Today's world is more digital than ever: petabytes of information are exchanged via the internet. This traffic will continue to increase in the future (partly due to the surge in AI applications). Fiber optic networks, carrying 99% of the internet traffic, are a crucial infrastructure. Optical communication connections now operate at 800 Gbps, but within 10 and 20 years, optical connections will be required that operate at 1.6 Tbps and 3.2 Tbps respectively.

Power Consumption

The major bottleneck preventing operation at higher data rates is power consumption. Optical transponders, in particular digital signal processing (DSP) units, consume a certain amount of energy per exchanged bits. As data rates double and quadruple, the total power consumption follows the same trend if the current DSP algorithms are used.

Nonlinear Propagation Over the Fiber

To operate at higher data rates, the received signal quality (signal-to-noise ratio, SNR) should also be higher. To achieve this, researchers try to push power launched into the fiber to higher values. However, due to the nonlinear propagation characteristics of light through the fiber, the improvement obtained in SNR is limited.

Project activities

In this project, low-complexity and low-power-consumption DSP algorithms will be designed, developed, and demonstrated. One line of study will focus on error correction algorithms that require at least a quarter of the current power consumption per bit, resulting in DSP modules for 1.6 and 3.2 Tbps transponders with below 20 W power consumption. The other line of study will focus on nonlinearity-tolerant signal generation algorithms that will allow launch powers to be increased towards 0-5 dBm region, allowing better signal quality at the receiver side.