Rozprawy doktorskie na temat „The Optimum Detection Probabilities”

PONTIFÍCIA UNIVERSIDADE CATÓLICA DO RIO DE JANEIRO
COORDENAÇÃO DE APERFEIÇOAMENTO DO PESSOAL DE ENSINO SUPERIOR
PROGRAMA DE EXCELENCIA ACADEMICA
Os sistemas de transmissão em bloco, permitem a transmissão de N símbolos de forma simultânea, seja em modulação de portadora única ou multiportadora. A recepção ótima, no sentido de máxima verossimilhança em canais com multipercursos apresenta um custo computacional elevado de AN, onde A é a ordem da constelação (A igual a 2 para BPSK). Para evitar este alto custo computacional é usual fazer a detecção símbolo a símbolo após a equalização. Nesta dissertação é proposto um receptor com detecção por grupos que apresenta uma complexidade intermediária entre o receptor ótimo e os receptores que utilizam detecção símbolo-a-símbolo em sistemas com transmissão em blocos. O tipo de estrutura idealizada agrupa as componentes do bloco equalizado em grupos e realiza detecção conjunta ótima dos símbolos em cada grupo. Com relação possíveis estratégias de agrupamento foram propostos três métodos, o primeiro método faz uma busca exaustiva pelo agrupamento ótimo e tem como consequência um custo computacional elevado para um número grande de símbolos por bloco. Na procura por algoritmos que evitem uma busca exaustiva pelo agrupamento ótimo, mas que resultem em bons ganhos de desempenho, e a sua aplicação em sistemas com um número elevado de símbolos por bloco, foram propostos dois métodos de agrupamento sub-ótimos e eficientes, cujos receptores apresentaram ganhos de desempenho apreciáveis quando comparados ao receptor convencional.
Block transmission systems allow transmissions of N symbols simultaneously, with single carrier or multi-carrier modulation. Maximum likelihood optimal reception in multipath channels have a high computational cost of AN, where A is the constellation order (A iqual 2 for BPSK). To avoid this cost is usual to make symbol-by-symbol detection after equalization. In this work we propose a receiver with group detection that has a good tradeof between computation complexity and bit error rate performance. The idealized structure groups the components of the equalized block in sub-blocks and does optimal joint detection of the symbols in each sub-block. With relation to possible grouping strategies three methods were proposed. The first one searchs for an optimal grouping and has, as a consequence, a high computational cost for block with a large number of symbols. Sub-optimal efficient algorithms that avoid the exhaustive search for the optimal grouping but show good performance gains and feasible application in systems with large number of symbols per block were proposed. The resulted receivers achieved substantial performance gain in comparison with the conventional symbol-by-symbol receiver.

Detectability, the probability that a species is encountered if it inhabits a site, is often overlooked in fisheries research despite its potential to obscure habitat use inferences. Detectability can be estimated using models that also provide an estimate of occupancy (Ψ), the probability that a species inhabits a site. I used these models to estimate both probabilities, and to examine factors affecting detectability and occupancy for three fishes in Chesapeake Bay tributaries: young-of-the-year striped bass (Morone saxatilis), yearling Atlantic croaker (Micropogonias undulatus), and spottail shiner (Notropis hudsonius). Occupancy models were fitted to data from a seine survey conducted during summer, 2008 and 2009, in two Chesapeake Bay tributaries. Key assumptions of occupancy models relate to the extent and timing of fish movement: sites are independent, and no site-specific emigration or immigration occurs. A mark-recapture study of striped bass, and previously published studies of Atlantic croaker and spottail shiner, suggested that these assumptions were reasonable. Detectability differed among species and variation was explained by both gear-related and environmental factors. Effective net length (i.e., the distance from shore the seine was deployed) explained variation in detectability for all species; generally, when the effective seine length exceeded 12 m, detectability was higher and less variable. Detectability varied from early to late summer for Atlantic croaker and spottail shiner but not for striped bass. This variation may be attributed to increased net avoidance by Atlantic croaker during late summer and increased relative abundance of spottail shiner due to recruitment of individuals to the gear. Occupancy of striped bass and Atlantic croaker, both of which are transient species, was high (Ψ>0.80), whereas the resident spottail shiner occupied fewer sites (Ψ=0.59±0.21; mean±SE) and occupancy varied by river (ΨMattaponi=0.36±0.11; ΨPamunkey=0.82±0.10). Occupancy models are useful to identify factors affecting detectability of fishes captured by seines in Chesapeake Bay tributaries, but other fisheries studies would benefit from sampling design modifications that maximize detectability and improve habitat-use inferences.

Whether a species is rare and requires protection or is overabundant and needs control, an accurate estimate of population size is essential for the development of conservation plans and management goals. Wildlife censuses in remote locations or over extensive areas are logistically difficult, frequently biased, and time consuming. My dissertation examined various techniques to determine the probability of detecting animals using remotely sensed imagery. We investigated four procedures that integrated unsupervised classification, texture characteristics, spectral enhancements, and image differencing to identify and count animals in remotely sensed imagery. The semi-automated processes had relatively high errors of over-counting (i.e., greater than 60%) in contrast to low (i.e. less than 19%) under-counting errors. The single-day image differencing had over-counting errors of 53% while the manual interpretation had over-counting errors of 19%. The probability of detection indicates the ability of a process or analyst to detect animals in an image or during an aerial wildlife survey and can adjust total counts to estimate the size of a population. The probabilities of detecting an animal in remotely sensed imagery with semi-automated techniques, single-day image differencing, or manual interpretation were high (e.g. ≥ 80%). Single-day image differencing resulted in the highest probability of detection suggesting this method could provide a new technique for managers to estimate animal populations, especially in open, grassland habitats. Remotely sensed imagery can be successfully used to identify and count animals in isolated or remote areas and improve management decisions. Sightability models, used to estimate population abundances, are derived from count data and the probability of detecting an animal during a census. Global positioning systems (GPS) radio-collared bison in the Henry Mountains of south-central Utah provided a unique opportunity to examine remotely sensed physiographic and survey characteristics for known occurrences of double-counted and missed animals. Bison status (detected, missed, or double-counted) was determined by intersecting helicopter survey paths with bison travel paths during annual helicopter surveys. The probability of detecting GPS-collared bison during the survey ranged from 91% in 2011 to 88% in 2012.

