The Fourier transform interrogator

Photonic sensors have recently attracted much attention in both industry and academia. High accuracy, low weight and the possibility of building a large sensor network are key benefits of photonic sensors. Another benefit is installing optical sensors in harsh environments where electronic sensors' usage is not plausible: aerospace applications where ionizing radiation is present or gas and oil pipes are some examples.

Integrated photonics brings new challenges to the interrogation of multiplexed sensors in WDM. Unlike FBG sensors, whose resonance wavelength can be chosen to an accuracy better than 1.0 nm, the resonance wavelength of integrated micro-ring resonators cannot be chosen during the design stage. The main reason is the imperfections of the manufacturing process. The fact that the resonance wavelength is unpredictable is a problem for interrogators based on interferometry. Such interrogators perform the demultiplexing and demodulation in different stages: first, a spectrometer separates the optical channels; subsequently, outputs of the spectrometer are conveyed to interferometers. From the photo-receiver voltages connected to MZI outputs, the signal from the sensors can be demodulated. As the resonance value of sensors cannot be determined during design, two sensors may have their resonances in the same spectrometer's channel. As a result, the demultiplexing step fails, compromising the interrogator's operation.

In Chapter 4 of this thesis, a new interrogation method is proposed. Much of the effort of interferometric interrogators is to separate the spectrum of the sensors correctly. In the Fourier Transform Interrogator, the spectrum of all sensors is sent to an array of Mach-Zehnder interferometers (MZI) with different OPDs. Using the output voltages from the photo-receivers attached to the MZIs, we derive a system of non-linear equations, whose solution provides the signal from each sensor. The demodulation and demultiplexing steps are performed simultaneously for the Fourier interrogator, which guarantees the interrogator's unique flexibility. On the other hand, the computational cost is high since the system of non-linear equations is solved using Newton's method. For each set of voltages sampled over time, a different system of equations is obtained. Chapter 4 leaves some unanswered questions:

Does the system of non-linear equations have a unique solution?
How many solutions are there?
What is the physical meaning of each of the solutions?
Is it possible to solve non-linear systems of equations for fast sensors in real-time?

All these questions are answered in Chapter 5. As a consequence of the new algebraic formulation, it is possible to solve about 1 000 000 algebraic systems of equations in about 10 ns, i.e., allowing the real-time interrogation of high-speed sensors. The interrogator is a candidate for interrogating arrays of ultrasound ring resonator sensors in the tens of MHz range.