Paulina Beaver from UCL’s Optical Networks Group delivered a comprehensive talk on future-proofing the internet, focusing on the evolution and critical role of optical fiber networks in supporting the vast data demands of modern digital life. She began by tracing the origins of global communications back to the 19th century, highlighting the undersea copper telegraph cables from Cornwall to India and Australia, and early wireless communication experiments. These historical milestones laid the foundation for today’s cloud infrastructure, which underpins everything from mobile phones to autonomous vehicles.
The talk emphasized the exponential growth in data generation, driven by technologies like augmented reality, IoT, and social media, with projections reaching 200 zetabytes by 2025. Optical fibers have become the backbone of this data transmission, carrying about 99% of global internet traffic. Beaver explained how optical networks operate, using lasers and multiplexing techniques to transmit multiple wavelengths of light simultaneously through fibers, enabling incredibly high data rates over vast distances, including transoceanic links.
Despite their efficiency, optical fibers face challenges such as nonlinear signal distortion caused by the intensity-dependent refractive index of the fiber core. Researchers are actively working on mitigating these effects through advanced modulation formats and expanding the usable bandwidth of fibers beyond traditional C-band frequencies to include S and L bands. New fiber technologies, such as hollow-core and multi-core fibers, promise even greater capacity and reduced nonlinearities, potentially revolutionizing future network performance.
Beaver also discussed the complexity of optical network design, including wavelength routing and network resilience. She highlighted the importance of optimizing network capacity and delay while ensuring robustness against physical disruptions like cable cuts. The talk touched on the growing role of optical interconnects within data centers, which are evolving to support CPUs, GPUs, and emerging quantum processors, all requiring ultra-high-speed, energy-efficient communication.
In closing, Beaver reflected on the broader implications of sustaining the internet and AI infrastructure, noting the immense energy and storage demands involved. She advocated for learning from nature’s efficiency, suggesting that future networks should be “educable,” capable of adapting and optimizing through experience and instruction, unlike current AI models. Her group’s ongoing research integrates physics, digital signal processing, graph theory, and AI to design networks that can meet the future’s performance and efficiency challenges.
