Courses

CMSC-691-002 - DP TOP: AI for Networking & Edge Systems - Fall 2024

This course aims to provide an in-depth exploration of the latest research and developments in the rapidly-evolving fields of networking and edge systems, focusing particularly on the transformative impact of AI and machine learning technologies. Equipped with both theoretical and practical components, this course is aiming to provide you with an extensive understanding of the state-of-the-art AI/ML methodologies applied to networking and edge systems. Throughout interactive and engaging lectures, we will delve into cutting-edge topics such as ML/AI fundamentals for networking, edge computing, network performance optimization, ML for IoT, security, and privacy issues in AI-enabled networking.

CMSC 440: Data Communication and Networking - Spring 2025

This course explores computer networking, focusing on the applications and protocols that run on the Internet. We will take a top-down approach to the layered network architecture, studying applications first and then proceeding down the network “stack” towards the physical link. We will look at the operation of applications such as the web, FTP, e-mail, and DNS. At the transport layer, we will study both connectionless UDP and connection-oriented TCP. Since TCP is the protocol that the majority of Internet traffic uses, we will study its operation in-depth, including flow control and congestion control. We will also look at how data is routed through the Internet, regardless of transport protocol. We will also introduce current “hot” topics, such as network security and wireless/mobile network.

Education

University of Maryland, College Park

Ph.D.
Computer Science
2006

Patents

Transparent Digital Rights Management for Extendible Content Viewers

https://patents.google.com/patent/US7171558B1/

2007-01-29

A digital rights management system for controlling the distribution of digital content to player applications. The system comprises a verification system, a trusted content handler, and a user interface control. The verification system is provided to validate the integrity of the player applications; and the trusted content handler is used to decrypt content and to transmit the decrypted content to the player applications, and to enforce usage rights associated with the content. The user interface control module is provided to ensure that users of the player applications are not exposed to actions that violate the usage rights. The preferred embodiment of the present invention provides a system that enables existing content viewers, such as Web browsers, document viewers, and Java Virtual Machines running content-viewing applications, with digital rights management capabilities, in a manner that is transparent to the viewer. Extending content viewers with such capabilities enables and facilitates the free exchange of digital content over open networks, such as the Internet, while protecting the rights of content owners, authors, and distributors. This protection is achieved by controlling access to the content and constraining it according to the rights and privileges granted to the user during the content acquisition phase.

A New Scheme for Operating a Wireless Station Having Directional Antennas

US7630343B2

2009-12-07

Disclosed is a wireless station having a receiver and a network allocation vector associated with each of a plurality of directional antennas. The receivers concurrently listen for frames from remote stations. When any receiver detects a frame from a remote station, the receiver activates its associated NAV. The station has one or more transmitters that can transmit using the antennas. While transmitting a signal using an antenna, the receivers associated with any non-transmitting antennas continue to listen for signals from remote stations. The station cancels any signals from the transmitting antenna received by the non-transmitting antennas. To perform the cancellation, the station performs a self-calibration procedure. The station can self-calibrate by either silencing neighboring stations or by inserting null tones into a transmitted calibration signal.

Estimation of Travel Times using Bluetooth

US8519868B2

2013-08-26

Methods for estimating travel time using at least two remote systems to record the timestamps associated with obtaining identifying information of a wireless Bluetooth enabled, or other WPAN technology, electronic device in a vehicle. A remote system in one embodiment is a Bluetooth enhanced traffic controller. Characteristics of Bluetooth technology, such as a unique address for each Bluetooth capable device are used to detect a vehicle with a Bluetooth device by at least a first and a second remote system. Vehicle identifying data including at least a time stamp is transmitted by the remote systems to a central system. The central system determines a travel time, or an estimated travel delay. Travel time related data is provided by the central system to a display, such as a variable or dynamic message sign.

