Speeding up networks requires new kinds of hardware. As technology moves toward 6G wireless standards, some of the hardware required is still under development.
6G's speeds require advanced techniques like configurable filters and beamforming, which are challenging to accomplish with traditional electronic circuits. The interface between the fiber optic cable is the "hardwired" connection of the network and the wireless signal that 6G represents is complicated. However, a team of researchers from the Universitat Politècnica de València in Spain claims to have developed a solution to this problem: a programmable photonic processor.
Optical to RF networking
Fiberoptic cables, which underpin most modern networks, are a type of photonic technology. They use visible light to transmit data and have several advantages over other technologies, such as copper wires. These include ultra-high bandwidth and lower power consumption, both of which are key to network scalability.
However, once a fiberoptic cable reaches its destination, it is typically converted into a traditional copper cable (e.g., an Ethernet port) or a wireless signal. The reverse is also true — when a mobile endpoint like a laptop or a smartphone sends data back to the access point (e.g., a wireless router), it must translate that data back into photonics to send back along the fiberoptic line to link up with the rest of the internet.
That interface, between the optical and radio frequency realms, is a limiting factor in rolling out faster standards such as 6G. Traditional electronic ways of doing so, such as a flat 2.4 GHz Wi-Fi signal, can’t keep up. Newer technologies, like microwave photonics, show promise by being able to perform 12 key functions necessary to interface the two realms.
However, so far, devices have only ever been able to perform one of those 12 functions and couldn't be configured to do another. They were housed in an application-specific photonics integrated circuit (ASPIC), which is equivalent to the ASICs that were commonly used in early computers to perform specific tasks.
More modern systems use field-programmable gate arrays (FPGAs) to create custom circuitry that can be modified on the fly based on how the system is programmed. Similarly, microprocessors, while not configurable from a hardware perspective, can perform customizable tasks using their software layer. While those technologies are well established for electronic systems, so far, no equivalent has been created for photonic systems — until now.
Programmable photonics
Research led by José Capmany of the iTEAM Research Institute of the name="_Hlk188970888">Universitat Politècnica de València claims to have done just that. In a paper published in Nature Communications, Dr. Capmany and his team describe what they call a "general-purpose programmable photonic processor."
The system they describe in the paper is made up of three separate layers — the photonics layer, an electronics layer and a software layer. The breakthrough was the interfaces between the three, which allow for configurability, something the team claims has never been captured by a single photonics chip.
On the photonics layer, the team built a series of 72 programable unit cells (PUCS). Each PUC uses a tool called a Mach-Zehnder Interferometer, which allows the chip to measure the phase shift between two beams. Notably, they also include a thermo-optic phase actuator, which uses temperature to distort the phase of the light passing through it.
These thermo-optic systems can have their temperature manipulated via a control system — in this case, provided by the electronics layer in a module called the logic unit. As the temperature changes, and therefore the phases change, the readings from the interferometer change, allowing it to perform logical operations that form the basis of the 12 functions required of a microwave photonics system.
The electronics system itself is comprised of three different subsystems:
- Driving subsystem to control the phase actuators in each PUC.
- Readout subsystem that uses photodetectors to provide a feedback control loop into the third subsystem.
- Logic subsystem that acts like a traditional electronics processor.
A software framework sits on top of the electronics system, acting similarly to how a firmware layer would in a traditional microprocessor. This layer is responsible for programming the LU to perform the proper functions at the right time, depending on what is needed in the current context. Since it runs on a typical processor, this interface would be familiar to most firmware engineers; however, the manipulation tools require a fundamental understanding of photonics.
Ultimately, this three-layer stack allows the photonics processor to perform any of the 12 main functions typically required of a microwave photonics system. These range from beamforming to providing a tunable optical filter, all normally performed by bulky one-off ASPICs.
Use cases
Dr. Capmany and his team said the chip could change several industries, not just RF networking but also AI, lidar and spectroscopy.
"It's the first chip in the world with these characteristics,” Dr. Capmany said. “It can implement the twelve basic functionalities required by these systems and can be programmed on demand, thus increasing the efficiency of the circuit.”
The Universitat Politècnica de València team is working with a team from technology vendor iPRONICS, where Dr. Capmany is also involved. The start-up has included the underlying processor in a technology called Smartlight, which is currently commercially available and is being tested at several companies, including Vodafone.
Next steps
The team is working to make the system more scalable, and it will still face some commercial challenges. One deals with specialization. While having a single chip that can do all the different functionalities might be convenient, it is undoubtedly more efficient to have a chip that specializes in just one. iPRONICS hopes to offset this by utilizing mass production of Smartlight chips to lower the unit price to well below the ASPIC prices, even at the cost of slightly less efficient design.
Other struggles with commercialization will include the ecosystem built around the chip, the most successful microprocessors and FPGAs are well supported by documentation and lots of free learning material on the Internet. Smartlight will have to develop that same infrastructure as a more traditional approach to RF communications like the ESP32 if it wants to succeed.
However, if widely adopted, Smartlight might represent a breakthrough in how the next generation of networking is done.
About the author
Andy Tomaswick is an engineer and freelance writer who is passionate about education, space exploration and making the world better through technology. When not engineering or writing, he spends time with his family or running in circles to stay in shape.
