Microwave Photonic Interference Cancellation: RF Analysis, III-V and Silicon Integration, Development of Balanced and Hybrid Architectures

Abstract

The demand for wireless data has maintained a consistent exponential growth between 40%- 60% over the past ten years. This demand is generated, as of 2023, from the 5.7 billion unique mobile users, an equal amount of machine-to-machine connections, and the ubiquitous use of highly data-intensive video-based applications. Although demand rapidly increases, our spectral resources remain a fixed commodity. This imbalance has strained modern wireless networks and resulted in a dramatic inflation of spectral licensing cost, as managed and auctioned by the Federal Communications Commission (FCC). In 2021, exclusive rights to 280 MHz of potential 5G licenses, centered at 3.7 GHz, were auctioned to wireless providers for 80.9 billion USD. Increasing economic barriers to wireless communication can limit connectivity in developing and highly populated regions.

Creating more efficient wireless networks would alleviate congestion, reduce cost, and allow for the continued development of high-bandwidth applications. To mitigate the growing disparity between supply and demand of spectral resources requires developing novel radios which operate at a higher and broader bandwidth, that implement more advanced communication techniques which fully optimize spectral and spatial resources. Implementing Cross-Division Duplex (XDD) techniques eliminates the need for unused spectrum, such as guard bands. Furthermore, implementing Full-Division Duplex (FDD) techniques eliminates the need for dedicated transmit and receive channels and hence effectively doubling the spectral efficiency. This increased performance comes at the cost of significant self-interference. The analog signal processing required to cancel self-interference, enabling high-bandwidth next-generation wireless networks, requires an optical solution; Hence, the development of the Microwave Photonic Canceller (MPC).

First developed within the Lightwave Laboratory at Princeton University in 2009, the MPCs demonstrated wideband removal of self-interference, tunable across GHz range. The pitfall of microwave photonic systems of the last few decades has been the high radio-frequency (RF) loss and noise. The work within this thesis focuses on the continued development of MPCs via the exploration of novel architectures to improve RF performance metrics, resulting in commercially viable solutions. Specifically, we demonstrated a balanced architecture for relative-intensity noise suppression and a hybrid architecture for RF chain preservation. These architectures resulted in over 20 dB reduction in the RF noise figure. While the architectures implemented were specific to microwave photonic cancellation, the techniques and performance improvements can generally be applied to microwave photonic analog signal processing.

In addition to architectural improvements, we explored leveraging the recent maturation in integrated photonic fabrication to achieve higher levels of productization with improved cancellation performance, robustness, and a significant reduction in size, weight, and power. A Balanced MPC was integrated onto an Indium Phosphide photonic integrated circuit, demonstrating increased circuitry complexity. Both passive and active silicon photonic MPCs were developed to explore the increased integrated functionality of silicon resulting in a 10-fold increase in instantaneous cancellation bandwidth compared to previous architectures.

In the last chapter of this work, the RF insights and experience gained through the development of the silicon photonic MPC are applied to the field of silicon neuromorphic photonics. The two systems draw parallels with each other. High-speed RF-optimized analysis, designs, and simulations are presented.