This dissertation applies a commercial flow simulation software package together with common signal processing techniques to the task of accurately detecting leakage in a large commercial gas pipeline. The techniques developed significantly improved the ability to produce accurate, reliable and stable leak detection predictions for the gas transmission pipeline studied and can be applied generally to other pipelines as well. Recommendations for minimum pipeline requirements to implement successful leak detection are also detailed. There are several commercial software packages available that perform some form of leak detection via system modelling. However, due to the commercial aspects of these products, vendors do not publish the detailed methods of leak detection. This thesis identifies the fundamental techniques required to have accurate and reliable leak detection on a gas transmission pipeline, whilst taking into account the lack of measurement data typically encountered on most gas pipelines. The investigation confirmed that a mass balance technique could be successfully used to produce stable leak detection results for compressible flow in gas transmission pipelines. This leak detection [using mass balance] can be achieved without flow measurement along the pipeline, instead, using only pressure and temperature measurements. Although it is recognized that flow measurement data will greatly improve the ability to detect leaks, the focus of this work is on pipelines where this flow measurement data at intermediate points along the pipeline is not available. It was also demonstrated the reliability of the leak detection was improved by the application of on-line signal processing techniques at various stages of the data processing.It was clear early into the investigation that the majority of the errors within the leak detection model were created by random errors from the input field data. These non-systematic errors from the measurement data that included pressure and temperature, produced interference with model output. This interference resembled random “white” noise that was removed by a combination of well established data filtering techniques. The most appropriate process of calculating leak detection flow was determined after analysing the results of different techniques applied to large quantities of actual pipeline operating data. The validation of the on-line techniques developed provides a valuable resource for those wishing to implement similar leak detection schemes elsewhere. Furthermore a software environment was chosen which incorporated an open input and output platform for data that could be interfaced with any operating system. Therefore these techniques can be applied to the numerous Supervisory Control and Data Acquisition (SCADA) systems in operation throughout the gas transmission industry, to provide a low cost solution to leak monitoring.

Thesis (M.S. in Operations Research)--Naval Postgraduate School, June 2007.
Thesis Advisor(s): Lyn R. Whitaker. "June 2007." Includes bibliographical references (p. 95-96). Also available in print.

The optimum ordering strategies for the coded V-BLAST system with capacity achieving temporal codes on each stream are studied in this thesis. Mathematical representations of the optimum detection ordering strategies for the coded V-BLAST under instantaneous rate allocation (IRA), uniform power/rate allocation (URA), instantaneous power allocation(IPA) and instantaneous power/rate allocation (IPRA) are derived. For two transmit antennas, it is shown that the optimum detection strategies are based on the per-stream before-processing channel gains. Based on approximations of the per-stream capacity equation, closed-form expressions of the optimal ordering strategy under the IRA at low and high signal to noise ratio (SNR) are derived. Necessary optimality conditions under the IRA are given. Thresholds for the low, intermediate and high SNR regimes in the 2-Tx-antenna system under the IPRA are determined, and the SNR gain of the ordering is studied for each regime. Performances of simple suboptimal ordering strategies are analysed, some of which perform very close to the optimum one.

This dissertation applies a commercial flow simulation software package together with common signal processing techniques to the task of accurately detecting leakage in a large commercial gas pipeline. The techniques developed significantly improved the ability to produce accurate, reliable and stable leak detection predictions for the gas transmission pipeline studied and can be applied generally to other pipelines as well. Recommendations for minimum pipeline requirements to implement successful leak detection are also detailed. There are several commercial software packages available that perform some form of leak detection via system modelling. However, due to the commercial aspects of these products, vendors do not publish the detailed methods of leak detection. This thesis identifies the fundamental techniques required to have accurate and reliable leak detection on a gas transmission pipeline, whilst taking into account the lack of measurement data typically encountered on most gas pipelines. The investigation confirmed that a mass balance technique could be successfully used to produce stable leak detection results for compressible flow in gas transmission pipelines. This leak detection [using mass balance] can be achieved without flow measurement along the pipeline, instead, using only pressure and temperature measurements. Although it is recognized that flow measurement data will greatly improve the ability to detect leaks, the focus of this work is on pipelines where this flow measurement data at intermediate points along the pipeline is not available. It was also demonstrated the reliability of the leak detection was improved by the application of on-line signal processing techniques at various stages of the data processing.
It was clear early into the investigation that the majority of the errors within the leak detection model were created by random errors from the input field data. These non-systematic errors from the measurement data that included pressure and temperature, produced interference with model output. This interference resembled random “white” noise that was removed by a combination of well established data filtering techniques. The most appropriate process of calculating leak detection flow was determined after analysing the results of different techniques applied to large quantities of actual pipeline operating data. The validation of the on-line techniques developed provides a valuable resource for those wishing to implement similar leak detection schemes elsewhere. Furthermore a software environment was chosen which incorporated an open input and output platform for data that could be interfaced with any operating system. Therefore these techniques can be applied to the numerous Supervisory Control and Data Acquisition (SCADA) systems in operation throughout the gas transmission industry, to provide a low cost solution to leak monitoring.

COORDENAÇÃO DE APERFEIÇOAMENTO DO PESSOAL DE ENSINO SUPERIOR
Este trabalho propõe um receptor multiusuário híbrido para utilização em um sistema DS/CDMA com modulação M-PSK e codificação convolucional. O receptor proposto é composto por um receptor decorrelator ao final do qual é realizada uma decodificação convolucional por grupos. Resultados de desempenho obtidos por simulações para grupos de dois usuários são apresentados e comparados com o desempenho do receptor convencional, formado por um banco de filtros casados, e com o desempenho do receptor decorrelator já proposto na literatura.
This work proposes a hybrid multiuser receiver for a DS/CDMA system employing M-PSK modulation and convolutional coding. The proposed receiver is composed by a decorrelator receiver followed by a group convolutional decoder. Perfomance results obtained by simulation for groups of two users are presented and compared to the performance of a conventional receiver, composed of a bank of matched filters, and to the decorrelator receiver itself.

In this thesis, firstly we introduce various reduced-complexity near-optimum Sphere Detection (SD) algorithms, including the well-known depth-first SD, the K-best SD as well as the recently proposed Optimized Hierarchy Reduced Search Algorithm (OHRSA), followed by comparative studies of their applications, characteristics, performance and complexity in the context of uncoded non-cooperative Multiple-Input Multiple-Output (MIMO) systems using coherent detection. Particular attention is devoted to Spatial Division Multiple Accessing (SDMA) aided Orthogonal Frequency Division Multiplexing (OFDM) systems, which are considered to constitute a promising candidate for next-generation mobile communications. It is widely recognized that the conventional List SD (LSD) employed in channel-coded iterative detection aided systems may still impose a potentially excessive complexity, especially when it is applied to high-throughput scenarios employing high-order modulation schemes and/or supporting a high number of transmit antennas/users. Hence, in this treatise three complexity-reduction schemes are devised specifically for LSD-aided iterative receivers in the context of high-throughput channel-coded SDMA/OFDM systems in order to maintain a near-optimum performance at a reduced complexity. Explicitly, based on the exploitation of the soft-bit-information fed back by the channel decoder, the iterative center-shifting and Apriori-LLR-Threshold (ALT) schemes are contrived, which are capable of achieving a significant complexity reduction. Additionally, a powerful three-stage serially concatenated scheme is created by intrinsically amalgamating our proposed center-shifting-assisted SD with the decoder of a Unity-Rate-Code (URC). For the sake of achieving a near-capacity performance, Irregular Convolutional Codes (IrCCs) are used as the outer code for the proposed iterative center-shifting SD aided three-stage system. In order to attain extra coding gains along with transmit diversity gains for Multi-User MIMO (MU-MIMO) systems, where each user is equipped with multiple antennas, we contrive a multilayer tree-search based K-best SD scheme, which allows us to apply the Sphere Packing (SP) aided Space-Time Block Coding (STBC) scheme to the MU-MIMO scenarios, where a near Maximuma-Posteriori (MAP) performance is achieved at a low complexity. An alternative means of achieving transmit diversity while circumventing the cost and size constraints of implementing multiple antennas on a pocket-sized mobile device is cooperative diversity, which relies on antenna-sharing amongst multiple cooperating single-antenna-aided users. We design a realistic cooperative system, which operates without assuming the knowledge of the Channel State Information (CSI) at transceivers by employing differentially encoded modulation at the transmitter and non-coherent detection at the receiver. Furthermore, a newMultiple-Symbol Differential Sphere Detection (MSDSD) is contrived in order to render the cooperative system employing either the Differential Amplify-and-Forward (DAF) or the Differential Decode-and-Forward (DDF) protocol more robust to the detrimental channel-envelope fluctuations of high-velocity mobility environments. Additionally, for the sake of achieving the best possible performance, a resource optimized hybrid relaying scheme is proposed for exploiting the complementarity of the DAF- and DDF-aided systems. Finally, we investigate the benefits of introducing cooperative mechanisms into wireless networks from a pure channel capacity perspective and from the practical perspective of approaching the Discrete-input Continuous-output Memoryless Channel (DCMC) capacity of the cooperative network with the aid of our proposed Irregular Distributed Hybrid Concatenated Differential (Ir-DHCD) coding scheme.