Data Collection and Traffic Control using Multiple Wireless Receivers

US8566011B2

2013-10-21

Methods, systems, and devices for monitoring roadway traffic. A method includes transmitting wireless signals from a plurality of roadside equipment (RSE) devices, including from a first RSE device and from a second RSE device that are located at separated positions of an intersection. The method includes receiving responses by the first RSE device and second RSE device from a wireless device. The responses include a unique identifier corresponding to the wireless device. The method includes determining a signal strength of each of the responses by the first RSE and the second RSE and transmitting data from the first RSE device and the second RSE device to a control system. The data includes the unique identifier, the signal strength of each of the responses, and times that the responses were received. The method includes determining traffic information associated with the wireless device based on the received data.

Mobile Sensing for Road Safety, Traffic Management, and Road Maintenance

US8576069B2

2013-11-04

Mobile monitoring systems and methods are disclosed. In accordance with one aspect of the invention, the system includes a plurality of vehicles and a base communication station that are in communication with each other. Each of the vehicles includes a camera that generates image data, a location device that generates geographic coordinates of the vehicle, a computing device that receives the image data from the camera and the geographic coordinates of the vehicle and forms a processed image signal that includes the image data, the geographic coordinates and a time stamp, and a communication device that receives the processed image signal from the computing device and wireless transmits the processed image signal to the base communication station. The base communication station receives the processed image signal. The base communication station can include an image processor to further process the processed image signal from each of the plurality of vehicles to form an output signal and a transmitter that transmits the output signal. The output signal can be used to control traffic control devices, vehicles and to provide other useful information.

Video Streaming at Mobile Edge

US11128682B2

2021-09-20

Aspects of the subject disclosure may include, for example, a method comprising sending context information from a mobile wireless device through a control channel to a network server; receiving a policy at the mobile wireless device from the network server, wherein the policy assigns a video streaming bit rate to the mobile wireless device based on the context information; and implementing the policy to control a video streaming session between the mobile wireless device and a media server over a data channel. The context information may include information about the mobile wireless device and/or a user of the mobile wireless device. The policy may be different for each mobile wireless device. Other embodiments are disclosed.

Research Focus

Generative AI for Robust Communication and Efficient Networking

Generative AI is revolutionizing communication by enabling intelligent, automated content creation across text and multimedia, improving efficiency, accuracy, and personalization. It supports diverse applications, from customer service to real-time translation, while advancing technologies like transformers and reinforcement learning to enhance global knowledge sharing. We introduce LoRaFlow, a generative modeling approach using diffusion transformers and rectified flow to reconstruct LoRa signals from noisy inputs, improving demodulation. Integrated with existing LoRa infrastructure, LoRaFlow enhances signal recovery under low SNR conditions with minimal hardware changes, extending the range and reliability of IoT communications.

AI-Based Tactics for Advanced Network Deception

Modern computer networks, characterized by their high connectivity and heterogeneity, integrate a wide range of devices and protocols such as IoT, sensors, and robotic systems. While this complexity enables advanced services and adaptability, it also introduces significant security vulnerabilities, particularly in critical environments like the Internet of Battlefield Things (IoBT). To address these challenges, this project explores advanced cyber deception techniques aimed at mitigating reconnaissance-stage cyber threats. Two key contributions include MirageNet, a GAN-based framework for generating synthetic network traffic to enhance security and privacy, and MiragePkt, which synthetically generates network packets by learning from raw data. Additionally, a novel Graph Neural Network (GNN)-based approach for optimizing honeypot placement demonstrates superior performance over traditional methods, achieving 92.34% accuracy and a 139x speedup in inference. Together, these innovations highlight the potential of AI-driven solutions in fortifying modern network defenses.

HomePal: Developing a Smart Speaker-Based System for In-Home Loneliness Assessment for Older Adults

Experiences of loneliness are prevalent among older adults. Loneliness is a painful and pernicious state occurring when there is a perceived discrepancy between one’s optimal levels of social interactions and actual social relationships. Lonely older adults are more likely to experience functional decline, including activities of daily living, mobility, and stair climbing. Loneliness has been associated with negative health outcomes, such as increased morbidity and mortality, dementia risk, and cognitive impairment. With longer life expectancy rates, the number of older adults at risk for loneliness will increase substantially, presenting challenges to the healthcare system. However, studies reported that loneliness is a reversible condition, which could be achieved through appropriate interventions, such as improving physical health and social relationships. Detection measures are imperative to identify both lonely and at-risk individuals early enough to intervene before adverse health outcomes occur. The assessment of loneliness is challenging due to the social stigma of being labeled as lonely, resulting in underestimation. Although multidimensional scales exist that do not explicitly use the word “lonely,” healthcare providers do not routinely assess loneliness using validated instruments, making it challenging to be aware of the extent of loneliness among their older adult patients.

