Thermal transport behavior of austenite stainless steel stripe (304) and the carbon nano-tubes (CNTs) films of varying thickness are studied using a micropipette thermal sensor. Micropipette sensors of various tip sizes were fabricated and tested for the sensitivity and reliability. The sensitivity deviated by 0.11 for a batch of pipette coated under same physical vapor deposition (PVD) setting without being affected by a tip size. Annealing, rubber coating and the vertical landing test of the pipette sensor proved to be promising in increasing the reliability and durability of the pipette sensors. A micro stripe (80µm × 6µm × 0.6µm) of stainless steel, fabricated using focused ion beam (FIB) machining, was characterized whose thermal conductivity was determined to be 14.9 W/m-K at room temperature. Similarly, the thermal characterization of CNT films showed the decreasing tendency in the thermal transport behavior with the increase in the film thickness.

Demand of automobiles are significantly growing despite various factors, steadily increasing the average number of vehicles on the road. Increase in the number of vehicles, subsequently increases the risk of collisions, characterized by the driving behavior. Driving behavior is influenced by factors like class of vehicle, road condition and vehicle maneuvering by the driver. Rapidly growing mobile technology and use of smartphones embedded with in-built sensors, provides scope of constant development of assistance systems considering the safety of the driver by integrating with the information obtained from the vehicle on-board sensors. Our research aims at learning a vehicle system comprising of vehicle, human and road by employing driving patterns obtained from the sensor data to develop better systems of safety and alerts altogether. The thesis focusses on utilizing together various data recorded by the in-built embedded sensors in a smartphone to understand the vehicle motion and dynamics, followed by studying various impacts of collision events, types and signatures which can potentially be integrated in a prototype framework to detect variations, alert drivers and emergency responders in an event of collision.

