Simple Rules for Optimizing Asymmetries in SOA-Based Mach-Zehnder Wavelength Converters

In this paper, we present an analysis of semiconductor optical amplifier (SOA)-based differential Mach-Zehnder wavelength converters with a specific focus on optimizing performance through intentional asymmetries in optical power splitting, SOA bias, and interferometer phase bias. By introducing a simple conceptual framework for understanding the amplifier pulse dynamics, two simple yet effective design rules are derived. These design rules are validated using pseudorandom code in a comprehensive computer model, demonstrating the performance penalties that result when attempting optimization using only unequal SOA biasing or phase biasing. This work illustrates that dramatic improvements in extinction and eye margin can be achieved with optimized splitter asymmetries, and have significant implications for improved network performance and converter cascadability.