ClassyNet - Class-Aware Early Exit Neural Networks for Edge Devices

Edge-based and IoT devices have experienced phenomenal growth in recent years due to rapidly increasing demand in emerging applications that utilize machine learning models such as Deep Neural Network (DNN). However, one of DNN's main drawbacks lies in the large storage/memory requirement and high computational cost, which become a major challenge in adopting these models to edge devices. This led to the development of early-exit models such as BranchyNet that allow for the decision to be made on earlier stages by attaching dedicated exits to the inner layers of the architecture. However, existing early-exit models do not have control over what class should exit when. The need for these novel class-aware models can be observed in multiple edge applications where specific important classes need to be detected earlier due to their temporal importance. In this project, we aim to develop and design ClassyNet, the first early-exit architecture capable of returning only selected classes at each exit. This allows for speedups in inference time for sensitive classes by enabling the first layers to be deployed on edge devices, saving significant computational time and edge resources while maintaining high accuracy.

MedKnights - Towards Secure and Flexible Medical IoT (IoMT) Infrastructure using Generative Adversarial Network

With growing healthcare demands, the Internet of Medical Things (IoMT) has grown significantly in recent years and is dominating the healthcare industry. However, these smarter and advanced medical devices are “more” complicated in terms of software and hardware, with several defects and vulnerabilities that have been found and can lead to possible malicious attacks. Healthcare organizations are the new focus of attackers for carrying out IoMT-focused cyberattacks, which are becoming more common. In recent years, ransomware and distributed denial of service (DDoS) attacks are malware-based popular attacks on IoMT devices. Cyberattacks and disruptions in clinical care can have a catastrophic effect on patient safety, which trickles down to the medical staff’s responsiveness. Moreover, because different medical devices have varying vital capabilities, it is critical to enable differentiated network services for these devices with varying critical levels of operation under network dynamics. Hence, it becomes crucial to efficiently detect and identify any malicious network activities to eliminate or minimize the impact of these attacks, as well as to detect and identify network traffics belonging to different medical equipment.

DeepMAC – Towards A Deep Learning-Based Framework for Automated Design of Networking MAC Protocols

Networking protocols, practically, are designed through long-time and hard-work human efforts. However, these designed protocols, typically, are non-optimum with limited flexibility under several network scenarios and conditions. Moreover, due to evolving network technologies as well as increasing demands of modern applications, ”general-purpose” protocol stacks are not always adequate and need to be replaced by application tailored protocols. Therefore, replacing this inefficient human-based protocol designing process by a novel paradigm that enables rapid design of efficient, flexible, and high performance protocols that intelligently adapt to different device characteristics, application requirements, user objectives, and network conditions is highly desired. In this project, DeepMAC, we explore the first basic steps toward our vision of replacing the human driven network communication design by machine using ML techniques. This vision considers an intelligent system that automates the design of on-line adaptive protocols only by interacting with and learning from the environment, without having any prior knowledge. In this envisioned framework, network protocol stack is decomposed into core functionalities (e.g., switching, routing, congestion control, reliable connection, Backoff, etc.) in which the intelligent agent designs an efficient protocol by selecting the optimum set of functionalities in response to device characteristics, application requirements, user objectives, and network conditions.

Tracking and Diverting All-You-Can-Eat Cafeteria Food Waste and Loss from an Urban Public University in Central Virginia

Virginia Commonwealth University (VCU) is launching an innovative project to tackle food waste and insecurity, positioning itself as a leader in sustainable food management. Central to this initiative is the implementation of advanced food waste tracking technologies, including RFID-tagged plates and machine learning tools, in Shafer Dining Hall. The project will divert unused food to on-campus Ram Fridges, community refrigerators managed by RVA Community Fridge, and older adult living facilities linked to VCU Health System. Through education, outreach, and real-time data displays, VCU aims to foster a culture of sustainability on campus. This interdisciplinary effort will serve as a model for other institutions, with plans to expand to hospital and athletic facilities, promoting long-term environmental responsibility and food equity across the Richmond region.

