NEuromorphic PhoTonics with fast and efficient vertical cavity sUrface emittiNg lasErs

NEPTUNE aims to develop novel neuromorphic photonic technologies able to both sense and process information in real-time, drawing direct inspiration from the brain’s powerful computational capabilities. Current photonic technologies find applications in sensing, communications, and information processing, including security, environmental monitoring, high-speed fibre-optic and wireless data links. However, sensing and computation functionalities are frequently decoupled, creating bottlenecks in energy and latency. NEPTUNE aims to tackle this critical challenge, addressing the urgent need for compact, low-cost, fast, and efficient photonic platforms that integrate both functionalities in a single hardware framework. NEPTUNE’s platform exploits Vertical-Cavity Surface-Emitting Lasers (VCSELs), widely deployed in our society (in barcode scanners, mobile phones, light sources in optical networks, datacentres, etc.) for their low cost and energy efficiency. These VCSEL systems, operating at multiple infrared wavelengths, will be used to generate complex speckle patterns through optical fibres, acting as light diffusive media. The characteristics of these light patterns are highly sensitive to environmental perturbations (such as temperature, strain, and audio signals) enabling their sensing functionality. Event-based photo-detecting systems, which convert light signals into neural-like spikes, will capture the formed speckle patterns, creating a high-resolution photonic sensing system based on an ultrafast, discrete, neuromorphic data representation. Simultaneously, by generating fast optical neural-like spiking regimes with the VCSELs, we can exploit spike-based encoding mechanisms to represent input data in speckle patterns, embedding further computational capability in the platform. This allows the use of neuromorphic paradigms and algorithms, such as extreme learning machines and spiking neural networks, for low-latency photonic processing tasks.