after vibration endurance testing involves manual inspection only after the test is completed. An adaptable, reliable and low cost real time monitoring and diagnostic system that would interrupt the testing operation at the first onset of a failure is desired. This thesis describes accelerometer based, microphone (single and dualmicrophone) based and acoustic emission sensor based monitoring systems for automotive light assembly failure detection during endurance testing. Preliminary results from accelerometer based and dual-microphone based diagnostic systems show that significant differences between healthy and faulty fog light assemblies can be detected. Based on these initial testing results, subsequent testing and data analysis were conducted for accelerometer based and dual microphone based systems. Four data analysis methods have been used: (1) Averaging signals in the time domain, (2) FFT of time domain waveforms over a specified time, (3) Averaging frequency spectra, and (4) Statistical methods for time domain signals. Individual frequency spectra (from FFT) and the average of multiple frequency spectra have shown potential to distinguish between signals from faulty and healthy light assemblies. Statistical measures, such as, Arithmetic mean (μ) and Kurtosis (K) can also be used to differentiate healthy and faulty light assemblies. In general, this work has shown the good potential to develop methods for adaptable, reliable and low cost real time monitoring and diagnostic systems that would interrupt the testing operation at the first onset of a failure. / Thesis (Master, Mechanical and Materials Engineering) -- Queen's University, 2007-09-28 16:13:47.511

The thesis covers data fusion for aircraft navigation systems in distributed sensor systems. Data fusion methodologies are developed for the design, development, analysis and simulation of multisensor aircraft navigation systems. The problems of sensor failure detection and isolation (FDI), distributed data fusion algorithms and inertial state integrity monitoring in inertial network systems are studied. Various existing integrated navigation systems and Kalman filter architectures are reviewed and a new generalised multisensor data fusion model is presented for the design and development of multisensor navigation systems. Normalised navigation algorithms are described for data fusion filter design of inertial network systems. A normalised measurement model of skewed redundant inertial measurement units (SRIMU) is presented and performance criteria are developed to evaluate optimal configurations of SRIMUs in terms of the measurement accuracy and FDI capability. Novel sensor error compensation filters are designed for the correction of SRIMU measurement errors. Generalised likelihood ratio test (GLRT) methods are improved to detect various failure modes, including short time and sequential moving-window GLRT algorithms. State-identical and state-associated fusion algorithms are developed for two forms of distributed sensor network systems. In particular, innovative inertial network sensing models and inertial network fusion algorithms are developed to provide estimates of inertial vector states and similar node states. Fusion filter-based integrity monitoring algorithms are also presented to detect network sensor failures and to examine the consistency of node state estimates in the inertial network system. The FDI and data fusion algorithms developed in this thesis are tested and their performance is evaluated using a multisensor software simulation system developed during this study programme. The moving-window GLRT algorithms for optimal SRIMU configurations are shown to perform well and are also able to detect jump and drift failures in an inertial network system. It is concluded that the inertial network fusion algorithms could be used in a low-cost inertial network system and are capable of correctly estimating the inertial vector states and the node states.

The thesis covers data fusion for aircraft navigation systems in distributed sensor systems. Data fusion methodologies are developed for the design, development, analysis and simulation of multisensor aircraft navigation systems. The problems of sensor failure detection and isolation (FDI), distributed data fusion algorithms and inertial state integrity monitoring in inertial network systems are studied. Various existing integrated navigation systems and Kalman filter architectures are reviewed and a new generalised multisensor data fusion model is presented for the design and development of multisensor navigation systems. Normalised navigation algorithms are described for data fusion filter design of inertial network systems. A normalised measurement model of skewed redundant inertial measurement units (SRIMU) is presented and performance criteria are developed to evaluate optimal configurations of SRIMUs in terms of the measurement accuracy and FDI capability. Novel sensor error compensation filters are designed for the correction of SRIMU measurement errors. Generalised likelihood ratio test (GLRT) methods are improved to detect various failure modes, including short time and sequential moving-window GLRT algorithms. State-identical and state-associated fusion algorithms are developed for two forms of distributed sensor network systems. In particular, innovative inertial network sensing models and inertial network fusion algorithms are developed to provide estimates of inertial vector states and similar node states. Fusion filter-based integrity monitoring algorithms are also presented to detect network sensor failures and to examine the consistency of node state estimates in the inertial network system. The FDI and data fusion algorithms developed in this thesis are tested and their performance is evaluated using a multisensor software simulation system developed during this study programme. The moving-window GLRT algorithms for optimal SRIMU configurations are shown to perform well and are also able to detect jump and drift failures in an inertial network system. It is concluded that the inertial network fusion algorithms could be used in a low-cost inertial network system and are capable of correctly estimating the inertial vector states and the node states.

