Cornell University researchers have developed a low-power microchip called ‘microwave brain’ – a processor that can compute on both ultrafast data signals and wireless communication signals using the physics of microwaves.
Integrated fully on a silicon microchip, the processor is regarded as the first-of-its-kind microwave neural network. It computes at a real-time frequency domain for tasks like radio signal decoding, radar target tracking, and digital data processing within 200 milliwatts of power.
“Because it’s able to distort in a programmable way across a wide band of frequencies instantaneously, it can be repurposed for several computing tasks,” said lead author Bal Govind, a doctoral student.
Maxwell Anderson, also a doctoral student, discussed the microchip’s ability to process many signals. “It bypasses a large number of signal processing steps that digital computers normally have to do,” he said.
The inner workings
The chip works like a brain-inspired network; it uses special waveguides to connect its parts to spot patterns and learn from data.
Unlike the regular digital chips that process instructions step-by-step on a clock, this one uses fast, analogue signals in the microwave range, letting it handle data at tens of gigahertz – much faster than most digital chips.
“Bal threw away a lot of conventional circuit design to achieve this,” said Alyssa Apsel, professor of engineering, and a co-senior author.
“Instead of trying to mimic the structure of digital neural networks exactly, he created something that looks more like a controlled mush of frequency behaviors that can ultimately give you high-performance computation,” she continued.
An all-rounder chip
The chip can handle everything from basic logic operations to advanced jobs like spotting bit sequences or counting binary values in fast-moving data.
In tests, it reached 88 percent or higher accuracy on several tasks that classified wireless signal types – matching the performance of digital neural networks while using far less power and space.
“In traditional digital systems, as tasks get more complex, you need more circuitry, more power, and more error correction to maintain accuracy,” Govind said.
“But with our probabilistic approach, we’re able to maintain high accuracy on both simple and complex computations, without that added overhead,” he explained further.
Because the chip is extremely sensitive to incoming signals, researchers have deemed it perfect for hardware tasks. It can detect unusual activity in wireless communications across different microwave frequency bands.
“We also think that if we reduce the power consumption more, we can deploy it to applications like edge computing,” Apsel said.
“You could deploy it on a smartwatch or a cellphone and build native models on your smart device instead of having to depend on a cloud server for everything.”
The chip may be in its experimental stage as of now, but the researchers are affirmative regarding its scalability. They are already improving their accuracy to integrate into existing microwave and digital processing platforms.
The study was published in the journal Nature Electronics.