS).  Per FCC, the incumbent users should be provided with robust protection from interference by PAL and GAA users; PAL holders should be provided with robust protection from interference by other PAL holders and GAA users; and the GAA users are not protected from interference by higher-tier users and other GAA users. A key challenge to harness the full potential of the CBRS band is channel allocation for PAL and GAA users. Under this three-tier architecture, the goal of SAS is to maximize spectrum efficiency (in terms of the number of PAL and GAA users that can be supported) while ensuring interference resilience for the incumbent and PAL users. It turns out that this problem involves complex mathematical formulation with a large search space for an optimal solution.  In this research, we present our approach to this problem by considering Navy's shipborne radars as incumbents along the coast of Virginia.  By exploiting the unique geographical features associated with the Dynamic Protection Areas (DPAs) and through several novel reformulation techniques, we show that the complex Mixed Integer Non Linear Program (MINLP) formulation can be converted into a Mixed Integer Linear Program (MILP) with no approximation errors. Through simulation experiments with real-world CBRS map and DPAs along the east coast of Virginia, we show that an optimal solution can be found within the timing requirement set forth by FCC. 

Faculty:  Tom Hou      

Akshay Kumar Jain, Virginia Tech, Ph.D. program, Electrical and Computer Engineering   

Distributed energy resources (DERs) such as photovoltaic systems (PVs) are being deployed in electric grids at an increasing pace. Use of centralized controllers known as DER management systems (DERMS) has been proposed recently to ensure safe operation of DERs.  However, communication channels used by DERMS controllers, along with the unmanned locations where they are deployed, provide a target for cyber attackers to cause adverse grid impacts. These attacks can damage equipment, trip inverters, and cause undesirable operations of control devices. A detailed DERMS cyber-physical system (CPS) model is developed here to identify potential cyberattack paths and determine their severity. Robust and fast acting centralized cyber layer intrusion detection system (IDS) is also proposed to mitigate them within the cyber layer. The proposed IDS can detect and mitigate these cyberattacks before the electrical grid is impacted.      

Faculty: Chen Ching Liu         

Kellie Johnson, Virginia Tech, Ph.D. program, Agricultural, Leadership, and Community Education

Cyber biosecurity and workforce development in agriculture and the life sciences (ALS) is one weak spot in the curriculum at land grant institutions. Students that pursue majors related to ALS often don’t include training in cyber-related concepts or expose the ‘hidden curriculum’ of seeking internships and jobs. Exposing students through course work and preparing them for internships to provide experiential learning opportunities to bridge the two is one area this project aims to fulfill. The objectives of this work are 1) to learn about data security in ALS through class activities and 2) to understand how students apply the knowledge gained participating from the classroom to prepare them for their internship experience.

The course provided concrete learning activities with reflection and discussion in areas of professional development as well as agricultural processes related to cybersecurity and data management. Students worked together to promote learning and engagement from peers of different majors. The instructor provided reflection questions at the end of each class meeting based on the activity for that day.

Experiential learning is a method of teaching that has been shown to create transformative learning experiences for students. Over the course of the class, students developed technical and professional skills, followed by an eight-to-ten-week internship with a variety of industry partners. This study is intended to enhance the growth of the cyber biosecurity field, provide students with applicable skills that are transferable between the classroom and workforce, and also be used as a valuable education tool that various organizations and institutions can utilize as a model to best develop their own educational experiences through experiential.

Faculty:  Tiffany Drape                       

Shaoran Li, Virginia Tech, Ph.D. program, Electrical and Computer Engineering

Mobile Edge Computing (MEC) is a key technology in 5G to provide rich real-time services to mobile users. The main idea of MEC is to install powerful computing and storage hardware close to a base station (BS) at the network edge. By allowing a mobile service to access computing/storage resources faster, MEC can support ultra-low latency applications such as Virtual Reality (VR) and Augmented Reality (AR), autonomous iving, among others.