This thesis is composed of two parts: a high-speed transceiver front-end design with fault detection for FlexRay-based automative communication systems, and a temperature detector. In the first topic, a high-speed transceiver design with a fault detection circuit compliant with FlexRay standards V2.1 is proposed and realized on silicon. According to FlexRay physical layer standards, a LVDS-like transmitter is utilized to drive the twisted pair to generate a pair of differential signals. By contrast, a three-comparator-based receiver will recover the signals and then detect the mode of the bus. A failure detector is added at the transmitter to detect the operating current magnitude to avoid any over-current hazard. Meanwhile, a short-circuit detector is added at the receiver to detect short-circuit faults to vdd or gnd. The second topic discloses a temperature detector circuit, particularly dedicated for the reliability of any in-car system by sensing the temperature and the process at the same time. A discharging circuit composed of four PMOS transistors discharges until the output voltage is stable. It is composed of several reference voltages and comparators to determine the range of temperature and process. Eventually, by identifying the process corner and the temperature range, the reliability and safty of the car electrics will be ensured.

Fault-tolerant distributed systems often handle failures in two steps: first, detect the failure and, second, take some recovery action. A common approach to detecting failures is end-to-end timeouts, but using timeouts brings problems. First, timeouts are inaccurate: just because a process is unresponsive does not mean that process has failed. Second, choosing a timeout is hard: short timeouts can exacerbate the problem of inaccuracy, and long timeouts can make the system wait unnecessarily. In fact, a good timeout value—one that balances the choice between accuracy and speed—may not even exist, owing to the variance in a system’s end-to-end delays. ƃis dissertation posits a new approach to detecting failures in distributed systems: use information about failures that is local to each component, e.g., the contents of an OS’s process table. We call such information inside information, and use it as the basis in the design and implementation of three failure reporting services for data center applications, which we call Falcon, Albatross, and Pigeon. Falcon deploys a network of software modules to gather inside information in the system, and it guarantees that it never reports a working process as crashed by sometimes terminating unresponsive components. ƃis choice helps applications by making reports of failure reliable, meaning that applications can treat them as ground truth. Unfortunately, Falcon cannot handle network failures because guaranteeing that a process has crashed requires network communication; we address this problem in Albatross and Pigeon. Instead of killing, Albatross blocks suspected processes from using the network, allowing applications to make progress during network partitions. Pigeon renounces interference altogether, and reports inside information to applications directly and with more detail to help applications make better recovery decisions. By using these services, applications can improve their recovery from failures both quantitatively and qualitatively. Quantitatively, these services reduce detection time by one to two orders of magnitude over the end-to-end timeouts commonly used by data center applications, thereby reducing the unavailability caused by failures. Qualitatively, these services provide more specific information about failures, which can reduce the logic required for recovery and can help applications better decide when recovery is not necessary.

