Undersea cables dramatically increase the reach of the internet by connecting new populations to provide them the economic benefits of high-speed connectivity. New applications driving large amounts of data transfer continue to emerge – Metaverse, new VR/AR experiences, and ML/AI on demand. This has led to more data being managed at the edge of the network, but warehouse-scale computing still requires massive amounts of data to be exchanged between data centers or between data center and the network edge. Gartner predicts global cloud spending to increase to $917B by 2025. While communications service providers are evolving to offer cloud and content services with 5G architectures, hyperscale cloud and content providers continue to connect their data centers across the oceans with subsea cables while adding capacity and new routes for reliability.
Building and deploying subsea cables can take a village. Planners have to forecast capacity requirements today and meet the demands of the future. There must be physical infrastructure, including power to land the cable and house the SLTE and PFE equipment, terrestrial fiber connectivity into POP or Data Center, and there must be a redundancy plan with alternate deployed cable routes in case of any failure or fiber cut. And while subsea cables are a part of the growing digital economy, they have now become a critical asset to our internet infrastructure. New highly available workloads and services with consistent performance are delivered across subsea cables and this is driving additional routes to serve increasing capacity needs.
Space Division Multiplexing Helps Increase Capacity of Subsea Cables
Until 2020, trans-Atlantic and trans-Pacific cable deployments had 2, 4 or 6 fiber pairs per cable. The MAREA cable that went RFS in 2018 was unique with 8 fiber pairs and short repeater spacing to maximize performance. This cable formed the benchmark for multiple trans-Atlantic capacity records. MAREA also formed the background for an impending problem for subsea cables.
Scaling optical fiber capacities has been the focus of multiple generations of Digital Signal Processors (DSP) and high-speed optical components over the last decade. However, as we approached Shannon Capacity limits, capacity gains from coherent transponder innovation alone is getting incrementally smaller and a new approach was needed.
Accessing additional spectrum and packing more fibers into the cable was the next step to continue raising fiber capacities while observing available electrical power constraints. A new generation of undersea cables was developed to use higher count of fiber pairs. At slightly reduced capacity of each fiber pair, the total cable capacity is increased drastically by taking advantage of the linear bandwidth gain from additional fiber pairs and trade off the logarithmic scale repeater power gain per fiber pair (i.e. OSNR). Space Division Multiplexing, or SDM, is the term used to describe these new cables. SDM increases the cable fiber pair count from 4-8 pairs to 12, 16 and upward of more than 20 fiber pairs.
DUNANT was the first SDM cable to go live in February 2021. With 12 fiber pairs, the cable provided 250Tbps across the Atlantic. In July 2022, 20 fiber pair JUNO system was announced that would deliver an astounding 350Tbps of capacity trans-Pacific by 2024.
We expect most new subsea cables to leverage SDM technology and further drive down cost per bit. This in turn enables cable owners to offer an entire fiber or a slice of spectrum as a capacity service. SDM with higher fiber count cables creates more availability of fiber and spectrum for wholesale and retail buyers.
Behind the Technology
Higher fiber count cables imply more total bandwidth to fulfill with coherent transponders. Solutions will soon be available that can provide industry-leading baud rates at 140Gbaud. It would only take approximately 30 wavelengths to light up an entire fiber pair. Current generation transponder technology running at lower baud rates require 50% more transponders to fill the same fiber pair. This problem is exacerbated when the fiber pair count increases to 16 and 20 and further with future SDM cables that can support up to 24 fiber pairs in a single subsea cable. With several hundreds more wavelengths to manage and deploy, this implies more power, space, and operational complexity.
To maximize the capacity of each fiber pair, network operators can leverage Acacia’s advanced 3D shaping technology with the new Coherent Interconnect Module (CIM) 8 module (powered by the Acacia Jannu DSP) in the Cisco NCS 1004 Transponder, which will allow network operators to achieve the best sensitivity and capacity across any cables and any part of the available repeater bandwidth. Traditional transponder technologies can only operate at a few discrete baud rates, with 50G or 100G line side payload increment. Unfortunately, the combination of payload rate and baud rates creates a large step function in required SNR sensitivity. The early generation of coherent product operating at 34 or 56Gbaud, with a few discrete modulation formats were limited to SNR sensitivity gaps as large as 3-4dB. Current generation of product narrows that gap slightly, but it still suffers from the same fundamental limitations.
By combining probabilistic shaping with a powerful FEC algorithm, we can achieve the best SNR sensitivity and get even closer to Shannon’s limit. And by leveraging this continuously variable baud rate, we can accommodate any cable to maximize the capacity, regardless of the cable delivered SNR evolution across any fiber pairs and spectrum region. In reality, not all the cables are perfectly flat, and we have to get the most capacity out of what’s available.