After Qualcomm's proposal of the IS-95 standard, code-division multiple access (CDMA) gained popularity as an alternative multiple-access scheme in cellular and personal communication systems (PCS). Besides the advantage of allowing asynchronous operation of the users, CDMA direct-sequence spread spectrum (DS-SS) offers resistance to frequency selective fading and graceful degradation of the performance as the number of users increases.

Orthogonality of the signals in time-division multiple access and frequency-division multiple access is inherent from the nature of the multiple access scheme. In a CDMA system, orthogonality of the signals is not guaranteed in general. Consequently, the performance of conventional correlation receivers suffers.

Sub-optimum receivers which use knowledge of the interfering signals have been investigated by other researchers. These receivers attempt to cancel the multi-user interference by despreading the interfering users. Hence, these receivers require knowledge about all the spreading codes, amplitude levels, and signal timing, and are, in general, computationally intensive.

In this thesis, a technique is presented for which a high degree of interference rejection can be obtained without the necessity of despreading each user. It is shown that exploiting spectral correlation can help mitigate the effects of the multiple-access interference. If code-on-pulse DS-SS modulation is used, a cyclic form of the Wiener filter provides substantial improvements in performance in terms of bit error rate and user capacity. Furthermore, it is shown, that a special error-criterion should be used to adapt the weights of the filter.

The computational complexity of the receiver is equivalent to that of conventional equalizers.

Master of Science

Aquatic biodiversity continues to decline as humans modify the landscape. A population-level approach is necessary to address the mechanisms of impairment in urban stream habitats. When estimating population-level parameters, incomplete detection of individuals must be accounted for to ensure unbiased estimates. In this thesis, I examined differences in the detection probabilities of stream fishes, and used estimates of size/age-specific detection probabilities to reduce bias in estimates of the reproductive success of various fish species in urban and forested stream habitats. In Chapter 1, I examined differences in detection probabilities of stream fishes among electrofishing passes and size/age groups in the middle New River basin, Virginia. I also examined differences in detection probabilities between two physiographic regions: the middle New River basin, and the upper Wabash River basin, Indiana; and evaluated differences between single- and multiple-season estimation methods. I found that for most species, detection probabilities do not differ among electrofishing passes, size/age classes, between the two regions, or between single- and multiple-season estimation methods. I used size/age-specific estimates of detection probabilities to remove bias from relative abundance estimates of steam fish populations in Chapter 2. In Chapter 2, I examined the reproductive success of six lithophilic and speleophilic fishes in urban and forested reaches of 2nd-4th Strahler-order streams in the middle New River basin. I found that binary classification is a poor method of quantifying reproductive success, that the age distributions of many lithophilic and speleophilic species are dominated by adult individuals in urban habitats, and that the population growth rates of speleophils is reduced in urban habitats. These results suggest that although detection probabilities may be equal among various sources of variation, managers should verify this assumption before assuming equal detectability. The results also suggest that reduced reproductive success of speleophilic and lithophilic species in urban stream habitats may be a mechanism of their impairment.
Master of Science

According to IRIG 106-15, the ARTM CPM waveform, a kind of multi-h partial response continuous phase modulation (CPM), has almost three times the spectral efficiency of PCM/FM and approximately the same detection efficiency of PCM/FM. But the improved spectral efficiency of ARTM CPM comes at the price of computational complexity in the receiver. This paper focuses on multi-h full response CPM, which generally has less detection complexity than ARTM CPM, but also has good spectral efficiency and detection efficiency. Taking the minimum Euclidean distance, spectral efficiency and detection complexity as judgment criterions, optimum parameter combinations for multi-h full response CPM are presented.

This thesis discusses two aspects of spatial statistics: sampling and prediction. In spatial statistics, we observe some phenomena in space. Space is typically of two or three dimensions, but can be of higher dimension. Questions in mind could be; What is the total amount of gold in a gold-mine? How much precipitation could we expect in a specific unobserved location? What is the total tree volume in a forest area? In spatial sampling the aim is to estimate global quantities, such as population totals, based on samples of locations (papers III and IV). In spatial prediction the aim is to estimate local quantities, such as the value at a single unobserved location, with a measure of uncertainty (papers I, II and V). In papers III and IV, we propose sampling designs for selecting representative probability samples in presence of auxiliary variables. If the phenomena under study have clear trends in the auxiliary space, estimation of population quantities can be improved by using representative samples. Such samples also enable estimation of population quantities in subspaces and are especially needed for multi-purpose surveys, when several target variables are of interest. In papers I and II, the objective is to construct valid prediction intervals for the value at a new location, given observed data. Prediction intervals typically rely on the kriging predictor having a Gaussian distribution. In paper I, we show that the distribution of the kriging predictor can be far from Gaussian, even asymptotically. This motivated us to propose a semiparametric method that does not require distributional assumptions. Prediction intervals are constructed from the plug-in ordinary kriging predictor. In paper V, we consider prediction in the presence of left-censoring, where observations falling below a minimum detection limit are not fully recorded. We review existing methods and propose a semi-naive method. The semi-naive method is compared to one model-based method and two naive methods, all based on variants of the kriging predictor.

The goal of this work is to expand the knowledge on the field of Venn Prediction employed with Survival Data. Standard Venn Predictors have been used with Random Forests and binary classification tasks. However, they have not been utilised to predict events with Survival Data nor in combination with Random Survival Forests. With the help of a Data Transformation, the survival task is transformed into several binary classification tasks. One key aspect of Venn Prediction are the categories. The standard number of categories is two, one for each class to predict. In this work, the usage of ten categories is explored and the performance differences between two and ten categories are investigated. Seven data sets are evaluated, and their results presented with two and ten categories. For the Brier Score and Reliability Score metrics, two categories offered the best results, while Quality performed better employing ten categories. Occasionally, the models are too optimistic. Venn Predictors rectify this performance and produce well-calibrated probabilities.
Målet med detta arbete är att utöka kunskapen om området för Venn Prediction som används med överlevnadsdata. Standard Venn Predictors har använts med slumpmässiga skogar och binära klassificeringsuppgifter. De har emellertid inte använts för att förutsäga händelser med överlevnadsdata eller i kombination med Random Survival Forests. Med hjälp av en datatransformation omvandlas överlevnadsprediktion till flera binära klassificeringsproblem. En viktig aspekt av Venn Prediction är kategorierna. Standardantalet kategorier är två, en för varje klass. I detta arbete undersöks användningen av tio kategorier och resultatskillnaderna mellan två och tio kategorier undersöks. Sju datamängder används i en utvärdering där resultaten presenteras för två och tio kategorier. För prestandamåtten Brier Score och Reliability Score gav två kategorier de bästa resultaten, medan för Quality presterade tio kategorier bättre. Ibland är modellerna för optimistiska. Venn Predictors korrigerar denna prestanda och producerar välkalibrerade sannolikheter.