Le besoin croissant de biocapteurs optiques compacts, sélectifs, ultrasensibles, rapides et bas coût dans les domaines médical et environnemental a engendré une émergence de solutions technologiques, notamment les capteurs à bases de microrésonateurs optiques. Ces types de biocapteurs sont capables de fournir une détection sélective de très faibles concentrations de biomolécules si leurs surfaces sont fonctionnalisées. En revanche, les deux méthodes optiques d'interrogation actuelles, balayage spectral et variation de l’intensité, ne peuvent ni fournir la sensibilité de la phase du signal optique propagé dans le capteur, ni les paramètres opto-géométriques (perte par propagation, l'indice effectif, coefficient de couplage, etc) nécessaires pour une modélisation de la réponse du capteur. Pour accéder à ces informations, nous avons proposé d’utiliser l'interféromètre optique à faible cohérence sensible à la phase comme une technique alternative d’interrogation et de caractérisation de microrésonateurs. La première partie des travaux de cette thèse est consacrée à l’étude de conception et de réalisation de microrésonateurs monomodes possédant un facteur de qualité supérieur à 20000 dans l'eau. Cette étude a été validée par la réalisation technologique, à l'aide des procédés de photolithographie classique et de gravure sèche au plasma d'oxygène, de microrésonateurs polymères possédant des facteurs de qualité allant jusqu'à 38 200. La deuxième partie des travaux de thèse est dédiée à l'adaptation du dispositif PS-OLCI, initialement développé au Laboratoire National de Métrologie et d'Essais (LNE) pour interroger les composants des télécommunications optiques, pour la caractérisation de microrésonateurs optiques. Les résultats obtenus en évaluant les performances spatiales et spectrales de différents microrésonateurs ont montré que le dispositif PS-OLCI n'est pas seulement un outil d'interrogation et de caractérisation mais aussi un véritable outil d’aide à la conception de microrésonateurs optiques. Une modélisation, validée par l'ajustement des mesures expérimentales, de la réponse PS-OLCI d'un microrésonateur, met en évidence la relation existant entre l'interférogramme et les intégrales de Fresnel. La dernière partie de nos travaux concerne l'association du dispositif PS-OLCI et d’un composant optofluidique, constitué de microrésonateurs et d’un circuit microfluidique en polymères, pour la détection d’espèces biologiques. A cet effet, la molécule de glucose a été choisie pour démontrer la détection homogène ou volumique en solution aqueuse en obtenant respectivement les limites de détection de l'ordre de 50 µg/ml et de 2 µg/ml en exploitant l’intensité ou la phase des mesures PS-OLCI. Ces performances démontrent la capacité de notre capteur à déceler des biomolécules en faible concentration ainsi que la pertinence de la mesure de la phase, d'où l'intérêt du dispositif PS-OLCI. Pour remédier au problème de sélectivité du capteur en détection homogène, la méthode de détection surfacique est utilisée. La problématique de chimie de surface des polymères, c'est à dire la fonctionnalisation des surfaces des guides polymères en vue d’une détection surfacique, a d'abord été effectuée. Les limites de détection obtenues en détection surfacique sont ensuite évaluées à leur tour pour différents types de molécules particulièrement les protéines telles que la streptavidine ou la biotine. Les performances de détection de streptavidine obtenues sont au moins 10 fois meilleures que celles obtenues à l’aide de la technique de Résonance de Plasmons de Surface considérée à ce jour comme la technique de référence en biodétection sans marqueur. Ces premiers résultats, présentant des marges importantes d’amélioration, contribuent à démontrer que les capteurs à base de microrésonateurs optiques sont des candidats potentiels très prometteurs pour la détection de très faibles concentrations de biomolécules pour l’analyse biochimique. / The increasing need for compact, selective, ultrasensitive, fast and affordable optical biosensors in the medical and environmental sectors gave rise to new technological solutions, especially regarding sensors based on optical microresonators. If their surfaces are functionalized, these biosensors can provide a selective detection of low concentrations of biomolecules. However, two common optical interrogation methods – spectral scanning and intensity variation – cannot provide the same sensitivity as the method using phase detection of the guided modes nor the opto-geometrical parameters (propagation loss, effective refractive index, coupling coefficient), needed for the modeling of the sensor response. To get this information, we proposed to use the Phase Sensitive-Optical Low Coherence Interferometer (PS-OLCI) as a new alternative technical solution for interrogation and characterization of microresonators. The first part of this thesis is dedicated to the conception and fabrication of single mode microresonators with a quality factor higher than 20 000 in water. This work was validated by the manufacture of polymer microresonators with a quality factor up to 38 200, using UV photolithography and Reactive Ion Etching (RIE) processes. The second part of this work covers the adaptation of PS-OLCI setup, initially developed at Laboratoire national de métrologie et d'essais (LNE), to interrogate optical telecommunication devices, for the characterization of optical microresonators. The results, obtained through the analysis of spatial and spectral performances of various microresonators, showed that the PS-OLCI setup is not only an interrogation and characterization tool but also a real support tool for designing optical microresonators. The performed modelling of the PS-OLCI and microresonator association response, validated by the fitting of the experimental data, demonstrated the relation between PS-OLCI measurements and Fresnel integrals. The last part of this work is dedicated to label free biosensing experiments using PS-OLCI setup associated to an optofluidic component, made of polymer optical microresonators and polymer microfluidic circuit, to detect biological species. The glucose molecule was chosen to demonstrate the homogeneous sensing experiments in aqueous solution. The obtained detection limits are around 50 µg/ml when we exploited intensity and around 2 µg/ml when we exploited the PS-OLCI measurements phase. These results demonstrate the high sensitivity of the proposed biosensor as well as the value of the optical phase measurement, hence the interest of PS-OLCI set up. To address the problem of sensor selectivity in homogeneous sensing method, surface sensing experiences were performed. The first step of this method was the functionalization of the sensor surface, by binding adequate recognition molecular sites to the sensor surface in order to immobilize target molecules. Proteins were then chosen to perform this very same type of experiences. The preliminary results in the 0.02 pg/mm2 range clearly show that our sensors is ten times more sensitive than Surface Plasmon Resonance, which is actually considered among the most successful label free sensing methods. These first results, which can be improved, demonstrate that the sensors based on optical microresonators are promising candidates for the detection of low concentrations of biomolecules for biochemical investigation.

Cutting operations are important and commonly used operations in the field of manufacturing. Automated machining is today commonly used in CNC-machines. One common drawback with automated machining is that the tool condition is challenging to predict which leads to a conservative tool replacement times. This leads to a low utilisation of the tool economical lifetime and an unnecessary high number of tool replacements. Methods for indirect continuous monitoring of the tool wear exist but usually require retrofitting of external sensors that can be both costly and also interrupt the machine operation due to the additional wiring. It is therefore of interest to investigate the possibility to use the, often high resolution, sensors already fitted in a CNC-machine to extract valuable data that can indirectly give an estimation of the tool condition. This thesis work has, with attention to the X-, Y- and Z-position sensors, resulted in development of algorithms that show relations between tool wear and data acquired from these sensors. The algorithms operate in the frequency domain to determine changes in the dynamic response over the time of tool degradation.

