All optical signal processing for ultrafast imaging systems

Shoeiby, M 2017, All optical signal processing for ultrafast imaging systems, Doctor of Philosophy (PhD), Electrical and Computer Engineering, RMIT University.


Document type: Thesis
Collection: Theses

Attached Files
Name Description MIMEType Size
Shoeiby.pdf Thesis application/pdf 4.64MB
Title All optical signal processing for ultrafast imaging systems
Author(s) Shoeiby, M
Year 2017
Abstract This thesis is constructed around the theme of all-optical ultrafast signal processing techniques to tackle the bandwidth limitation of electronics in the context of serialised time encoded imaging. Two sets of approaches are taken to develop these processing methods. The first approach is the all-optical emulation of pattern identification functions such as multiplication (AND gate) and subtraction (XNOR gates) for ultrafast imaging systems. The second approach focuses on all-optical pattern extraction methods such as edge detection using the Hilbert transform (HT). To accomplish the objectives, a combination of linear and nonlinear optical platforms such as the dispersive Fourier Transform (FT) in single-mode fibres and the four-wave mixing (FWM) process in highly nonlinear fibres (HNLFs) are exploited.

First, a novel technique for correlation of 1-D information encoded onto the intensity spectrum of ultra-short pulses is presented. The information is mapped to the time-domain using dispersion and then mixed with a spectrally engineered broadband pump using FWM to create a narrow bandwidth idler, which is then filtered and electronically integrated.

The concept is then expanded to perform processing of 2-D opaque intensity and transparent phase images by simulating a physical ultrafast temporal imaging system, the STEAM camera. Similar to the proposed correlator, the intensity system exploits FWM to perform multiplication. However, the phase system utilises the ability of the FWM process to subtract phase information and exhibits a higher extinction ratio compared to the intensity system. Unlike previous methods, the solutions in this thesis provide all-optical functionality at every stage of processing.

The performance of both systems in the presence of noise is investigated, where the phase system has shown to be highly robust. The demonstrated frame rate of the experimental implementation, 20 MHZ, and the simulated systems, 100 MHz, is only limited by the accumulated dispersion required to perform the frequency-to-time mapping.

Through studying all optical processing using FWM mixing it became apparent that the spectrally encoded images would be most effective if carried by a comb of coherent laser lines rather than a continuous spectrum ultrashort pulse. Hence, the thesis next studies signal processing using integrated comb sources. First, a temporal microwave photonics Hilbert transformer based on transversal filtering technique is demonstrated. The optical source used is a CMOS-compatible integrated nonlinear microring resonator frequency comb. The comb source allowed the design of filters with up to 20 taps with high stability. To experimentally demonstrate the temporal transformer, it is applied to an RF Gaussian pulse with intensity FWHM of 0.12 ns with an excellent agreement between the measured and the simulated results.

Having shown that photonic signal processing can be achieved using comb sources and in particular that a Hilbert transform can be implemented, the thesis then explores the use of comb sources and Hilbert transform techniques for edge detection. First, a 2 tap Hilbert transform, which is ideal for edge detection is conceived and numerically analysed. The 2-tap Hilbert transform was used to perform edge detection of temporal signals after detection by a fast photodiode. This is the first reported case of the all-optical temporal Hilbert transform for edge detection applications.

Finally, inspired by the transversal 2-tap Hilbert transform, a proof of concept simulation is demonstrated that exploits the FWM process to implement an all-optical scheme for direct edge detection of spectral information without the need for frequency-to-time mapping.
Degree Doctor of Philosophy (PhD)
Institution RMIT University
School, Department or Centre Electrical and Computer Engineering
Subjects Signal Processing
Photonics and Electro-Optical Engineering (excl. Communications)
Keyword(s) Correlation
Nonlinear optical image processing
Nonlinear optical signal processing
Pattern matching
All optical signal processing
Versions
Version Filter Type
Access Statistics: 126 Abstract Views, 84 File Downloads  -  Detailed Statistics
Created: Thu, 13 Jul 2017, 13:26:20 EST by Adam Rivett
© 2014 RMIT Research Repository • Powered by Fez SoftwareContact us