Other Research Projects

View a complete list of research projects here.

Research Grants

CC*IIE Networking Infrastructure: CHARMED – Campus High-Availability Scientific Research Environment and DMZ

National Science Foundation

2014-12-31

The project builds a full featured "scienceDMZ" that leverages the Old Dominion University's investment in operational high availability resources in the existing data centers routing and switching fabric, incorporates an enhanced and scalable High Performance Computing (HPC) model, and facilitates data intensive research while precluding the need for science projects to create their own stand alone environments. This high performance and high availability computing and networking environment accelerates scientific discovery by increasing the rate at which data can be transferred between collaborators, and by facilitating the creation of communities focused on the data. This project represents a significant shift in promoting and facilitating greater scientific research through collaboration by accommodating the needs of data driven discovery in Physics, Oceanography, Computer Science and Engineering, along with researchers from across campus. Three scientific domains are selected as first users to show the impact of the developed enhanced scientific research environment. The developed environment will have a broad impact on scientific discovery by being a common reliable infrastructure for scientists to host collaboration environments in support of their works. It is envisioned that this infrastructure could be more broadly utilized to accommodate further "Big Data" science in social sciences, biomedical research and public health. This project provides opportunities for enriching the undergraduate and graduate curricula and for exposing students to cutting-edge research in different science domains. Moreover, all of the designs, documentation, and developed tools are made available for other institutions.

EAGER: Bluetooth Open-Source Stack (BOSS) - A Flexible and Extensible Bluetooth Research Platform

National Science Foundation

2014-11-30

Bluetooth technology continues to evolve and expand, taking advantage of the desirable attributes and features it possesses in comparison to other wireless technologies. With the latest update to the Bluetooth specification (version 4.1), Bluetooth devices are expected to become major players in the much-hyped Internet of Things market. A large body of the research community, utilizing this technology, will require modifications to the lower layers of the Bluetooth protocol. Given the lack of an open-source implementation of Bluetooth stack, the community efforts will be limited in validating and verifying their research under realistic scenarios. This ambitious project aims to design, develop, and disseminate a flexible and extensible open-source Bluetooth platform (BOSS) that will enable new research opportunities for the wireless and mobile computing community. The platform will enable development and evaluation of schemes, services, and applications across all layers of the Bluetooth stack, through the creation of a community-maintained, open-access repository. These goals will be accomplished by: 1) developing a complete Bluetooth open-source platform, by implementing lower layers of the protocol stack as firmware on an existing open-source Bluetooth hardware platform, and extending existing open source upper layers of the stack; and 2) creating and maintaining an open-access repository, where the community will play a key role in generating novel open-source modules.

This project will empower the network community with a tool-set facilitating the development and evaluation of new Bluetooth protocols, applications and services. These new protocols and services could be used to contribute to future Bluetooth standards. BOSS project will have an impact on multiple areas in wireless networks and mobile computing research, such as protocol design, security, wireless network coexistence, and multi-interface smart devices, among others. The project will expose students to cutting-edge research in wireless networks, mobile computing, and computer architecture. In addition, the project?s outcome will be utilized in lectures targeting high school students to educate them about Bluetooth technology and its application as an example of wireless technology as well as Internet of Things. Results from this work will be disseminated through presentations at conferences, journal publications, and reports. An open-access repository, which includes all the designed an

Bridging the Disciplinary Gaps in Cybersecurity Curricula through General Education, High Impact Practices, and Training for Incoming Freshmen