Dans l'analyse des émetteurs alpha et beta dans l'eau par scintillation liquide, la problématique de la discrimination des rayonnements se pose afin d'atteindre les limites de détection souhaitées. Dans le cas rare ou les énergies permettent une discrimination, nous avons mis en place un protocole d'analyse simultanée tritium/alpha sur des frottis. A l'aide d'une méthode a trois fenêtres il permet d'atteindre les limites de détection de 0,4 Bq/cm2 en tritium et de 0,04 Bq/cm2 en alpha.De façon générale, aucune distinction entre les spectres n'est possible et un ajustement du fond sous le pic alpha ne permet pas d'atteindre les limites de détection définies par les normes. Cependant le processus physique de scintillation permet une distinction des rayonnements, phénomène utilise par certains appareils, ceux-ci proposent donc le réglage par l'utilisateur d'un paramètre discriminant seuil. Nous avons développé un protocole de réglage permettant de reconstituer les spectres alpha et beta discrimines au mieux, rendant accessible une limite de détection de 0,03 Bq/L en alpha.Nous avons par ailleurs développé une approche innovante utilisant un détecteur en coïncidence dont les photomultiplicateurs sont connectes directement a une carte numérique d'acquisition. Chaque impulsion numérisée est ensuite traitée selon plusieurs critères simultanément, à la place des deux habituels. Aucun ne permet a lui seul d'obtenir une discrimination selon un seuil unique, mais la distribution des évènements dans chaque spectre est différente : un fenêtrage des évènements alpha permet d'atteindre une limite de détection de 0,2 Bq/L, quatre fois moindre que par une discrimination en énergie seule.

Every day we share our personal information through digital systems which are constantly exposed to threats. For this reason, security-oriented disciplines of signal processing have received increasing attention in the last decades: multimedia forensics, digital watermarking, biometrics, network monitoring, steganography and steganalysis are just a few examples. Even though each of these fields has its own peculiarities, they all have to deal with a common problem: the presence of one or more adversaries aiming at making the system fail. Adversarial Signal Processing lays the basis of a general theory that takes into account the impact that the presence of an adversary has on the design of effective signal processing tools. By focusing on the application side of Adversarial Signal Processing, namely adversarial information fusion in distributed sensor networks, and adopting a game-theoretic approach, this thesis contributes to the above mission by addressing four issues. First, we address decision fusion in distributed sensor networks by developing a novel soft isolation defense scheme that protects the network from adversaries, specifically, Byzantines. Second, we develop an optimum decision fusion strategy in the presence of Byzantines. In the next step, we propose a technique to reduce the complexity of the optimum fusion by relying on a novel nearly-optimum message passing algorithm based on factor graphs. Finally, we introduce a defense mechanism to protect decentralized networks running consensus algorithm against data falsification attacks.

Thesis (M.S.)--University of Wisconsin--Madison, 1991.
Typescript. eContent provider-neutral record in process. Description based on print version record. Includes bibliographical references (leaves 99-100).

We investigate sequential event detection problems arising in Wireless Sensor Networks (WSNs). A number of battery–powered sensor nodes of the same sensing modality are deployed in a region of interest(ROI). By an event we mean a random time(and, for spatial events, a random location) after which the random process being observed by the sensor field experiences a change in its probability law. The sensors make measurements at periodic time instants, perform some computations, and then communicate the results of their computations to the fusion centre. The decision making algorithm in the fusion centre employs a procedure that makes a decision on whether the event has occurred or not based on the information it has received until the current decision instant. We seek event detection algorithms in various scenarios, that are optimal in the sense that the mean detection delay (delay between the event occurrence time and the alarm time) is minimum under certain detection error constraints. In the first part of the thesis, we study event detection problems in a small extent network where the sensing coverage of any sensor includes the ROI. In particular, we are interested in the following problems: 1) quickest event detection with optimal control of the number of sensors that make observations(while the others sleep),2) quickest event detection on wireless ad hoc networks, and3) optimal transient change detection. In the second part of the thesis, we study the problem of quickest detection and isolation of an event in a large extent sensor network where the sensing coverage of any sensor is only a small portion of the ROI. One of the major applications envisioned for WSNs is detecting any abnormal activity or intrusions in the ROI. An intrusion is typically a rare event, and hence, much of the energy of sensors gets drained away in the pre–intrusion period. Hence, keeping all the sensors in the awake state is wasteful of resources and reduces the lifetime of the WSN. This motivates us to consider the problem of sleep–wake scheduling of sensors along with quickest event detection. We formulate the Bayesian quickest event detection problem with the objective of minimising the expected total cost due to i)the detection delay and ii) the usage of sensors, subject to the constraint that the probability of false alarm is upper bounded by .We obtain optimal event detection procedures, along with optimal closed loop and open loop control for the sleep–wake scheduling of sensors. In the classical change detection problem, at each sampling instant, a batch of samples(where is the number of sensors deployed in the ROI) is generated at the sensors and reaches the fusion centre instantaneously. However, in practice, the communication between the sensors and the fusion centre is facilitated by a wireless ad hoc network based on a random access mechanism such as in IEEE802.11 or IEEE802.15.4. Because of the medium access control(MAC)protocol of the wireless network employed, different samples of the same batch reach the fusion centre after random delays. The problem is to detect the occurrence of an event as early as possible subject to a false alarm constraint. In this more realistic situation, we consider a design in which the fusion centre comprises a sequencer followed by a decision maker. In earlier work from our research group, a Network Oblivious Decision Making (NODM) was considered. In NODM, the decision maker in the fusion centre is presented with complete batches of observations as if the network was not present and makes a decision only at instants at which these batches are presented. In this thesis, we consider the design in which the decision maker makes a decision at all time instants based on the samples of all the complete batches received thus far, and the samples, if any, that it has received from the next (partial) batch. We show that for optimal decision making the network–state is required by the decision maker. Hence, we call this setting Network Aware Decision Making (NADM). Also, we obtain a mean delay optimal NADM procedure, and show that it is a network–state dependent threshold rule on the a posteriori probability of change. In the classical change detection problem, the change is persistent, i.e., after the change–point, the state of nature remains in the in–change state for ever. However, in applications like intrusion detection, the event which causes the change disappears after a finite time, and the system goes to an out–of–change state. The distribution of observations in the out–of–change state is the same as that in the pre–change state. We call this short–lived change a transient change. We are interested in detecting whether a change has occurred, even after the change has disappeared at the time of detection. We model the transient change and formulate the problem of quickest transient change detection under the constraint that the probability of false alarm is bounded by . We also formulate a change detection problem which maximizes the probability of detection (i.e., probability of stopping in the in–change state) subject to the probability of false alarm being bounded by . We obtain optimal detection rules and show that they are threshold d rules on the a posteriori probability of pre–change, where the threshold depends on the a posteriori probabilities of pre–change, in–change, and out–of–change states. Finally, we consider the problem of detecting an event in a large extent WSN, where the event influences the observations of sensors only in the vicinity of where it occurs. Thus, in addition to the problem of event detection, we are faced with the problem of locating the event, also called the isolation problem. Since the distance of the sensor from the event affects the mean signal level that the sensor node senses, we consider a realistic signal propagation model in which the signal strength decays with distance. Thus, the post–change mean of the distribution of observations across sensors is different, and is unknown as the location of the event is unknown, making the problem highly challenging. Also, for a large extent WSN, a distributed solution is desirable. Thus, we are interested in obtaining distributed detection/isolation procedures which are detection delay optimal subject to false alarm and false isolation constraints. For this problem, we propose the following local decision rules, MAX, HALL, and ALL, which are based on the CUSUM statistic, at each of the sensor nodes. We identify corroborating sets of sensor nodes for event location, and propose a global rule for detection/isolation based on the local decisions of sensors in the corroborating sets. Also, we show the minimax detection delay optimality of the procedures HALL and ALL.

