Developing error-resilient and highly energy efficient computing platforms for ultra-low power applications will be the focus of a new research at Florida Polytechnic University funded by a grant from the National Science Foundation.
Dr. Ashiq Sakib, assistant professor of electrical and computer engineering, secured the $175,000 grant that could change the landscape of self-powered electronic devices operating on very minimal harvested energy. Low maintenance, safety critical, and extreme environmental applications, such as implantable medical electronics, military surveillance devices, and exploratory deep-sea and outer-space systems could benefit from this research.
“My focus is on energy efficiency and reliability,” Sakib said. “The proposal focuses on creating a venue to implement low power and sustainable embedded computing to support batteryless smart devices.”
Sakib explained that most digital electronic devices are synchronous, which use global clock for synchronizing between components, and are powered by rechargeable batteries.
“With people demanding high speeds, technology is not always able to keep up. Because as you have high operating frequency clocks, the power consumption of your device becomes high,” Sakib said.
The devices he envisions will be based on asynchronous clockless designs, which can resolve the power inefficiencies associated with conventional synchronous designs. This makes them an excellent choice for applications powered by very limited harvested energy, he said.
As the energy generated through harvesting is dependent on the environmental source, movement, or human activities, the supplied energy may fluctuate, and continuous supply of desired energy might not be guaranteed. He hopes to overcome this problem, allowing devices to maintain their functionality despite the energy fluctuations.
“If you cannot produce huge energy through the sources, the device has to be extremely energy efficient,” he said. “If a device has to operate on limited harvested energy, it can benefit from asynchronous designs, and we can create a whole new class of applications that can be serviced using energy harvesting based on asynchronous domain.”
The two-year project will be conducted in two phases. The first will examine existing error-resilient architectures and analyze their error response. The second phase will involve creating modified or new architectures to ensure complete error resilience and evaluating their performance for power consumption, speed, and area utilization.
