Smart Optical Fabric & Devices

Led by Tod Sizer

The IP and Optical Network research lab designs the future of optical communications to meet future demands of applications that have not yet been invented. We push the state-of-the-art in optical and electrical communications in physics, materials science, math, software and optics to create new networks that adapt to changing conditions and go far beyond today’s limitations.

Future smart network fabric

We can approach the theoretical maximum information transfer rates, as defined as the Shannon Limit, discovered in 1948 by Claude Shannon, Bell Labs pioneer and “father of information theory”.  An essential enabler for optimal transmission is a smart network fabric.  Our research in this area includes:

• Superchannels – the combination of multiple optical carriers forming one higher-capacity channel, overcoming bandwidth limitation of today’s electronics and optoelectronics.
• Software defined transponders – the utilization of programmable digital-to-analog converter (DAC) and analog-to-digital converter (ADC) technology and coherent receivers to enable flexible high speed transport systems utilizing high-order Quadrature Amplitude Modulation (QAM) to adapt transmission capacity.
• Rate adaptivity – the ability to trade capacity for transmission reach and spectral efficiency, traditionally implemented by changing modulation formats from BPSK, QPSK to n-QAM.

Operating optical systems close to the Shannon limit

An elegant approach for n-QAM based optical systems is to change the probability of occurrence of constellation points in a fixed constellation, or Probabilistic Constellation Shaping (PCS), which modifies the probability with which constellation points – the alphabet of the transmission – are used. Traditionally, all constellation points are used with the same frequency. PCS cleverly uses constellation points with high amplitude less frequently than those with lesser amplitude to transmit signals that are more resilient to noise and other impairments. Reducing the probability of occurrence of outer constellation points e.g. in a 64-QAM constellation from the typically equal probable case will reduce the transmission rate with an arbitrary fine granularity with almost no changes to the signal processing.

Probabilistic Constellation Shaping has been a focus of research activities in this area, driven mainly by Bell Labs. PCS in optical transport systems achieves higher transmission capacity over a given channel to improve significantly the spectral efficiency of optical communications. With proper selection of the probability distribution of the constellation points we are able to operate closer to the Shannon limit in comparison to traditional unshaped constellations. For instance using a shaped 64-QAM at the same rate as a conventional 16-QAM constellation will yield an 1 dB SNR improvement, which is equivalent to a reach increase of about 25%.

Beyond state-of-the-art optical line rates

We investigate solutions for very high optical channels using a single optical carrier with beyond state-of-the-art systems of 100-200 Gbit/s.  We look at very high-speed electronic multiplexers or converters and also at the synthesis of high speed electrical modulation signals by concatenating multiple DAC outputs.

In the latter approach, we achieved PAM signal generation at 190 GBaud symbol rates with an all electronic 100-GHz Bandwidth Digital-to-Analog Converter.  We also demonstrated the first all-electronically time division multiplexed (ETDM) transponder with a line rate exceeding 1Tb/s (90 GBaud PDM 64-QAM), using a novel 3-bit multiplexing DAC circuit. And using recent electronic converter technology we demonstrated a record single-channel line rate for electronically multiplexed signals, 1.2-Tbit/s polarization-multiplexed coherent transmission experiment over 300 km of fiber.

Spatial optical multiplexing

With classical optical technologies like DWDM, coherent optical transport systems using higher-order of QAM modulation single fibre systems with petabit/s capacity over large transmission distances cannot be realized. We are looking at optical transmission of parallel channels in a few-mode and multi-mode fibre or over multi-core fibre so that we can realize a scalable solution when the entire spectral domains of classical single-mode fibers and of optical amplifiers are exhausted. Our recent research has been focused on investigation of parallel optical transport schemes either by using several modes of few-mode or multimode fibres and also with of multi-core fibres which exhibit more or less coupling between the core.

Different approaches to spatial optical multiplexing: parallel transmission over over multi-core fibres (left) or few-mode optical fibres (right). Optical multiplexing and demultiplexing can be performed by optical devices (optical lanterns based or phase-plates Fan-In/Fan-Out device). MIMO processing can be utilized for crosstalk cancellation.

We are researching optical multiplexing and demultiplexing components, optical multi-core amplifier and parallel fibre transmission to achieve transmission with multi-peta-bit capacity. As an example we have accomplished 138 Tbit/s transmission over 6 spatial modes, 2 polarizations, and 120 wavelength channels carrying 16-QAM signals, over 650 km few-mode fiber, using low insertion and mode-dependent loss mode-selective photonic lanterns.

New datacenter communication architectures

Centralized data centers (DC) emerged in the 2010s as the most important switching and processing centers for applications in enterprise and social communications. We expect future applications, including ‘Industry 4.0’, telemedicine, and autonomous transport to rely on much faster data processing, with far higher data volumes than today’s applications, and ultra-low latency needs. This future Edge Cloud will need a smart network fabric supporting high capacity and low latency.

Our datacenter research addresses flexible adaptive network architectures, efficient interconnection of distributed small DCs by new flexible optical transport systems, flexible adaptive data transport and efficient usage of fiber infrastructure which is commonly known as data-center-interconnect (DCI).

Edge Cloud comprising several small DCs and guaranteeing low latence connections for residential, automotive and industrial applications.