A fundamental problem in MEC is to determine which tasks should be offloaded to the edge server. This problem is further complicated by the fact that the number of processing cycles of a task is typically unknown in advance, until it has been completed by a processor. That is, at the time of making an offloading decision, there is much uncertainty in estimating the required processing cycles of a task. To adess this uncertainty, we consider probabilistic task deadlines in MEC with limited statistical knowledge of the uncertain processing cycles. We employ chance-constrained programming (CCP) to formulate the problem and design an online solution called EPD -- Energy-minimized solution with Probabilistic Deadline guarantee. We show that EPD is resilient to randomness in task processing cycles (in terms of meeting probabilistic deadlines) and can minimize energy consumption of mobile users.

Faculty: Tom Hou

Weitong Li, Virginia Tech, Ph.D. program, Computer Science      

Despite its critical role in Internet connectivity, the Border Gateway Protocol (BGP) remains highly vulnerable to attacks such as prefix hijacking, to adess this issue, the Resource Public Key Infrastructure (RPKI) was developed starting in 2008, with deployment beginning in 2011.  In recent few years, the RPKI adoption rate among the network resource owners has been rapidly increased by publishing an RPKI object, called ROA, that binds their IP prefixes to themselves; but RPKI can operate only if the routers also verify the incoming BGP announcements with ROAs.

However, it has been known to be challenging to measure route origin validation (ROV) status of ASes since it is not allowed to measure routers directly. To democratize Internet-wide ROV measurement, we present multiple techniques that do not require any IP prefixes to own and control, nor volunteers in ASes that we would like to measure; we also present a simple technique to identify the ROV policy of a given ASes. With these techniques, we characterize the ROV-policy of more than 20,000 ASes and cross check our findings with other techniques.          

Faculty: Taejoong (Tijay) Chung        

Connor Mackert, Virginia Tech, B.S. program, Electrical and Computer Engineering   
T.K Trinh, Virginia Tech, B.S. program, Electrical and Computer Engineering   
Chris Keating, Virginia Tech, B.S.program, Electrical and Computer Engineering   
Daniel Mauro, Virginia Tech, B.S. program, Electrical and Computer Engineering                

The recent implementation of 5G technology has opened numerous opportunities for the advancement of power distribution systems. 5G’s low latency and high bandwidth have allowed new research to arise in remote synchronization of microgrids and nanogrids using network communication protocols. We present synchronization of active power between three nodes (two Distributed Energy Resources and one load) and the power grid using the User Datagram Protocol. We also explore a rudimentary communication-loss backup system, and even faster synchronization using the Precision Time Protocol and the White Rabbit Project. 

Faculty: Igor Cvetkovic 

William Mahaney, Virginia Tech, M.S. program, Mathematics  

Post-quantum cryptosystems are public-key cryptographic schemes whose security relies on computational problems which are hard for quantum computers. One major family of post-quantum cryptosystems are the lattice-based cryptosystems, whose hardness comes from computational problems involving lattices: discrete analogues of real vector spaces. A hard problem of major interest in cryptography is the Closest Vector Problem (CVP), where one is given a lattice and challenge vector in Rn and asked to find the lattice point closest to the challenge vector. To instantiate the lattice-based signature scheme FALCON, there is a need to quickly produce short lattice bases so that a signer can solve the CVP efficiently to produce cryptographic signatures. This work in progress aims to accelerate the key generation process of FALCON in order to better facilitate ephemeral uses of the signature scheme.

Faculty: Travis Morrison        

Ardavan Mohammadhassani, Virginia Tech, Ph.D. program, Electrical and Computer Engineering 

The future power grid envisions a large share of power generation from renewable energy resources in the form of microgrids. Microgrids are a collection of sources, consumers, and electrical equipment that can either purchase power from their nearby utility or supply themselves with their renewable energy resources. Renewable energy resources are intermittent and interface with power electronics. Therefore, we need high-speed and cybersecure communications to control microgrids in a superior fashion. In this project, we have been working on developing a physical microgrid testbed at Virginia Tech that is equipped with 5G through connecting to the 5G testbed at CCI in Arlington, VA. This testbed will allow us to evaluate the practicality of our designs to make sure they can be used in the field.         