We investigate sequential event detection problems arising in Wireless Sensor Networks (WSNs). A number of battery–powered sensor nodes of the same sensing modality are deployed in a region of interest(ROI). By an event we mean a random time(and, for spatial events, a random location) after which the random process being observed by the sensor field experiences a change in its probability law. The sensors make measurements at periodic time instants, perform some computations, and then communicate the results of their computations to the fusion centre. The decision making algorithm in the fusion centre employs a procedure that makes a decision on whether the event has occurred or not based on the information it has received until the current decision instant. We seek event detection algorithms in various scenarios, that are optimal in the sense that the mean detection delay (delay between the event occurrence time and the alarm time) is minimum under certain detection error constraints. In the first part of the thesis, we study event detection problems in a small extent network where the sensing coverage of any sensor includes the ROI. In particular, we are interested in the following problems: 1) quickest event detection with optimal control of the number of sensors that make observations(while the others sleep),2) quickest event detection on wireless ad hoc networks, and3) optimal transient change detection. In the second part of the thesis, we study the problem of quickest detection and isolation of an event in a large extent sensor network where the sensing coverage of any sensor is only a small portion of the ROI. One of the major applications envisioned for WSNs is detecting any abnormal activity or intrusions in the ROI. An intrusion is typically a rare event, and hence, much of the energy of sensors gets drained away in the pre–intrusion period. Hence, keeping all the sensors in the awake state is wasteful of resources and reduces the lifetime of the WSN. This motivates us to consider the problem of sleep–wake scheduling of sensors along with quickest event detection. We formulate the Bayesian quickest event detection problem with the objective of minimising the expected total cost due to i)the detection delay and ii) the usage of sensors, subject to the constraint that the probability of false alarm is upper bounded by .We obtain optimal event detection procedures, along with optimal closed loop and open loop control for the sleep–wake scheduling of sensors. In the classical change detection problem, at each sampling instant, a batch of samples(where is the number of sensors deployed in the ROI) is generated at the sensors and reaches the fusion centre instantaneously. However, in practice, the communication between the sensors and the fusion centre is facilitated by a wireless ad hoc network based on a random access mechanism such as in IEEE802.11 or IEEE802.15.4. Because of the medium access control(MAC)protocol of the wireless network employed, different samples of the same batch reach the fusion centre after random delays. The problem is to detect the occurrence of an event as early as possible subject to a false alarm constraint. In this more realistic situation, we consider a design in which the fusion centre comprises a sequencer followed by a decision maker. In earlier work from our research group, a Network Oblivious Decision Making (NODM) was considered. In NODM, the decision maker in the fusion centre is presented with complete batches of observations as if the network was not present and makes a decision only at instants at which these batches are presented. In this thesis, we consider the design in which the decision maker makes a decision at all time instants based on the samples of all the complete batches received thus far, and the samples, if any, that it has received from the next (partial) batch. We show that for optimal decision making the network–state is required by the decision maker. Hence, we call this setting Network Aware Decision Making (NADM). Also, we obtain a mean delay optimal NADM procedure, and show that it is a network–state dependent threshold rule on the a posteriori probability of change. In the classical change detection problem, the change is persistent, i.e., after the change–point, the state of nature remains in the in–change state for ever. However, in applications like intrusion detection, the event which causes the change disappears after a finite time, and the system goes to an out–of–change state. The distribution of observations in the out–of–change state is the same as that in the pre–change state. We call this short–lived change a transient change. We are interested in detecting whether a change has occurred, even after the change has disappeared at the time of detection. We model the transient change and formulate the problem of quickest transient change detection under the constraint that the probability of false alarm is bounded by . We also formulate a change detection problem which maximizes the probability of detection (i.e., probability of stopping in the in–change state) subject to the probability of false alarm being bounded by . We obtain optimal detection rules and show that they are threshold d rules on the a posteriori probability of pre–change, where the threshold depends on the a posteriori probabilities of pre–change, in–change, and out–of–change states. Finally, we consider the problem of detecting an event in a large extent WSN, where the event influences the observations of sensors only in the vicinity of where it occurs. Thus, in addition to the problem of event detection, we are faced with the problem of locating the event, also called the isolation problem. Since the distance of the sensor from the event affects the mean signal level that the sensor node senses, we consider a realistic signal propagation model in which the signal strength decays with distance. Thus, the post–change mean of the distribution of observations across sensors is different, and is unknown as the location of the event is unknown, making the problem highly challenging. Also, for a large extent WSN, a distributed solution is desirable. Thus, we are interested in obtaining distributed detection/isolation procedures which are detection delay optimal subject to false alarm and false isolation constraints. For this problem, we propose the following local decision rules, MAX, HALL, and ALL, which are based on the CUSUM statistic, at each of the sensor nodes. We identify corroborating sets of sensor nodes for event location, and propose a global rule for detection/isolation based on the local decisions of sensors in the corroborating sets. Also, we show the minimax detection delay optimality of the procedures HALL and ALL.