Abstract In the interest of improving reliability, electrical monitoring methods were utilized to observe the degradation of electronics interconnections while simultaneously characterizing accelerated testing-induced changes in test structures by means of optical examination, X-ray, scanning acoustic microscopy and scanning electron microscopy. To improve the accuracy of lifetime prediction for the PCSB interconnections investigated in this work, a modified Engelmaier’s solder joint lifetime prediction model was recalibrated. The results show that with most of the presented lead-free (SAC387, SAC405, SAC-In) solder and structure combinations with a large global thermal mismatch (ΔCTE > 10 ppm/°C), lifetime was adequate in the presented TCT ranges of 0‒100 °C and −40‒125 °C, while the amount of non-preferred crack types, i.e. ceramic cracks, was minimized. Degradation of interconnections was characterized using RF measurements both during TCT and intermittently during TCT breaks. A grounded coplanar waveguide was arranged either in a straight back-to-back configuration or together with a filter module with a passband at 22‒24 GHz—both with two transitions—and characterized during cycling breaks up to 25 GHz and 30 GHz, respectively. Besides off-cycle measurements, in-cycle measurements were done on an antenna structure with an in-band at 10‒11 GHz, up to 14 GHz. The results show that the signal response was initially affected at some frequencies as short-duration (< 1 s) glitches in the monitored signal when measured during cycling in 0‒100 °C TCT. Later on the degradation could be observed in the whole frequency band as TCT was continued. Development of the semi-empirical lifetime prediction model for PCSB interconnections showed the temperature range dependency of the correction term to be a second order polynomial instead of a logarithmic one. For components with PCSB BGA, promising prediction results were achieved which differed from the realized lifetime by less than 0.5% at best. / Tiivistelmä Elektroniikkaliitosten rikkoontumisen seurantaan tarkoitettuja sähköisiä monitorointimenetelmiä kehitettiin samanaikaisesti karakterisoimalla testauksella liitoksiin aikaansaatuja muutoksia optisesti, akustisella mikroskoopilla sekä röntgen- ja pyyhkäisyelektronimikroskoopeilla. Liitosten eliniän ennustamiseen soveltuva muokattu Engelmaierin malli kalibroitiin PCSB-liitosten elinikäennusteen tarkkuuden parantamiseksi. Tulosten perusteella useimmille tässä työssä käytetyille lyijyttömille (SAC387, SAC405, SAC-In) juotteille ja suuren termisen epäsovituksen (ΔCTE > 10 ppm/°C) rakenneyhdistelmille eliniät lämpösyklaustesteissä 0‒100 °C ja −40‒125 °C alueilla olivat riittävät ja haitallisimpien murtumien, eli keraamimurtumien, määrä saatiin minimoiduksi. RF-mittauksia käytettiin liitosten vikaantumisen seurantaan sekä lämpösyklauksen aikana että syklausten välillä. Maadoitettua koplanaarista aaltojohtoa käytettiin joko suoraan perättäiskytkennässä tai suodatinmoduulin kanssa, jonka päästökaista oli 22–24 GHz. Rakenteet karakterisoitiin syklausten välillä 25 GHz ja 30 GHz asti tässä järjestyksessä. Näiden mittausten lisäksi 10–11 GHz kaistalla toimivaa antennirakennetta karakterisoitiin syklauksen aikana 14 GHz asti. Tulokset osoittavat, että signaalivasteen muutos ilmenee aluksi joillakin taajuuksilla lyhyinä, alle 1 s mittaisina häiriöpiikkeinä, 0‒100 °C syklauksen aikana. Syklauksen edetessä vasteen huononeminen havaitaan myöhemmin koko mittausalueella. Puolikokeellista elinikäennustemallia tarkasteltaessa havaittiin, että PCSB-liitosten lämpötila-alueesta riippuvia korjauskertoimia kuvasivat logaritmisen riippuvuuden sijaan parhaiten toisen asteen polynomifunktiot. PCSB BGA ‒rakenteille saadun ennusteen ja toteutuneen eliniän välinen ero oli pienimmillään alle 0.5 %.

