There’s some exciting stuff happening in the world of technology! Researchers are developing advanced ways to process radiofrequency signals that could really speed up wireless communication. This means devices linked to the internet can exchange information faster and use less energy. At the moment, much of this signal processing is managed by software-defined radios, or SDRs, which rely on software for filtering and analyzing signals.
Even though SDRs are commonly used, they have a hitch: they split computing and memory functions, which leads to lots of back-and-forth data transfers and energy waste. Plus, they use analog-to-digital converters to turn radio signals into digital data, which is often slow and power-hungry. This has prompted engineers to think outside the box and create better systems that can use analog signals directly, making everything more efficient.
A cutting-edge project from the University of Massachusetts Amherst, Texas A&M University, and TetraMem Inc. is looking promising! These researchers have rolled out a new system that incorporates memristors — revolutionary non-volatile memory devices — for processing analog radiofrequency systems. Their findings, published in Nature Electronics, show that this new approach can boost speed and energy efficiency compared to traditional SDRs.
According to the team, this idea was inspired by how our brains work when processing sensory signals. Instead of converting analog signals to digital bits and then back for processing, similar to what current devices do, our brains handle analog signals in real-time and remember only what’s really important. This approach uses way less energy.
In their study, they developed a memristive system-on-a-chip (SoC). Think of this as a super-efficient circuit that combines all the necessary components for computing, with memristors ready to go. It mimics how our brains function, processing and drawing out important data quickly.
The integrated circuit has a unique crossbar array of memristors, designed to manage signals as they come in, which is more efficient and aligned with specific AI algorithms.
According to the authors: “The memristive SoC results from over a decade of hard work in AI hardware. We’ve gone from high-performance memristors to creating integrated chips for commercial use, aiming to bring memristive hardware into the AI and communications scenes for the future.” They claim this chip shows much faster processing speeds and energy efficiency compared to current sensor-processing solutions.
This innovative circuit processes the analog signals directly and sorts crucial information, much like how we do in our brains. With tests revealing low latency and minimal power use, this medical device concept dares to rewrite rules.
As they put it, “This concept allows for ultra-fast and energy-conscious signal processing directly on edge devices. Our architecture splits the processing workload across ten computing cores, supported by on-chip peripheral components.”
To showcase what they can do, the team ran tests on high-accuracy RF transmitter ID and anomaly detection, ending up with fantastic results: less lag and better energy efficiency than leading digital platforms.
This new SoC promises to bring AI right into the wireless communication processes using the amazing properties of memristors. Looking forward, we might see the principles behind this dev lead to the creation of more advanced memristor-based chips, paving the way for super speedy and energy-smart wireless systems.
As they said, “Our recent work is just a starting point, and we’re already working on a more powerful memristive system and advanced RF circuits to support future needs and expand functionality. We’re aiming for our tech to mesh seamlessly with existing Wi-Fi standards and the upcoming sixth-generation networks, where AI helps with smart RF signal processing, adapting to complex wireless setups.”
This article was brought to you by Ingrid Fadelli, edited by Gaby Clark, and fact-checked by Andrew Zinin—it’s the result of dedicated efforts! We appreciate readers like you keeping independent science journalism alive. If you find value in our work, consider a donation.
For more details: Yi Huang et al, Radiofrequency signal processing with a memristive system-on-a-chip, Nature Electronics (2025). DOI: 10.1038/s41928-025-01409-y.
© 2025 Science X Network
This story was originally published on Tech Xplore.