This report analyzes the impact of some important factors on the prediction of detection probabilities for radiation portal monitors (RPMs). The application of innovative detection technology to improve operational sensitivity of RPMs has received increasing attention in recent decades. In particular, two alarm algorithms, gross count and energy windowing, have been developed to try to distinguish between special nuclear material (SNM) and naturally occurring radioactive material (NORM). However, the use of the two detection strategies is quite limited due to a very large number of unpredictable threat scenarios. We address this problem by implementing a new Monte Carlo radiation transport simulation approach to model a large set of threat scenarios with predefined conditions. In this report, our attention is focused on the effect of two important factors on the detected energy spectra in RPMs, the mass of individual nuclear isotopes and the thickness of shielding materials. To study the relationship between these factors and the resulting spectra, we apply several advanced statistical regression models for different types of data, including a multinomial logit model, an ordinal logit model, and a curvilinear regression model. By utilizing our new simulation technique together with these sophisticated regression models, we achieve a better understanding of the system response under various conditions. We find that the different masses of the isotopes change the isotopes’ effect on the energy spectra. In analyzing the joint impact of isotopes’ mass and shielding thickness, we obtain a nonlinear relation between the two factors and the gross count of gamma photons in the energy spectrum.
text

碩士
國立中央大學
電機工程研究所
89
Sub-optimum Coherent Detection of OMOAM in AWGN Abstract A novel class of coherent Orthogonally Multiplexed Orthogonal Amplitude Modulation [1](OMOAM) signals is investigated in this thesis. An exact analysis on the bit error probability is presented for a coherent OMOAM sub-optimum receiver, and verified by computer simulation. The performance comparison between the optimum [1] and sub-optimum OMOAM receivers are also illuminated. The performance of the sub-optimum coherent receiver in the presence of imperfect power estimation is also considered.

It is widely recognized that multicarrier systems such as orthogonal frequency division multiplexing (OFDM) are suitable for severely time-dispersive channels. However, it is also recognized that multicarrier signals have high envelope fluctuations which make them especially sensitive to nonlinear distortion effects. In fact, it is almost unavoidable to have nonlinear distortion effects in the transmission chain. For this reason, it is essential to have a theoretical, accurate characterization of nonlinearly distorted signals not only to evaluate the corresponding impact of these distortion effects on the system’s performance, but also to develop mechanisms to combat them. One of the goals of this thesis is to address these challenges and involves a theoretical characterization of nonlinearly distorted multicarrier signals in a simple, accurate way. The other goal of this thesis is to study the optimum detection of nonlinearly distorted, multicarrier signals. Conventionally, nonlinear distortion is seen as a noise term that degrades the system’s performance, leading even to irreducible error floors. Even receivers that try to estimate and cancel it have a poor performance, comparatively to the performance associated to a linear transmission, even with perfect cancellation of nonlinear distortion effects. It is shown that the nonlinear distortion should not be considered as a noise term, but instead as something that contains useful information for detection purposes. The adequate receiver to take advantage of this information is the optimum receiver, since it makes a block-by-block detection, allowing us to exploit the nonlinear distortion which is spread along the signal’s band. Although the optimum receiver for nonlinear multicarrier schemes is too complex, due to its necessity to compare the received signal with all possible transmitted sequences, it is important to study its potential performance gains. In this thesis, it is shown that the optimum receiver outperforms the conventional detection, presenting gains not only relatively to conventional receivers that deal with nonlinear multicarrier signals, but also relatively to conventional receivers that deal with linear, multicarrier signals. We also present sub-optimum receivers which are able to approach the performance gains associated to the optimum detection and that can even outperform the conventional linear, multicarrier schemes.

碩士
國立中央大學
資訊工程研究所
99
Skin detection plays an important role in a wide range of image processing applications. Common skin detection methods need to set skin color cluster decision boundaries in different color space components. When the pixel values fall within these decision boundaries, they would be defined as skin pixels. However, this method may not work well in the scenes with complex and time-varying illumination. Common property-based shadow detection methods have the same problem. To overcome these shortcomings, a composite improved approach to detect moving skin regions is presented in this thesis. This thesis proposes Gaussian mixture models with skin and shadow probabilities (GMM-SS) to detect moving skin regions. The system is separated into two parts. The first part of GMM-SS uses Gaussian mixture models with skin probability to detect moving skin regions. The learning rate in GMM-SS is lower for pixels with higher skin robabilities. The moving skin regions would be constructed into background Gaussian more slowly by the lower learning rate, and vice versa. The second part of GMM-SS uses Gaussian mixture models with shadow probability to remove casting shadows to get pure moving skin region. The shadow Model is used to identify distributions of pixel values that could represent shadowed surfaces. In this model, it modifies shadow distribution learning rate with each pixel’s shadow probability and makes it more flexible for shadow detection. The experiment results show that our system is more efficient and robust for moving skin region detection and removing shadow regions. We compare it with common skin detection and Gaussian mixture models. It keeps steady skin detection rates and low false alarm rates in most situations than common skin and shadow detection methods.

碩士
國立清華大學
電機工程學系
88
A real-time symbol-optimum detection algorithm for a CDMA system is developed in this paper by a truncation in the backward schedule. Simulations are present to show that the proposed truncated symbol-optimum detection provides negligible performance loss compared to the original symbol-optimum detection. It allows a tradeoff between the latency, the computation load, and the performance. A systolic array implementation of the algorithm is derived to exploit the suitability of the algorithm for parallel processing. With such an implementation, the processing speed is greatly improved.

