UC Researcher Tackles Global Problem of Data Storage

CEAS Associate Professor Rashmi Jha leads as PI on NSF-funded research that will enable massive data storage with reduced power consumption and increased complexity of computing tasks performed by devices ranging from smartphones to the Internet of Things.

data storage
Rashmi Jha, PhD. Photo/UC Photography

With the rise of the Cloud, artificial intelligence and the Internet of Things (the foreseeable future in which not just computers but all of our devices and appliances are connected to each other via the Internet), the need to store data has been forecasted to increase to ~163 zettabytes (or 163 trillion gigabytes) by 2025. And in order to store such a vast amount of data, non-volatile memory (NVM) devices will be needed.

Consequently, by the year 2025, non-volatile memory will represent an $82.3 billion market.

Conventional memory technologies face serious challenges in addressing these needs due to increased energy consumption, scalability limitations and limited bandwidth. Therefore, there is an urgent demand to develop new memory devices, circuits and architectures to support these needs.

Rashmi Jha, PhD and University of Cincinnati College of Engineering and Applied Science (CEAS) associate professor of Electrical Engineering and Computer Science, aims to develop new types of memory devices that will be massively scalable, energy-efficient and fast, which will open tremendous opportunities to integrate them in existing memory architectures to enable future applications.

Funded by the National Science Foundation, Jha’s UC group will collaborate with Swaroop Ghosh’s team at Penn State University to achieve the objectives of this research. Combined NSF-funding for UC and Penn State totals $450,000 through 2020.

“Our latest and greatest technology advancements such as AI, the Internet of Things, machine learning and deep neural networks require massive data storage and in-memory computing, making new computer architectures like tensor processing units of the utmost necessity,” Jha said. “Availability of on-chip memory and the ability to do in-memory computing will be critical to address the massive computer resources demands for training and inferencing using these architectures.”

data storage
Jha’s PhD students in lab: Andrew Rush, Abhijeet Barua, Joshua Mayersky, Thomas Schultz, Alex Jones.

When integrated with smartphones, Jha’s technology will allow the devices not only to store more data (i.e. pictures and videos) through low power consumption, but also to perform more complex computing tasks. Possible tasks include: facial-cue recognition, behavior recognition and smart health monitoring with actionable insights. The integration of these systems in wearable computing platforms such as wearable sensors, smart watches or point-of-care devices will help programmers to implement more complex algorithms that can be used for applications such as monitoring rehabilitation after injuries or detecting skin cancers.

Beyond handheld and wearable devices, the integration of this technology in data centers will greatly reduce their power consumption and enable data storage on a much larger scale.

“The types of memory devices we are developing will have tremendous opportunities not only in high-density data storage but also in developing computing platforms for advanced artificial intelligence,” Jha said. “In the long run, it would be highly beneficial for society if artificial intelligence in machines become as scalable and versatile as human brains. Then, for example, wearable robots will be able to help people suffering from Alzheimer’s disease to navigate and make decisions if they are lost, or help micro-drones to make decisions like birds or autonomous cars to have human-like intelligence.”

UC is leading this effort by providing excellent research facilities including ERC Microfabrication Center and Materials Characterization Center and the state-of-the-art Advanced Logic and Memory Devices Electrical Characterization facility at the Microelectronics and Integrated-systems with Neurocentric Devices (MIND) lab, under the direction of Jha.

data storage
Jha’s NSF REU student Tony Bailey making measurements on synaptic memory devices in MIND lab.

Jha also plans to infuse her project with education by incorporating research outcomes into her academic courses. Undergraduate students will be involved in the research via Research Experiences for Undergraduates programs at Penn State University and University of Cincinnati.

“I hope to also involve under-represented groups in science and engineering through several outreach programs at both the universities, such as Summer Research Opportunities Program at PSU, and Emerging Ethnic Engineers (E3) program at UC. The E3 program provides opportunities to high-school students from the inner-city schools with economically challenged backgrounds to participate in internships and lab-experience activities,” she said.

In addition to the NSF grant, Jha’s research group is collaborating with the Air Force Research Lab (both the Rome Research Lab in Rome, NY, and the Wright-Patterson Air Force Base in Dayton, Ohio), and DAGSI (Dayton Area Graduate Studies Institute).

Jha joined the University of Cincinnati in 2015. She holds a PhD (2006) and a master of science (2003) in electrical engineering from North Carolina State University. Jha attended the Indian Institute of Technology (IIT), where she earned her bachelor of science in electrical engineering in 2000. She received an NSF-CAREER award early on in her career to perform research in the areas of neuromorphic computing. Most recently, Jha was awarded Master Educator Award by CEAS at UC.

“I would like to thank all of my students and collaborators for their contributions on various projects. I work with a team of very talented undergraduate and graduate students and I’m excited to see what we accomplish together!” she said.

Source : University of Cincinnati