US National Science Foundation

2017-07-31

Experts agree that cybersecurity is a multifaceted problem that should be addressed through an interdisciplinary framework; however, most of cybersecurity curricula are offered within specific disciplines. Developing interdisciplinary programs can be challenging with issues such as: student isolation, a disconnect between faculty from the disciplinary programs, a lack of coordination in course delivery, gaps in providing student support, and a lack of a basic interdisciplinary foundation among faculty and students alike. To address these challenges, the "Bridging the Disciplinary Gaps in Cybersecurity Curricula through General Education, High Impact Practices, and Training for Incoming Freshman" project will fully integrate student-focused high impact practices into an interdisciplinary cybersecurity major and minor at Old Dominion University. In addition, an interdisciplinary cybersecurity training module for incoming freshmen will be developed and delivered, and another interdisciplinary general education cybersecurity course will be made available through open education resources (OER). Developing an interdisciplinary general education cybersecurity course will provide students the foundation they need to learn about the topic through a multi-faceted lens.

Integrating high impact practices into interdisciplinary curricula is a novel approach that will bolster the quality of cybersecurity academic programs. These will include learning communities, service learning, internships, undergraduate research projects, and e-Portfolios. The learning communities will include freshmen learning communities, sophomore learning communities, and living learning communities. Each learning community will have a peer mentor assigned. A process will be developed so cybersecurity majors can develop an electronic portfolio in their first-year general studies cybersecurity course, providing a framework for usage over the course of their studies. The development of an open education resource (OER) general studies cybersecurity course will provide a framework that other higher education institutions can use to develop similar courses. Integrating high impact practices into the curricula will provide a framework that can be replicated in other institutions. Graduates of these curricula will be able to fill those occupational cybersecurity positions that require an interdisciplinary background in multiple fields, as well as strong communications and critical thinking skills. In

STTR: Drone-Delivered Naloxone for Opioid Overdose Treatment by Bystanders

National Institutes of Health

2020-06-30

One hundred thirty Americans die from an opioid overdose (OD) daily. The key to improving this survival is
the timely delivery of the antidote naloxone. Current US Emergency Medical Services (EMS) systems are
often unable to deliver the life-saving drug fast enough manner to save the victim’s life. Our uniquely-qualified
team (Virginia UAS, LLC commercial drone pilot training instructors; a Virginia Commonwealth University
(VCU) faculty physician EMS medical director/jet and commercial drone pilot with extensive experience
leading high-impact, multicenter, prehospital NIH clinical trials; and VCU engineers with experience in
development and commercialization of drone hardware and software) has developed a novel prototype drone
system that is interfaced to Richmond’s 9-1-1 dispatch computer. The system automatically loads a 9-1-1
bystander caller’s GPS phone location into a drone and launches it to the scene of an opioid OD carrying a
lightweight, removable, patent-pending, real-time audio/video package (COMMRx) containing naloxone nasal
spray. Just five drones in the City of Richmond could ensure delivery of the naloxone package to bystanders
in ≤2 min as EMS first responders begin racing to the scene. Trained 9-1-1 pilot dispatchers would direct the
bystander to remove COMMRx from the drone, take it to the victim, and administer the naloxone spray long
before first responders can physically reach the victim. This Phase I project proposes to: 1) finalize the
system customization; 2) train/FAA-certify seven 9-1-1 dispatchers drone pilots; and 3) perform realistic
simulation testing of the system to guide design of a Phase II clinical trial to take place in Richmond,
Henrico/Chesterfield counties, and Roanoke, VA. This novel system developed and executed by a uniquely
qualified team shifts the paradigm of prehospital emergency care delivery by empowering and directing
laypersons to initiate time-dependent, life-saving treatment to opioid OD victims long before EMS personnel
can physically treat the patient. Our commercial goal is to expand our flight school business to provide
similar capability throughout the 5,783 9-1-1 centers nationally.

