Multi-input injection locking in a single-mode Fabry-Pérot laser for all-optical gates
All-optical signal processing has attracted significant interest in recent years due to its higher data rates, reduced electromagnetic interference, crosstalk, heat dissipation, and risk-free short circuiting compared with its electronic counterparts.1 All-optical logic gates are the basic fundamental units of signal processing and computing for modern information and communication systems. Among other technologies, semiconductor optical amplifiers (SOAs) are widely used and have the advantage of compactness and integratability. However, they are expensive and require high driving current of ∼200mA. Fabry-Pérot laser diodes (FP-LDs)—whose threshold is just ∼9mA and cost a few tens of dollars—represent an energy-efficient, cost-effective all-optical alternative to SOAs. In addition, because this scheme uses single-mode (SM) FP-LDs,2 it requires no probe beam, as do SOAs and multi-mode (MM) FP-LDs: see Figure 1(i). Similarly, in Figure 2(ii), the supporting beams alone cannot suppress the dominant mode (λ1) even with two input beams. However, with one major beam and any supporting beam, it becomes possible.
The basic principle of signal processing using FP-LDs is injection locking. This refers to a synchronizing effect that suppresses the dominant mode present in an SMFP-LD when an external beam is injected into any of its side modes: see Figure 1(ii).2,3 Our experiments indicate there is some range of power within which the input beam can be injection-locked, but the dominant mode is not suppressed. Consequently, the power range can be used for multi-input injection locking. The basic idea is to manage the input beams by varying their wavelength detuning and power and then exploit the power range between injection locking and suppression of the dominant mode.4, 5 Figure 2 shows the spectrum schematics and power-management technique. For example, beam 2 alone cannot suppress beam 1, but beams 2 and 3 together can: Figure 2(i).
We used multi-input injection locking to implement a variety of logic gates, including NAND, XNOR, and XOR (see Figure 3). SMFP-LD1 is based on multi-input injection locking, SMFP-LD2 on the supporting beam principle, and SMFP-LD3 on injection locking. One major beam (C) and any of the supporting beams are required to suppress the dominant mode of SMFP-LD2. Polarization controllers minimize the losses on the Mach-Zehnder modulator and pass only transverse electric (TE) polarized light to SMFP-LDs, as injection locking in this scheme occurs only with such light.
Figure 4 shows the outputs of each SMFP-LD: C (NAND), E (XNOR), and F (XOR). The 1s and 0s are the states of input and output data that verify the different logic gates. The suppression ratios for all the logic gates are greater than 20dB, which is sufficient to differentiate between logic 0 and logic 1.
The key issue with the approach we have described here is the suppression of the dominant mode with proper adjustment of input beam power at various stages. SMFP-LDs do not need any external probe beams, which means that schemes using SMFP-LDs are efficient and cost-effective compared with other all-optical technologies. The concept can be applied to combinational circuits, computation, optical sensor networks, and automation. We are now working on a combinational circuit and reconfigurable control unit using multi-input injection locking in SMFP-LDs, with potential application in optical computing and signal processing.
Bikash Nakarmi is currently a PhD candidate in information and communication engineering in the Convergence Optoelectronics Device Engineering Lab at KAIST. His research interests are FP-LD applications for photonic computing and power-efficient optical networks.
Yong Hyub Won received his PhD in electrical engineering from Cornell University, NY (1990), and is currently a professor in the Department of Electrical Engineering at KAIST. In 2006, he developed a single-mode light source using MMFP-LD and the world's first test bed for optical burst-switching add-drop multiplexing. His current research interests are FP-LD applications, terahertz communication, and next-generation Internet with ultra-low-power saving capability and power-saving-related components.