Faculty: Ali Mehrizi-Sani

Mpeh Ntantang, Virginia Tech, B.S. program, Electrical and Computer Engineering

The undergraduate researcher will conduct research related to wireless communications and wireless communication testbeds. The research may include cconfiguration, integration, testing, and/or experimental use of one or more of thevfollowing: software defined radios (Ss), S-based wireless testbeds, unmanned aerial vehicles that can carry Ss and other radios, and related hardware and software.

Faculty: Carl Dietrich 

Tristan Dennis, Virginia Military Institute, B.S. program, Computer and Information Sciences
Tanner Mallari, Virginia Military Institute, B.S. program, Computer and Information Sciences
Kolby Quigg, Virginia Military Institute, B.S. program, Computer and Information Sciences

This project proposes a web-based modeling and analytical environment prototype for predicting the spread of malicious behaviors and computer viruses (as contagions) on 5G and other types of networks. We recast networks in an abstract simulated graph form. The goal is to run offline and real-time intensive data analytics and simulations to see the effects of viruses, network topological structures, devices, and applications behavior, and propose intervention strategies by blocking nodes or altering network structure to avoid cascading failures. This work will be beneficial to cybersecurity analysts, researchers, businesses IT professionals, and students. This work will use a novice approach to view networks as a complex system of components (devices) that interact together. The spread of viruses, malware, or other infections will be predicted according to prescribed behavior models assigned to each device. The environment will employ multiple scenarios to provide a threat evaluation and execute blocking strategies automatically. In addition, the environment will have a dedicated dashboard for security analysts to manually input various modeling parameters and evaluate the outcomes, promoting the concept of analyst in the loop. The proposed system will have additional features provided through the dashboard to query and visualize networks, subnetworks, and their properties.

Faculty: Sherif Abdelhamid   

Nitasha Sahani, Virginia Tech, Ph.D. program, Electrical and Computer Engineering

Microgrids in the distribution systems offer flexible and resilient electric energy locally, particularly when main power grid is unavailable. Potential disturbances caused by extreme weather events or cyber-attacks can lead to power supply disruptions from main grid. Microgrids have the capability to sustain critical services without relying on the utility grids. Reliable communication is the key to enable complete observability of the geographically distributed network. Wireless communication has the potential to provide an economic and wider coverage connectivity solution for distributed smart grid components. This study evaluates the efficiency of 5G wireless communication in a distributed grid setup for time-sensitive applications including the islanded operation as a resiliency source and monitoring of microgrid dynamic operations. It focuses on the design requirements and integration capability of 5G with existing distribution system to analyze the impact of 5G in smart grid.  

Faculty:  Chen-Ching Liu        

Matthew Salerno, Virginia Tech, B.S., Electrical and Computer Engineering 

Our server infrastructure for wireless@vt, called Cornet, has seen better days. The high turnover of students administrating the system has led to numerous legacy components, undocumented features, and snowflake servers. This means that new student researchers spend most of their time trying to understand how the current server infrastructure works instead of implementing new features or updating out of date practices. This also makes changing configurations a grueling process as the complexity only increases as our research continues. In a addition, we want Cornet to serve as a valuable educational experience for student administrators. In this project, we examine the underlying issues that makes maintaining a long running server with high administrative turnover so difficult, and we work towards implementing a solution using simple and modern software that is easy to learn and experiment with.          

Faculty: Carl Dietrich             

Wyatt Sweat, Virginia Tech, Ph.D. program, Computer Science      

Current commercial policy/rule-based methods can be inflexible at detecting novel attacks such as ransomware (ex: 2021 Colonial Pipeline hack), supply-chain (ex: 2021 Microsoft Exchange Server, 2020 SolarWinds), insider threats (corporate espionage, accidental data exposure), and advanced persistent threats (large organized groups hacking with a purpose). Likewise a more generic application of machine learning could require extensive training or customer-by-customer level customization, making it infeasible for wide-range implementation. Both these approaches, whether from not knowing the rules or insufficient tuning can leave a client vulnerable to unforeseen threats until after the damage has been done to the company’s IP, reputation, and stock ticker. Our solution uses precise and scalable AI to efficiently comb through the available data searching for dangerous outliers of user or system events to locate threats, novel or known. The setup likewise requires minimal training or customer-level tuning to be effective allowing for efficient and practical deployments.