Tese de mestrado. Biologia (Ecologia e Gestão Ambiental). Universidade de Lisboa, Faculdade de Ciências, 2014
A agricultura tem uma longa história, desde as suas origens no leste Mediterrânico há 10 000 anos atrás, até a uma das formas mais generalizada dos atuais usos do solo na Europa. Geralmente, as paisagens agrícolas apresentam-se sob a forma de mosaicos com diferentes usos do solo, dominadas por culturas e pastagens nativas ou melhoradas utilizadas para pastagem. Além de espacialmente heterogéneos, os mosaicos agrícolas também são dinâmicos em diversas escalas temporais, sendo que, espécies que vivem em áreas agrícolas podem também interagir com as mudanças ambientais ao longo do tempo. Na Europa, as paisagens agrícolas desempenham um papel extremamente importante para a conservação da biodiversidade, visto que muitas espécies dependem diretamente dos sistemas agrícolas tradicionais. Contudo, a diversidade destas paisagens tradicionais está a decrescer drasticamente, quer devido ao abandono das terras, quer devido à intensificação do uso dos solos. Em particular, a intensificação da agricultura é comumente apontada como uma ameaça à sobrevivência de muitas espécies, normalmente adaptadas aos sistemas agrícolas tradicionais. Desde modo, compreender de que forma é que a intensificação da agricultura afeta a biodiversidade tornou-se num dos objetivos principais quer da ecologia, quer da conservação da vida selvagem. No entanto, alcançar este objetivo traz também muitos desafios, tanto conceptualmente, como metodologicamente, sendo que neste contexto, o conceito de metapopulação será particularmente útil na compreensão de como a perda e a fragmentação dos habitats pode afetar a persistência de espécies espacialmente estruturadas. Atualmente, o conceito geral de metapopulação consiste na ideia de que habitats de menores dimensões e mais isolados apresentam menor probabilidade de serem ocupados, enquanto que habitats maiores e menos isolados estão mais propensos a serem colonizados. Em condições de equilíbrio, a persistência de uma metapopulação e uma escala regional, resulta do equilíbrio entre eventos de extinção e colonização local. Assim, os modelos metapopulacionais podem ser particularmente úteis para avaliar de que forma é que a transformação da paisagem pode afetar a dinâmica extinção/colonização e a persistência a longo prazo das populações, desde que (i) fragmentos de habitats adequados possam ser diferenciados a partir da matriz envolvente; (ii) todas as populações locais apresentem determinado risco de extinção, em algum momento; e (iii) a dispersão entre habitats e os eventos de colonização locais ocorram lentamente, de modo a que a dinâmica das populações dentro dos habitats seja assíncrona. Normalmente, o foco dos modelos de metapopulações incide sobre a dinâmica de ocupação das espécies nos habitats, enquanto a dinâmica das populações locais é muitas vezes negligenciada. No entanto, quando se trabalha com ocupação de espécies, é fundamental ter em consideração a possibilidade de que uma espécie pode estar presente num habitat e não ser detetada durante a amostragem. Deste modo, se a deteção imperfeita não for contabilizada nos modelos, estas "falsas ausências” podem levar a inferências incorretas sobre a dinâmica local da espécie alvo. Para solucionar este problema, foram desenvolvidos modelos que incorporam explicitamente deteções imperfeitas no processo de modelagem para permitir uma estimativa imparcial da probabilidade de ocupação (dada como a fração de sítios ocupados), e das co variáveis que a afetam. Em modelos de ocupação que têm em conta deteções imperfeitas, a probabilidade de deteção é normalmente assumida como sendo menor do que 1. Para incorporar deteções imperfeitas, é necessário que, pelo menos, alguns locais sejam amostrados mais do que uma vez, dentro de um período de tempo relativamente curto, envolvendo, portanto, tanto uma replicação espacial como temporal. A ideia básica associada a estes modelos é que dentro do período de tempo da amostragem, assume-se que não ocorrem mudanças na ocupação desses mesmos habitats (ou seja, os habitats ou estão sempre ocupados ou desocupados pelas espécies), sendo que as mudanças só podem ocorrer entre diferentes períodos de amostragem, devido a eventos de colonização e extinção local. Assim, com base em amostragens repetidas de presença/ausência das espécies, estes modelos ajudam a estimar a probabilidade de deteção de pelo menos um indivíduo da espécie-alvo durante um período de amostragem, sabendo que os indivíduos estão presentes na área de estudo. Contudo, apesar de serem simples de planear, os métodos de amostragem necessários para desenvolver estes modelos são muitas vezes difíceis de implementar, na medida em que é necessário uma replicação dos locais amostrados, o que muitas vezes requer um compromisso entre o número de pontos de amostragem distribuídos em toda a área de interesse/inferência (replicação espacial), e o número de replicas de amostragem em cada local (replicação temporal). Para compensar este potencial problema, uma abordagem comum é a realização de várias amostragens independentes numa única visita, usando um ou múltiplos observadores, ou então proceder-se à amostragem de várias parcelas inseridas num local maior, numa única visita (replicação espacial). O rato-de-água (Arvicola sapidus Miller 1908) é um pequeno mamífero restrito à Península Ibérica e França, sendo classificado como "Vulnerável" pela União Internacional para a Conservação da Natureza (IUCN). As populações desta espécie encontram-se em declínio em vários locais da sua distribuição, principalmente devido à perda e fragmentação de habitats resultantes das atividades humanas. Tipicamente, os habitats adequados ao rato-de-água estão associados às margens estáveis de cursos de água, com solo lamacento dominado por herbáceas ou vegetação arbustiva. À semelhança do seu congénere A. amphibious, o rato-de-água normalmente forma colónias discretas, facilmente reconhecíveis, sendo muitas vezes vistas como subunidades destintas de uma população maior ou metapopulação. Assim, o rato-de-Água apresenta-se como uma espécie ideal para uma abordagem metapopulacional, a qual poderá gerar informações úteis relativas à avaliação da dinâmica (meta)populacional desta espécie. Geralmente, o rato-de-Água forma pequenos trilhos onde os indivíduos costumam circular e onde produzem latrinas com típicos dejetos cilíndricos, muitas vezes usadas como indícios indiretos da presença da espécie. Desta forma, a presença destas latrinas é geralmente utilizada para realizar amostragens em grande escala, no sentido de se inferir acerca do estado das populações de rato-de-Água, incluindo estudos que avaliam os fatores que afetam os seus padrões de ocupação em diferentes sistemas ecológicos, como zonas montanhosas, zonas agrícolas, ou sistemas de lagoas temporárias. No entanto, nenhum destes estudos teve em conta as deteções imperfeitas, o que pode limitar a força das inferências sobre as variações de ocupação, e, assim, a eficácia dos esforços de conservação recomendadas para a espécie. Dada a necessidade de avaliar as tendências das populações de ratos-de-Água, bem como a falta de estudos de modelação das probabilidades de ocupação da espécie, tendo em conta deteções imperfeitas, este estudo tem como objetivo facultar mais pistas sobre os fatores que afetam os padrões e dinâmica de ocupação desta espécie em paisagens agrícolas Mediterrânicas, testando quais as co-variáveis ambientais e relacionados com as amostragens (esforço de amostragem, experiência do observador, chuva) podem eventualmente produzir variações na probabilidade de deteção da espécie. Especificamente, os habitats adequados para a espécie foram identificados e mapeados dentro de uma área de cerca de 247,6 ha, e em seguida foi amostrada a presença da espécie em cada habitat. Esta amostragem foi feita com base em típicos da presença da espécie, usando dois observadores, e empregando diferentes esforços de amostragem. Este procedimento foi repetido em duas sessões de amostragem, que correspondem ao início da época de reprodução (novembro-dezembro) e à época em que as populações estão presumivelmente perto de seu pico de abundância (fevereiro-março). Em seguida, foram desenvolvidos modelos de ocupação espacialmente explícitos para estimar as probabilidades de deteção e ocupação sazonais dos ratos-de-Água em relação às características dos habitats. De acordo com as previsões da teoria da metapopulação, os resultados obtidos mostram que a ocupação dos habitats pelos ratos-de-água está positivamente relacionada com a área do habitat e negativamente relacionada com o isolamento dos habitats. Além disso, e de acordo com as preferências da espécie em relação ao habitat, a presença de água foi significativa para a probabilidade de ocupação, contudo, as variáveis testadas para a vegetação (% coberto vegetal e altura do coberto) não mostraram suporte no ranking dos modelos. Por outro lado, esperava-se que a probabilidade de deteção variasse, não apenas com as características da vegetação presente nos habitats, mas também com a experiência do observador na condução das amostragens aos sinais típicos da presença da espécie, com o esforço de amostragem empregue, com o tempo decorrido desde o início do estudo, e com as condições meteorológicas durante as amostragens. Contudo, nenhuma destas variáveis mostrou qualquer suporte aquando do ranking dos vários modelos testados. No geral, espera-se que os resultados obtidos proporcionem uma melhor compreensão no que diz respeito às respostas das populações de ratos-de-água aos atuais usos do solo em paisagens Mediterrânicas fragmentadas. É ainda importante discutir os possíveis efeitos de probabilidades de deteção heterogéneas na força das inferências que podem ser feitas, bem como as suas implicações em programas de monitorização que visem a conservação da espécie a grandes escalas (de paisagem e nível regional).
In Europe, agricultural landscapes may play an extremely important role to biodiversity conservation, as many species depend directly on traditional farming systems. However, the diversity of traditional agricultural landscapes is decreasing dramatically, either due to land abandonment or crop production intensification. Understanding the effects of agricultural intensification on farmland biodiversity has thus become a main goal in both wildlife ecology and conservation. In this context, the metapopulation concept has provided a useful framework to understand how habitat loss and fragmentation may affect the persistence of species that is spatially structured. The southern water vole (Arvicola sapidus) usually forms discrete and easily recognizable breeding colonies, which are often seen as subunits of a larger patchy population or metapopulation, and thus is an ideal species to examine whether a metapopulation approach can yield useful insights for assessing its population dynamics. In this study we addressed this issue, analyzing the factors shaping the occupancy patterns and dynamics of this species in Mediterranean farmland, based on presence-sign searches in two seasons, and taking into account for imperfect detection. A spatially explicit modeling approach was used to test the following predictions: detectability of southern water vole presence signs should be affected by the sampling effort employed during the surveys, observer experience, local vegetation characteristics, and weather conditions; while occupancy should be mostly affected by patch size, isolation, presence of water, and vegetation variables. Results indicated that detectability was relatively high (≈71-81%), though contrary to the initial predictions, there was no support for none of the effects tested. Occupancy probabilities were relatively low (≈29-31%), and according to predictions from metapopulation theory and the species habitat preferences, was positively related to the patch area and presence of water, and negatively related to patch isolation, while vegetation variables tested showed no significant effects. Overall, this study provided important insights regarding the responses of southern water vole patchy populations to current land-uses in fragmented Mediterranean farmland; and the possible effects of imperfect detection on the strength of inferences made, and its implications for large-scale (landscape- and regional-level) monitoring programs targeting the conservation of the species.