NeTS: Small: SMILE -- Towards Smarter Network Edges for Next Generation Networks

National Science Foundation

2017-10-31

As the number of smart devices and their applications continue to grow, transmission of mobile traffic data over wireless links (i.e., Wi-Fi and cellular links) is continuing to increase significantly. To cope with the explosion of smart devices coupled with a growing proliferation of cloud or edge-based applications, best-effort Quality Of Service is no longer a satisfactory solution and a new breed of intelligent networks is required. More specifically, it is now necessary to have greater visibility and control over the traffic generated from the smart devices in order to deliver optimal performance and a high Quality of Experience to a variety of users and applications. Software Defined Networks (SDN) are a way to facilitate wired network evolution by enabling flexible, controllable, and efficient networks. Motivated by the unique difference of wireless (vs. wired), this project aims to develop and evaluate SMILE (SMart and Intelligent wireLess Edge framework) that supports SDN-like structure at network wireless edge that include both wireless access devices (e.g., Wi-Fi Access Points (APs)) and end-devices. Extending SDN paradigm all the way to smart end-devices would realize our objective in making the network, both wired and wireless, open, flexible and programmable to provide truly end-to-end management and control in which users could reap the full benefits of SDN. The proposed research is expected to advance the state of the art of networking and mobile computing.

The proposed project, SMILE, aims to achieve the following main goals: i) Design and development of SMILE framework that supports wireless networks to be open, flexible, and programmable to cope with complexity, dynamic nature, and uncertainty associated with wireless networks, ii) Draw on the developed SMILE framework to design and develop three different network services such as edge-based fine-grained context-aware video adaptation service, and iii) Implement and evaluate the developed smart protocols and services on medium-scale testbed of smartphones and Wi-Fi APs as well as simulation tools. All schemes/protocols, software development, testbed realization, experimental results will be made open available to the research community. This project will also carry out a number of educational activities involving K-12, undergraduate, and graduate students, make strong outreach efforts for recruiting and mentoring under-represented students, and enriching undergraduate and graduate curri

MRI: Track 1 Acquisition of NVIDIA DGX H100 GPU system for research and education at VCU

US National Science Foundation

2023-06-30

Graphic Processing Units (GPUs) have become an essential computational tool in support of a wide range of research areas. This project funds the acquisition of a high-performance GPU cluster to be installed at the Virginia Commonwealth University (VCU) High-Performance Research Computing (HPRC) core user facility, which provides shared university-wide access to high-performance computing clusters for research and education purposes. The GPU instrumentation will benefit the research and training activities of 22 research groups and impact 400+ users across the entire university. The acquisition advances next-generation research in artificial intelligence, modeling, and simulations across a broad range of disciplines including computer science, engineering, life sciences, and physical sciences. Further benefits include those to cutting-edge research, research training of graduate and undergraduate students, course and curriculum development, K-12 outreach impacts, and the participation of underrepresented minorities. The new resources will also strengthen interdisciplinary research and broaden collaborative activities across disciplines with local industry and federal laboratories.

The acquisition of a NVIDIA DGX H100 GPU system will significantly expand the GPU computing capacity at VCU to help satisfy the increasing high-performance computing demand across various science & engineering disciplines. The key research projects that will benefit from this equipment are in the areas of machine learning for high-speed data streams, deep learning and quantum machine learning, neuromodulation (transcranial magnetic stimulation and deep brain stimulation), computational chemistry (quantum chemistry), optics and defects in semiconductors (gallium nitride), software vulnerability prediction using deep learning, nanomagnetic materials and nanoscale magnetic devices, real-time pandemic management architecture, secure and flexible medical IoT infrastructure (protection from cyberattacks deployed at network edge components), and ultrafast dissociation dynamics of isolated organic molecular cations. This project will also support the training of high-school, undergraduate, graduate students and postdoctoral fellows in cutting-edge interdisciplinary research spanning computer science, engineering, physics, and chemistry, strengthening existing ties with nearby HBCUs, as well as expanding K-12 minority outreach programs.

OAC Core: MedKnights - Towards Secure and Flexible Medical IoT (IoMT) Infrastructure using Generative Adversarial Networks

US National Science Foundation

2022-06-30

With growing healthcare demands, the Internet of Medical Things (IoMT) has grown significantly in recent years and is dominating the healthcare industry. However, these smarter and advanced medical devices are very complicated in terms of software and hardware, with defects and vulnerabilities that are found regularly and they are also vulnerable to possible malicious attacks. Healthcare organizations are the new focus of attackers for carrying out IoMT-focused cyberattacks, which are becoming more common. In recent years, ransomware and distributed denial of service (DDoS) attacks are malware-based popular attacks on IoMT devices. Cyberattacks and disruptions in clinical care can have a catastrophic effect on patient safety, which trickles down to the medical staff?s responsiveness. Moreover, because different medical devices have varying vital capabilities, it is critical to enable differentiated network services for these devices with varying critical levels of operation under varying network dynamics. Hence it is crucial to efficiently detect and identify any malicious network activities to eliminate or minimize the impact of these attacks, as well as to detect and identify network traffic belonging to different medical devices.