Previous research on new novel substrates for giant magnoresistance structures has indicated that a net increase in the effect is present. The substrates studied were V-grooved or stepped, however research presented in this thesis used an embossed surface manufactured from alumina oxide which consisted of regular hexagonal arrays with spacing of 110 nm and pore diameter of 60 nm. The physical properties measurements unveiled a net enhancement of the giant magnetoresistance effect thru the whole range of the copper spacer thicknesses deposited with direct current magnetron sputter. The maximum net increase appeared for a spacer thickness of 4.0 nm where the flat silicon substrate yielded a 3 % increase but the embossed surface substrate generated a 12% increase with an overall effect of a 4-fold net enhancement of the effect. Both the aluminum oxide substrates and the thin films structures can be manufactured inexpensively and can be also mass-produced, which are welcoming advantages for the technology sector of magnetic sensing.

Esta tese apresenta os estudos e esforços visando ao desenvolvimento de sensores químicos para aplicações diversas na área forense. Foram desenvolvidos métodos eletroanalíticos para detecção e quantificação de alguns compostos comumente encontrados na adulteração de amostras de drogas de abuso (procaína, fenacetina, aminopirina, paracetamol, levamisol), além da cocaína e estudos fundamentais sobre o comportamento eletroquímico desses compostos. Empregaram-se também métodos eletroquímicos para quantificação de compostos tóxicos e perigosos como explosivos (ácido pícrico) e melamina por exemplo. Os trabalhos utilizando sensores eletroquímicos contemplam modificações eletroquímicas das superfícies eletródicas, utilização de sensores com polímeros molecularmentes impressos (MIP) e eletrodos descartáveis em papel utilizando diferentes técnicas voltamétricas e amperométricas, eletrodo disco rotatório (EDR) e microbalança de cristal de quartzo. Além da fabricação de dispositivos analíticos descartáveis em papel empregando detecção eletroquímica utilizou-se também a detecção colorimétrica para quantificação de alguns dos principais adulterantes de amostras de apreensão de cocaína, como procaína e fenacetina, bem como análises e discriminações de compostos explosivos (peroxi e nitro compostos) nessas plataformas portáteis e de baixo custo. Os métodos foram sempre desenvolvidos visando característicos como: facilidade, praticidade, baixo custo e portabilidade para análises diretamente no local de medida com mínima infraestrutura laboratorial. Por fim, são apresentados alguns estudos realizados durante estágio de pesquisa no exterior (Universidade da Califórnia - San Diego (UCSD)) na área de Wearable Sensors, em que foram desenvolvidos métodos para análises de micronutrientes no suor (zinco) e um metabólito (ácido úrico) na saliva usando sensores aplicados diretamente no corpo humano. / This thesis shows studies and efforts to the development of chemical sensors for different applications in the forensic field. Electroanalytical methods were developed for detection and quantification of some compounds (procaine, phenacetin, aminopyrine, acetaminophen, levamisole) commonly found in the drug of abuse adulteration process and cocaine, as well as, fundamental studies about the electrochemical behavior of these compounds. It was also employed electrochemical methods for quantification of hazardous compounds such as explosives (picric acid) and melamine. Analytical methods with electrochemical sensors included electrochemical modification of electrodic surfaces, molecularly imprinted polymers (MIP), and paper disposable electrochemical devices using different voltammetric and amperometric techniques, rotating disc electrode (RDE) and quartz crystal microbalance. In addition to the fabrication of paper disposable analytical devices with electrochemical detection, it was also used the colorimetric detection to quantify some of the major adulterants in cocaine seizure samples, such as procaine and phenacetin, as well as analysis and discrimination of explosive compounds (peroxy and nitro explosives) in these low cost portable platforms. All proposed methods were always developed aming at theses characteristics: ease, convenience, low cost and portability for analysis directly at the measurement site with minimal laboratory infrastructure. Finally, we presented some studies conducted during research internship abroad (University of California - San Diego (UCSD)) in the area of Wearable Sensors, which have been developed methods for micronutrient analysis in sweat (Zn) and a metabolite (Uric Acid) in saliva using sensors applied directly to the human body