Faculty:  Danfeng (Daphne) Yao

Stephen Timmel, Virginia Tech, Ph.D. program, Mathematics

When designing error-correcting codes, it is standard to assume that errors in each symbol are independent. The resulting models ignore many real-world instances of channel memory such as burst noise and intersymbol interference, which must often be accounted for separately in implementations. This oversimplification is especially problematic for the family of polar codes, which provide unprecedented asymptotic bounds using a very explicit model of the communication channel. We extend existing results to explore when polar codes can be applied directly to channels with memory, potentially leading to improved performance in 5G applications.

Faculty: Gretchen Matthews 

Pratheek Upadhyaya, Virginia Tech, Ph.D. program, Electrical and Computer Engineering

1) AoI Based RAN Scheduling xApp

Open RAN (O-RAN) is an emerging concept based on openness and intelligence that is poised to have a transformative impact on RAN operability and serve as the heart of future 6G Cellular Networks. The radio access network (RAN) part of the next-generation wireless networks will require efficient solutions for satisfying ultra-reliable low latency (URLLC) services. One such use case is the transmission of time-sensitive information collected by IoT sensors. To ensure timely delivery of information, optimal scheduling policies which ensure the freshness of information need to be employed. In this poster, we demonstrate the closed-loop control of a RAN scheduler in an O-RAN ecosystem. We design an Age of Information (AoI) based scheduler that incorporates user level KPIs to make policy decisions that minimizes the average AoI for all users. We showcase the feasibility of RAN control over the E2 interface by deploying this framework on our O-RAN enabled 5G testbed. We outline some of our design experiences while analyzing the performance of our scheduler when compared against baselines such as round-robin and proportional fairness schedulers.

2) ML Aided Beamforming on Near-Real-Time RIC

In this poster, we demonstrate a Neural Network that performs the calculations for multi-beamforming systems that are able to dynamically adapt to changing user requirements. For instance, if we anticipate a group of users need more throughput, our multi-beamforming system computes antenna weights required to allocate a more directional beam to the aforementioned, thus increasing the overall SINR-MCS rate and therefore throughput. Conversely, if a particular set of users, need to be nulled out, our beamforming network, can add nulls in the direction of said users and mitigate interference. This network finds use in improving uplink SINR, downlink link reliability, multicast systems, UAV communications, ionospheric communications. The novelty in our work is two-fold, the traditional computation of antenna weights for multi-beam forming systems is computationally expensive. Our neural network is trained to replicate these calculations. We find that the neural network’s performance mimics the ideal results very closely while being 2-3 orders of magnitude faster. By integrating this into the Near RT-RIC, we gain the added benefit of performing a coarse performance optimization for the users in question. In the future, we propose to extend this to the Real-Time RIC, where finer corrections can be made.

Faculty: Jeffrey H. Reed, Vijay K. Shah, Aloizio P. DaSilva

Ya Xiao, Virginia Tech, Ph.D. program, Computer Science        

Google and Apple jointly introduced a digital contact tracing technology and an API called exposure notification, to help health organizations and governments with contact tracing. The technology and its interplay with security and privacy constraints require investigation. In this study, we examine and analyze the security, privacy, and reliability of the technology with actual and typical scenarios (and expected typical adversary in mind), and quite realistic use cases. We do it in the context of Virginia’s COVIDWISE app. This experimental analysis validates the properties of the system under the above conditions, a result that seems crucial for the peace of mind of the exposure notification technology adopting authorities, and may also help with the system’s transparency and overall user trust.  

Faculty: Danfeng (Daphne) Yao