碩士
國立成功大學
電腦與通信工程研究所
95
Optical code-division multiple-access (OCDMA) techniques not only provide a flexible and secure transmission but also allow multiple users access the network asynchronously and simultaneously. Therefore, they are the most suitable solution in local area network (LAN). In early one-dimensional (1-D) coding systems, the code length should be increased rapidly in order to accommodate large active users. Hence, the nonlinear effect will come into existence and the systems are hard to realize. Two-dimensional (2-D) coding systems have been introduced to improve the defect and they also maintain good correlation properties. In this thesis, we propose a new 2-D code family for OCDMA system called folded optical orthogonal codes (FOOCs). We fold the original 1-D optical orthogonal codes (OOCs) into the matrix form to produce FOOCs. FOOCs use two dimensions, i.e. time and wavelength, for transmission. According to our analysis, FOOCs have good correlation properties and can supply large number of codes. In the OCDMA system using FOOCs, we propose a novel and simple method to detect the active users. A special wavelength called pilot wavelength is added into the original transmission patterns of every user. At the receiver, therefore, we can acquire the number of the active users from the power of the pilot wavelength. Consequently, if we adjust the threshold value to the optimum threshold value according to the active users, we can decrease the bit error rate (BER) and improve the system performance.

碩士
國立清華大學
電機工程學系
96
This thesis concerns the quantum detection problem for ternary pure state signals. We first review the relevant study on the quantum signal detection problem in the manner determined in Eldar's paper. We verify her result with the ternary geometrically uniform (GU) pure state signal having the uniform prior probability distribution, in which we have used the matrix-form expression of the optimum detection operators. Next, we consider the optimum detection problem for the ternary GU pure state signal under the condition that one prior probability is given and the others are unknown. In this situation, we formulate our optimization problem and derive the necessary and sufficient conditions for the solution of the problem. To seek the closed-form expression for the optimum detection operators, we perform the numerical simulation. As a result, we make a conjecture on the closed-form expression of the optimum measurement operators and on the optimum prior distribution for our problem.

碩士
亞洲大學
光電與通訊學系
103
The block transmission scheme is a widely adopted technique in modern wireless communication systems. From the perspective of theoretical results and hardware implementations, a communication system can achieve an improved performance when the block-by-block transmission scheme is used and the channel state information is utilized efficiently in its transceiver design. For a block transmission system using zero-forcing equalization, this thesis investigates the performance of a jointly optimized linear transceiver design with block delay detection. The algorithm of determining minimum BER block delay is derived and built under the transmitted power constraint. Finally, computer simulations are conducted for validating the theoretical derivations and justifying the superior BER performance of the proposed designs in this thesis.

碩士
國立臺北科技大學
電子工程系研究所
105
The thesis proposes a buck converter and a boost converter. The proposed buck converter is a four-phase buck converter with optimum-damping controlled and dynamic-slope compensation techniques. The proposed four-phase buck converter has been fabricated in TSMC 0.18 μm CMOS 1P6M processes. The main topology of the proposed four-phase buck converter is based on the current-mode control with the optimum-damping controlled and dynamic-slope compensation techniques, so that the proposed four-phase buck converter can provide wide output voltage range from 0.5V to 1.8V. The proposed buck converter uses four-phase structure to raise output current up to 1500 mA. The experimental results show that transient response time is about 3 μs and 4 μs, when the load current is changed from 100 to 1500 mA and from 1500 to 100 mA, respectively. The peak power efficiency is 90.14% at 1000 mA load current. The second converter is a boost converter with adaptive on-time (AOT) controlled and zero-current-detection techniques. The proposed boost converter has been fabricated in TSMC 0.18 μm CMOS 1P6M processes. The main topology of boost converter is based on dual-loop control with the zero-current-detection technique, so that the efficiency of the converter can be increased by 11% at the light load. The input voltage range is from 0.5V to 1V, the output voltage is 1.8V. The experimental results show that transient response time is about 2 μs and 3 μs, when the load current is changed from 5 to 300 mA and from 300 to 5 mA, respectively. The peak power efficiency is 91.6% at 200 mA load current.

碩士
國立中央大學
機械工程學系
101
This study investigates the effect of machining defects and the optimum mounting on a primary mirror of a remote sensing instrument. Processing defects discussed here are residual stress and subsurface damage. The specimens are ground by diamond tools with different grain-sized, and the effects of grain size of the grinding tool on the residual stress and subsurface damage are investigated. Generally, optical simulation process does not consider the degraded optical performance caused by the deformations of mounting and mirror. This research established an opto-mechanical analysis process to integrate finite element analysis and surface fitting. After fitting, the data of the deformed primary mirror can be imported into optical simulation software to analyze its real optical performance. Based on the integrated opto-mechanical analysis, the best position to glue the kinematic mount and the primary mirror is determined. Furthermore, Taguchi method is adopted to determine the design parameters of the kinematic mount for the primary mirror which will yield a best optical performance.

The atmospheric lightwave propagation is considerably influenced by the random variations in the refractive index of air pockets due to turbulence. This undesired effect significantly degrades the performance of free-space optical (FSO) communication systems. Interestingly, the severity of such random degradations is highly related to the range of atmospheric propagation. In this thesis, we introduce relay-assisted FSO communications as a very promising technique to combat the degradation effects of atmospheric turbulence. Considering different configurations of the relays, we quantify the outage behavior of the relay-assisted system and identify the optimum relaying scheme. We further optimize the performance of the relay-assisted FSO system subject to some power constraints and provide optimal power control strategies for different scenarios under consideration. Moreover, an application of FSO relaying technique in quantum communications is investigated. The results demonstrate impressive performance improvements for the proposed relay-assisted FSO systems with respect to the conventional direct transmission whether applied in a classical or a quantum communication channel.