Reconfigurable modular optical fibre resonance sensor system using field programmable gate array

The optical fibre sensor (OFS) has been studied for the last few decades and has found extensive use in many scientific and engineering fields including major application areas such as environmental, chemical, and biomedical. In spite of its popularity for use in a range of sensor applications, the OFS system is still facing hurdles for effective deployment in a range of environments. An OFS is a sensor that utilises an optical fibre as the sensing element or as a medium for light to propagate. A typical OFS system is physically large in size and is required to be used only within a fixed laboratory setting which limits its use in portable, on-site applications. To gain widespread use, a user friendly and portable optical sensor system is required and available for wide use. Such a system should be a device that is simple to use and the results easy to obtain and understand for personnel in the field. The work discussed in this thesis considers the development of a modular field programmable gate array (FPGA) based OFS system. The use of the FPGA in an OFS design is explored as the enabling technology. This leads to a portable sensor system design that can perform data processing functions in the field without the need for a personal computer (PC). For sensor data classification, the k-nearest neighbour (kNN) machine learning algorithm has been adopted for use in this system to achieve embedded and real-time system operation. This work has used a surface Plasmon resonance (SPR) sensor to demonstrate the feasibility of the system. The sensor is connected to a light source and photodetector using a plastic optical fibre (POF), allowing the detection of refractive index that has its resonance wavelength fall within the visible wavelength range. A tricolour red, green and blue (RGB) light emitting diode (LED) was used as the light source with single photodiode used as the photodetector. Time domain intensity modulation of individual colour light was used to interrogate three bands of interest in the SPR spectrum using the photodiode. The SPR sensor was developed for use in chemical sensing applications. To test and evaluate the system operation, concentrations of different glucose solutions were classified using the kNN algorithm in this sensor system to demonstrate its feasibility. Two approaches were used in this work to reduce both memory requirement and time complexity through pre-processing and hardware acceleration. This thesis is structured as follows. Chapter 1 will introduce the work and provide a rationale for the approach undertaken. The novel aspects of this contribution will be identified and discussed. Chapter 2 will review the OFS system considering both the SPR sensor and the portable, FPGA based system architecture. Chapter 3 will introduce the FPGA with its internal architecture that identifies the usefulness of the FPGA in embedded sensor system designs. Chapter 4 will introduce the kNN algorithm with different improvement techniques. The improvement technique is to lead towards an embedded classification implementation. The development of the sensor system will be discussed in Chapter 5. Chapter 6 will present the test results and embedded classification. Chapter 7 will conclude the thesis.