The aim of this thesis was to develop laterally resolved measurement systems for analysing Metal Insulator Semiconductor (MIS)-based potentiometric chemical sensors. Therefore, the Light-Addressable Potentiometric Sensors (LAPS) principle was used to provide chemical images. Two main system variants were under investigation. The first was a high-resolution so called LAPS-Microscope suitable for life science applications such as the investigation of pH changes of living cells. The second was developed for the analysis of the gas response of alloys on large scale semiconductor samples (25x25 mm2). Therefore, the system was called a LAPS-Macroscope. The LAPS-Microscope resolution depends on the optical focus and semiconductor properties. To reduce the semiconductor dependent resolution, 18 different types of samples were prepared varying the boron and carbon doping in thin silicon films. To analyse the LAPS resolution, different techniques were developed. A resolution improvement down to 3 μm compared to bulk silicon was achieved. A sputter target configuration was developed to produce ternary alloys on semiconductor samples with a continuous gradient in metal concentration. Different analysis methods such as EDX and AES were used to characterise the thin alloy films. Using the LAPS-Macroscope, the gas sensitivity of more than 106 different alloy compositions at a single sample can be investigated. As a result, metal concentration differences down to 0.03% can be distinguished. Typically, the sensor response of 625 different alloy compositions to changes in the hydrogen concentration was investigated within 16 min. The ternary alloy system PdxNiyCo1-x-y was analysed with the LAPS-Macroscope showing that nickel in the range up to 24 atom% reduces the hydrogen sensitivity. There was no significant influence of cobalt in the concentration range tested. Further, poisoning experiments with H2S showed improved behaviour of palladium alloys with nickel in the range of ~5-10 atom%. It was shown that an effective high-throughput method for the characterisation of ternary alloys was established, which is called a Continuous Gradient High Throughput Screening Macroscope CG-HTSM.

Wireless sensors and their applications have become increasingly attractive for industry, building automation and energy control, paving the way for new applications of sensor networks which go well beyond traditional sensor applications. In recent years, there has been a rapid growth in the number of wireless devices operating in close proximity to the human body. Wearable sensor nodes are growing popular not only in our normal living lifestyle, but also within healthcare and military applications, where different radio units operating in/on/off body communicate pervasively. Expectations go beyond the research visions, towards deployment in real-world applications that would empower business processes and future business cases. Although theoretical and simulation models give initial results of the antenna behaviour and the radio channel performance of wireless body area network (WBAN) devices, empirical data from different set of measurements still form an essential part of the radio propagation models. Usually, measurements are performed in laboratory facilities which are equipped with bulky and expensive RF instrumentation within calibrated and controllable environments; thus, the acquired data has the highest possible reliability. However, there are still measurement uncertainties due to cables and connections and significant variations when designs are deployed and measured in real scenarios, such as hospitals wards, commercial buildings or even the battle field. Consequently, more flexible and less expensive measurement tools are required. In this sense, wireless sensor nodes offer not only easiness to deploy or flexibility, but also adaptability to different environments. In this thesis, custom-built wireless sensor nodes are used to characterise different on-body radio channels operating in the IEEE 802.15.4 communication standard at the 2.45 GHz ISM band. Measurement results are also compared with those from the conventional technique using a Vector Network Analyser. The wireless sensor nodes not only diminished the effect of semi-rigid or flexible coaxial cables (scattering or radiation) used with the Vector Network Analyser (VNA), but also provided a more realistic response of the radio link channel. The performance of the wireless sensors is presented over each of the 16 different channels present at the 2.45 GHz band. Additionally, custom-built wireless sensors are used to characterise and model the performance of different on-body radio links in dynamic environments, such as jogging, rowing, and cycling. The use of wireless sensors proves to be less obstructive and more flexible than traditional measurements using coaxial cables, VNA or signal generators. The statistical analysis of different WBAN channels highlighted important radio propagation features which can be used as sport classifiers models and motion detection. Moreover, specific on-body radio propagation channels are further explored, with the aim to recognize physiological features such as motion pattern, breathing activity and heartbeat. The time domain sample data is transformed to the frequency domain using a non-parametric FFT defined by the Welch’s periodogram. The Appendix-Section D explores other digital signal processing techniques which include spectrograms (STFT) and wavelet transforms (WT). Although a simple analysis is presented, strong DSP techniques proved to be good for signal de-noising and multi-resolution analysis. Finally, preliminary results are presented for indoor tracking using the RSS recorded by multiple wireless sensor nodes deployed in an indoor scenario. In contrast to outdoor environments, indoor scenarios are subject to a high level of multipath signals which are dependent on the indoor clutter. The presented algorithm is based on path loss analysis combined with spatial knowledge of each wireless sensor.