This thesis proposes a new multi-homing mobile architecture for future heterogeneous network environment. First, a new multi-homed mobile architecture called Multi Network Switching enabled Mobile IPv6 (MNS-MIP6) is proposed which enables a Mobile Node (MN) having multiple communication paths between itself and its Correspondent Node (CN) to take full advantage of being multi-homed. Multiple communication paths exist because MN, CN, or both are simultaneously attached to multiple access networks. A new sub layer is introduced within IP layer of the host’s protocol stack. A context is established between the MN and the CN. Through this context, additional IP addresses are exchanged between the two. Our MNS-MIP6 architecture allows one communication to smoothly switch from one interface/communication path to another. This switch remains transparent to other layers above IP. Second, to make communication more reliable in multi-homed mobile environments, a new failure detection and recovery mechanism called Mobile Reach ability Protocol (M-REAP) is designed within the proposed MNS-MIP6 architecture. The analysis shows that our new mechanism makes communication more reliable than the existing failure detection and recovery procedures in multi-homed mobile environments. Third, a new network selection mechanism is introduced in the proposed architecture which enables a multi-homed MN to choose the network best suited for particular application traffic. A Policy Engine is defined which takes parameters from iv the available networks, compares them according to application profiles and user preferences, and chooses the best network. The results show that in multi-homed mobile environment, load can be shared among different networks/interfaces through our proposed load sharing mechanism. Fourth, a seamless handover procedure is introduced in the system which enables multi-homed MN to seamlessly roam in a heterogeneous network environment. Layer 2 triggers are defined which assist in handover process. When Signal to Noise Ratio (SNR) on a currently used active interface becomes low, a switch is made to a different active interface. We show through mathematical and simulation analysis that our proposed scheme outperforms the existing popular handover management enhancement scheme in MIPv6 networks namely Fast Handover for MIPv6 (FMIPv6). Finally, a mechanism is introduced to allow legacy hosts to communicate with MNS-MIP6 MNs and gain the benefits of reliability, load sharing and seamless handover. The mechanism involves introducing middle boxes in CN’s network. These boxes are called Proxy-MNS boxes. Context is established between the middle boxes and a multi-homed MN.

Chapter 4 of this thesis is a minor revision of the work published in "Daisuke Kotani and Yasuo Okabe, Fast Failure Detection of OpenFlow Channels, The 11th Asian Internet Engineering Conference (AINTEC 2015), pp.32-39, November 2015. http://dx.doi.org/10.1145/2837030.2837035" © ACM 2015. / 京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第19847号 / 情博第598号 / 新制||情||104(附属図書館) / 32883 / 京都大学大学院情報学研究科知能情報学専攻 / (主査)教授 岡部 寿男, 教授 美濃 導彦, 教授 喜多 一 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

One of the main challenges of the integration of unmanned aerial vehicles (UAVs) into today’s society is the risk of in-flight failures, such as motor failure, occurring in populated areas that can result in catastrophic accidents. We propose a framework to manage the consequences of an in-flight system failure and to bring down the aircraft safely without causing any serious accident to people, property, and the UAV itself. This can be done in three steps: a) Detecting a failure, b) Finding a safe landing spot, and c) Navigating the UAV to the safe landing spot. In this thesis, we will look at part b. Specifically, we are working to develop an active system that can detect landing sites autonomously without any reliance on UAV resources. To detect a safe landing site, we are using a deep learning algorithm named "You Only Look Once" (YOLO) that runs on a Jetson Xavier NX computing module, which is connected to a camera, for image processing. YOLO is trained using the DOTA dataset and we show that it can detect landing spots and obstacles effectively. Then by avoiding the detected objects, we find a safe landing spot. The effectiveness of this algorithm will be shown first by comprehensive simulations. We also plan to experimentally validate this algorithm by flying a UAV and capturing ground images, and then applying the algorithm in real-time to see if it can effectively detect acceptable landing spots.

Unmanned Aerial Vehicles (UAVs) are quite successful in maintaining steady flight operations, but propeller failure that exists causes them to experience a possible crash. The objective of this thesis project is to propose a UAV failure detection model as part of the developing framework of an autonomous emergency landing system for UAVs. Health monitoring is integrated where the quadcopter is flown for three cases of propeller faults. Vibration signals are measured during each flight, where a hardware system is designed with Arduino Uno and an Inertial Measurement Unit (IMU) sensor that contains a 3-axis accelerometer and a 3-axis gyroscope, and vibration graphs are made. Once the data is extracted, different parameters (aX, aY, aZ, gX, gY, and gZ) are selected with dimension n ∈ {1,2,3,4,5,6}, and 750 data points are chosen for the K-Means Clustering algorithm. Quadcopter Failure Detection Cluster (QFDC) plots and confusion matrices are created, and three different health states are classified as clusters – normal state, faulty state, and failure state. The parameter set gZ-aZ has the best performance metrics with an accuracy of 92.1%, which is chosen for the decision-making step that involves a Light Emitting Diode (LED) subsystem. Boundary conditions are set from the gZ-aZ QFDC plot where three LEDs turn on based on the specified health state to validate the model. The accuracies of the LED system range between 89% and 95%. Successful failure detection for UAVs would make UAVs safer and more reliable to fly with less imposed restrictions.