[Paper Review] Integrated lithium niobate microwave photonic processing engine

Hanke Feng, Tong Ge+10 more|arXiv (Cornell University)|Jun 26, 2023

Photonic and Optical Devices4 citations

TL;DR

This paper presents a monolithic integrated lithium niobate platform for microwave photonic processing that enables high-speed, low-power analog signal computation at up to 92 Gsps. It achieves 98.1% accuracy in temporal integration and differentiation, demonstrating applications in ODE solving, ultra-wideband signal generation, and photonic-assisted medical image segmentation with orders-of-magnitude speed and power advantages over electronics.

ABSTRACT

Integrated microwave photonics is an intriguing field that leverages integrated photonic technologies for the generation, transmission, and manipulation of microwave signals in chip-scale optical systems. In particular, ultrafast processing and computation of analog electronic signals in the optical domain with high fidelity and low latency could enable a variety of applications such as MWP filters, microwave signal processing, and image recognition. An ideal photonic platform for achieving these integrated MWP processing tasks shall simultaneously offer an efficient, linear and high-speed electro-optic modulation block to faithfully perform microwave-optic conversion at low power, and a low-loss functional photonic network that can be configured for a variety of signal processing tasks, as well as large-scale, low-cost manufacturability to monolithically integrate the two building blocks on the same chip. In this work, we demonstrate such an integrated MWP processing engine based on a thin-film lithium niobate platform capable of performing multi-purpose processing and computation tasks of analog signals up to 92 giga samples per second at CMOS-compatible voltages. We demonstrate high-speed analog computation, i.e., first- and second-order temporal integration and differentiation with computing accuracies up to 98.1 %, and deploy these functions to showcase three proof-of-concept applications, namely, ordinary differential equation solving, ultra-wideband signal generation and high-speed edge detection of images. We further leverage the image edge detector to enable a photonic-assisted image segmentation model that could effectively outline the boundaries of melanoma lesion in medical diagnostic images, achieving orders of magnitude faster processing speed and lower power consumption than conventional electronic processors.

Motivation & Objective

• To develop a chip-scale, monolithic platform for integrated microwave photonic processing with high fidelity and low latency.
• To enable high-speed analog computation of microwave and optical signals using electro-optic modulation and low-loss photonic circuits.
• To achieve CMOS-compatible operation at low voltages while maintaining high linearity and efficiency.
• To demonstrate multi-functional signal processing for real-world applications such as image edge detection and differential equation solving.
• To enable ultra-fast, low-power photonic-assisted image segmentation for medical diagnostics.

Proposed method

• The platform uses thin-film lithium niobate to integrate electro-optic modulators and low-loss photonic waveguides on a single chip.
• Electro-optic phase modulation is used to perform linear, high-bandwidth microwave-to-optical signal conversion at CMOS-compatible voltages.
• A reconfigurable photonic network enables dynamic implementation of signal processing functions like integration and differentiation.
• The system leverages temporal domain processing to compute first- and second-order derivatives and integrals of analog signals.
• Image edge detection is implemented using photonic temporal differentiation, enabling real-time boundary outlining in medical images.
• The photonic processor is interfaced with a machine learning inference model for image segmentation, replacing electronic computation.

Experimental results

Research questions

• RQ1 Can a monolithic lithium niobate platform achieve high-fidelity, high-speed analog signal processing at 92 Gsps with CMOS-compatible voltages?
• RQ2 To what extent can photonic temporal integration and differentiation achieve accuracies exceeding 98% in analog signal computation?
• RQ3 Can photonic-assisted image segmentation outperform electronic processors in speed and energy efficiency for medical imaging?
• RQ4 How effectively can the platform solve ordinary differential equations using all-optical computation?
• RQ5 Can the system be reconfigured for multiple signal processing tasks without hardware redesign?

Key findings

• The platform achieves 98.1% accuracy in first- and second-order temporal integration and differentiation of analog signals.
• Signal processing is demonstrated at up to 92 gigasamples per second, enabling real-time processing of high-bandwidth microwave and optical signals.
• The system enables photonic-assisted image segmentation that outlines melanoma lesion boundaries with significantly lower latency and power than electronic alternatives.
• Ultra-wideband signal generation is successfully demonstrated using the photonic differentiator as a key building block.
• The platform supports reconfigurable signal processing functions, including ODE solving and edge detection, on a single chip with low loss and high linearity.
• The integration of electro-optic modulators and photonic circuits on a single thin-film lithium niobate chip enables scalable, low-cost manufacturability.