The MedKnights project will extend knowledge in networking, machine learning, and the digital forensic domains in developing a holistic framework that can be deployed at the network edge components and effectively support different fine-grained security services for IoMT networks and devices. The main objectives of the project include: (1) creating a novel medical device testbed, IoMT network traffic datasets, IoMT device behavior signature datasets, and attack datasets; (2) employing generative adversarial networks (GANs) to identify many of the interesting features for medical networks and devices, to perform multi-class classification, automatic feature extraction, and be robust to noise, support drifts in networks, and support continual learning for new devices for securing IoMT networks and devices; and (3) enabling remote memory forensic capabilities for IoMT devices. The project will also carry out a number of educational activities involving K-12, undergraduate, and graduate students, make strong outreach efforts for recruiting and mentoring students from underrepresented groups, as well as enrich undergraduate and graduate curricula through exposing students to cutting-edge research in networking, security, and machin

HomePal: Developing a Smart Speaker-Based System for In-Home Loneliness Assessment for Older Adults

National Institue of Health/National Institute of Aging

2023-08-31

Project Abstract Experiences of loneliness are prevalent among older adults. Loneliness is a painful and pernicious state occurring when there is a perceived discrepancy between one’s optimal levels of social interactions and actual social relationships. Given that loneliness has been associated with negative health outcomes, detection measures are imperative to identify both lonely and at-risk individuals early enough to intervene before adverse health outcomes occur. However, the assessment of loneliness is challenging due to the stigma of being labeled as lonely and the lack of integration of loneliness assessments into primary care or community-based services. To tackle the problem, we will use smart speakers and passive sensing data to automatically assess older adults’ level of loneliness as correlates of self-report scores on the UCLA Loneliness Scale. Our goal in this project is to develop, deploy, and validate a smart speaker-based system for in-home loneliness assessment that integrates Internet of Things (IoT) devices to gather both speech (acoustic and prosodic features) and behavioral data (e.g., smart speaker use patterns, in-home mobility, sleep) from older adults. We will enroll 70 individuals (age 65+) living alone in the community to collect digital biomarker data continuously for 3 months. The specific aims are to: (1) Develop an innovative remote loneliness assessment system that allows passive and unobtrusive capture of speech and behavioral data in the home setting; (2) Using a train- test approach, develop and evaluate the performance of novel multi-class machine learning (ML) algorithms— semi-supervised type of Generative Adversarial Networks (SGANs) — to estimate older adults’ loneliness scores. The UCLA Loneliness Scale will be completed by participants every two weeks for 3 months to collect ground truth data; and (3) Identify potential implementation barriers to the effective use of the system among older adults (n=30). In particular, we will assess potential privacy and security concerns, social influence, cultural values, and the level of personification of virtual agents that could influence the adoption and use of the system. The novel ML models will allow us to identify not only “already” lonely individuals but also “at-risk” individuals. The proposed research seeks to provide preliminary evidence that digital biomarkers from smart speakers and IoT devices can be used for automatic loneliness assessment in the community.

Tracking and Diverting All-You-Can-Eat Cafeteria Food Waste and Loss from an Urban Public University in Central Virginia

Sustainable Agriculture Research and Education (SARE)

2024-11-30

This innovative project will leverage cutting-edge technologies and interdisciplinary collaboration to overhaul VCU’s food system. At the heart of the effort is the implementation of a state-of-the-art food waste tracking system within the university’s central dining hall. By combining advanced technologies such as RFID-tagged plates, machine learning, and high-precision measurement tools, the system will provide actionable insights into food preparation and consumption habits, allowing for smarter decision-making and significant reductions in waste.