Researchers Scale HF Signal Capability to Achieve Terabits-Per-Second Data Links

April 08, 2020 by Luke James

Researchers have developed a prototype system that in theory could increase data transfer speeds to highs of 10 terabits per second.

In contrast to current data rates, that figure is many thousand times faster than your average broadband connection. 

Growing consumer data usage has led to an increase in the demand for higher data rates in both wired and wireless systems. To explore a potential solution for meeting these demands, researchers have used the same technology that enables high-frequency signals to travel through regular phone lines.


Manipulating HF Signals to Improve Data Transfer

To test the transmission of data at higher frequencies, researchers described how they used experimental measurements and mathematical modeling to characterize the input and output signals in a waveguide. The research, which was published in Applied Physics Letters, explains how the researchers sent extremely high-frequency 200 GHz signals through a pair of copper wires. 

The researchers used a device with two wires running parallel inside a sheath with a large diameter. This sheath enables increased mixing of the waveguide modes and these mixtures, in turn, enable the transmission of parallel noninterfering data channels. Higher frequencies allow more bandwidth, which in turn, boosts the overall transfer rate. 


An input-output channel matrix for a waveguide structure.

A complete input-out channel matrix for the waveguide structure used in the research study. Image credited to Brown 
University, School of Engineering and ASSIA, Inc


A Significant Increase in Data Transfer Speeds

In doing this, the researchers found that data could move at rates of up to 10 terabits per second, a speed significantly faster than what can be achieved with currently available channels. However, this can only be achieved if the channel’s architecture is such that data is not interfered with and garbled. 

"To confirm and characterize this behavior, we measured the spatial distribution of energy at the output of the waveguide by mapping the waveguide's output port, showing where the energy is located," author Daniel Mittleman said.

The researchers created a 13 x 13-millimeter grid for the output of each possible input condition. This created a 169 x 169 channel matrix that allowed the team to completely characterize the waveguide channel. The results demonstrated a superposition of waveguide modes and allowed the team to estimate data ranges.