Esta tese apresenta os estudos e esforços visando ao desenvolvimento de sensores químicos para aplicações diversas na área forense. Foram desenvolvidos métodos eletroanalíticos para detecção e quantificação de alguns compostos comumente encontrados na adulteração de amostras de drogas de abuso (procaína, fenacetina, aminopirina, paracetamol, levamisol), além da cocaína e estudos fundamentais sobre o comportamento eletroquímico desses compostos. Empregaram-se também métodos eletroquímicos para quantificação de compostos tóxicos e perigosos como explosivos (ácido pícrico) e melamina por exemplo. Os trabalhos utilizando sensores eletroquímicos contemplam modificações eletroquímicas das superfícies eletródicas, utilização de sensores com polímeros molecularmentes impressos (MIP) e eletrodos descartáveis em papel utilizando diferentes técnicas voltamétricas e amperométricas, eletrodo disco rotatório (EDR) e microbalança de cristal de quartzo. Além da fabricação de dispositivos analíticos descartáveis em papel empregando detecção eletroquímica utilizou-se também a detecção colorimétrica para quantificação de alguns dos principais adulterantes de amostras de apreensão de cocaína, como procaína e fenacetina, bem como análises e discriminações de compostos explosivos (peroxi e nitro compostos) nessas plataformas portáteis e de baixo custo. Os métodos foram sempre desenvolvidos visando característicos como: facilidade, praticidade, baixo custo e portabilidade para análises diretamente no local de medida com mínima infraestrutura laboratorial. Por fim, são apresentados alguns estudos realizados durante estágio de pesquisa no exterior (Universidade da Califórnia - San Diego (UCSD)) na área de Wearable Sensors, em que foram desenvolvidos métodos para análises de micronutrientes no suor (zinco) e um metabólito (ácido úrico) na saliva usando sensores aplicados diretamente no corpo humano. / This thesis shows studies and efforts to the development of chemical sensors for different applications in the forensic field. Electroanalytical methods were developed for detection and quantification of some compounds (procaine, phenacetin, aminopyrine, acetaminophen, levamisole) commonly found in the drug of abuse adulteration process and cocaine, as well as, fundamental studies about the electrochemical behavior of these compounds. It was also employed electrochemical methods for quantification of hazardous compounds such as explosives (picric acid) and melamine. Analytical methods with electrochemical sensors included electrochemical modification of electrodic surfaces, molecularly imprinted polymers (MIP), and paper disposable electrochemical devices using different voltammetric and amperometric techniques, rotating disc electrode (RDE) and quartz crystal microbalance. In addition to the fabrication of paper disposable analytical devices with electrochemical detection, it was also used the colorimetric detection to quantify some of the major adulterants in cocaine seizure samples, such as procaine and phenacetin, as well as analysis and discrimination of explosive compounds (peroxy and nitro explosives) in these low cost portable platforms. All proposed methods were always developed aming at theses characteristics: ease, convenience, low cost and portability for analysis directly at the measurement site with minimal laboratory infrastructure. Finally, we presented some studies conducted during research internship abroad (University of California - San Diego (UCSD)) in the area of Wearable Sensors, which have been developed methods for micronutrient analysis in sweat (Zn) and a metabolite (Uric Acid) in saliva using sensors applied directly to the human body

Activity Recognition is concerned with identifying the physical state of a user at a particular point in time. Activity recognition task requires the training of classification algorithm using the processed sensor data from the representative population of users. The accuracy of the generated model often reduces during classification of new instances due to the non-stationary sensor data and variations in user characteristics. Thus, there is a need to adapt the classification model to new user haracteristics. However, the existing approaches to model adaptation in activity recognition are blind. They continuously adapt a classification model at a regular interval without specific and precise detection of the indicator of the degrading performance of the model. This approach can lead to wastage of system resources dedicated to continuous adaptation. This thesis addresses the problem of detecting changes in the accuracy of activity recognition model. The thesis developed a classifier for activity recognition. The classifier uses three statistical summaries data that can be generated from any dataset for similarity based classification of new samples. The weighted ensemble combination of the classification decision from each statistical summary data results in a better performance than three existing benchmarked classification algorithms. The thesis also presents change detection approaches that can detect the changes in the accuracy of the underlying recognition model without having access to the ground truth label of each activity being recognised. The first approach called `UDetect' computes the change statistics from the window of classified data and employed statistical process control method to detect variations between the classified data and the reference data of a class. Evaluation of the approach indicates a consistent detection that correlates with the error rate of the model. The second approach is a distance based change detection technique that relies on the developed statistical summaries data for comparing new classified samples and detects any drift in the original class of the activity. The implemented approach uses distance function and a threshold parameter to detect the accuracy change in the classifier that is classifying new instances. Evaluation of the approach yields above 90% detection accuracy. Finally, a layered framework for activity recognition is proposed to make model adaptation in activity recognition informed using the developed techniques in this thesis.