Dissertations / Theses on the topic 'Sensor decoding'

Motivated by potential applications in wireless sensor networks, this thesis considers the problem of communicating a large number of correlated analog sources over a Gaussian multiple-access channel using non-orthogonal code-division multiple-access (CDMA). A joint source-channel decoder is presented which can exploit the inter-source correlation for interference reduction in the CDMA channel. This decoder uses a linear minimum mean square error (MMSE) multi-user detector (MUD) in tandem with a MMSE joint source decoder (JSD) for multiple sources to achieve a computational complexity that scales with the number of sources. The MUD and the JSD, then iteratively exchange extrinsic information to improve the interference cancellation. Experimental results show that, compared to a non-iterative decoder, the proposed iterative decoder is more robust against potential performance degradation due to correlated channel interference and offers better near far resistance.

This thesis investigates a graph and information theoretic approach to design and analysis of low-density parity-check (LDPC) codes and wireless networks. In this work, both LDPC codes and wireless networks are considered as random graphs. This work proposes solutions to important theoretic and practical open problems in LDPC coding, and for the first time introduces a framework for analysis of finite wireless networks. LDPC codes are considered to be one of the best classes of error-correcting codes. In this thesis, several problems in this area are studied. First, an improved decoding algorithm for LDPC codes is introduced. Compared to the standard iterative decoding, the proposed decoding algorithm can result in several orders of magnitude lower bit error rates, while having almost the same complexity. Second, this work presents a variety of bounds on the achievable performance of different LDPC coding scenarios. Third, it studies rate-compatible LDPC codes and provides fundamental properties of these codes. It also shows guidelines for optimal design of rate-compatible codes. Finally, it studies non-uniform and unequal error protection using LDPC codes and explores their applications to data storage systems and communication networks. It presents a new error-control scheme for volume holographic memory (VHM) systems and shows that the new method can increase the storage capacity by more than fifty percent compared to previous schemes. This work also investigates the application of random graphs to the design and analysis of wireless ad hoc and sensor networks. It introduces a framework for analysis of finite wireless networks. Such framework was lacking from the literature. Using the framework, different network properties such as capacity, connectivity, coverage, and routing and security algorithms are studied. Finally, connectivity properties of large-scale sensor networks are investigated. It is shown how unreliability of sensors, link failures, and non-uniform distribution of nodes affect the connectivity of sensor networks.

This thesis regards the Wireless Sensor Network (WSN), as one of the most important technologies for the twenty-first century and the implementation of different packet correcting erasure codes to cope with the ”bursty” nature of the transmission channel and the possibility of packet losses during the transmission. The limited battery capacity of each sensor node makes the minimization of the power consumption one of the primary concerns in WSN. Considering also the fact that in each sensor node the communication is considerably more expensive than computation, this motivates the core idea to invest computation within the network whenever possible to safe on communication costs. The goal of the research was to evaluate a parameter, for example the Packet Erasure Ratio (PER), that permit to verify the functionality and the behavior of the created network, validate the theoretical expectations and evaluate the convenience of introducing the recovery packet techniques using different types of packet erasure codes in different types of networks. Thus, considering all the constrains of energy consumption in WSN, the topic of this thesis is to try to minimize it by introducing encoding/decoding algorithms in the transmission chain in order to prevent the retransmission of the erased packets through the Packet Erasure Channel and save the energy used for each retransmitted packet. In this way it is possible extend the lifetime of entire network.

The nervous system encodes information about external stimuli through sophisticated computations performed by vast networks of sensory neurons. Since the space of all possible stimuli is much larger than the space of those that are ultimately meaningful, dimensionality reduction techniques were developed to identify the subspace of stimulus space relevant to neural activity. However, dimensionality reduction methods provide limited insight into the nonlinear functions that build the nervous system’s internal model of the world. In Chapter 2, the functional basis is introduced that transforms the relevant subspace to a basis that describes the computational function of the subunits that make up the neural circuitry. This functional basis is used to uncover novel insights about the computations performed by neurons in low-level vision and, later on, high-level auditory circuitry. For the latter, significant barriers are found in the capability of current dimensionality reduction methods to recover the relevant subspaces of high-level sensory neurons. This barrier is caused by the relative difficulty of stimulating high-level sensory neurons, which are often unresponsive to noise stimuli, while still maintaining a thorough exploration of the stimulus distribution. In response, a new approach to dimensionality reduction is formulated in Chapter 3 called the low-rank maximum noise entropy method that makes it possible to overcome challenges presented by high-level sensory systems. In Chapter 4, functional bases derived from the relevant subspaces recovered by the low-rank maximum noise entropy method are employed to study the neural computations performed by high-level auditory neurons.

The first part of this work concerns analog decoding. It presents the design of the I/O interface for a fully analog iterative decoder for a serially concatenated convolutional code and of a fully analog Trellis Coded Modulation (TCM) decoder for error correction in multi-level (ML) flash memories. The iterative decoder represents a significant step ahead in the evolution of analog decoders due to its reconfigurability in both block length and code rate. Moreover, with an efficiency of 2.1nJ/bit, it outperforms digital decoders with the same block length of a factor up to 50. The potential performance and limitations of the analog approach for a TCM decoder have been investigated considering a 4-state and an 8-state decoder, both developed in a 0.18um standard CMOS process. In the second part of the thesis, the design of a low-power transceiver chipset for ultra wideband impulse radio (UWB-IR) is presented, with particular emphasis on the transmitter design. In particular, the transmitter uses a novel combined mixer and power amplifier to generate a Gaussian pulse with 1.25GHz bandwith and center frequency of 7.875GHz. The combined MRX-PA includes a monolithic transformer to reach a maximum output voltage swing of 3.2Vpp, necessary to ensure the required link distance of 10 meters. The transformer has been designed in order to maximize the power efficiency and at the same time to realize a fourth-order ladder filter, so as to reduce the transmitter out-of band emissions. The efficiency of this design has been compared with state-of-the-art UWB-IR transmitters, showing how the proposed solution leads to an improvement in the transmitter efficiency of a factor of almost 10.
La prima parte di questo lavoro di tesi e' dedicata alla decodifica analogica e presenta la progettazione di un'interfaccia di I/O per un decodificatore iterattivo completamente analogico per un codice convoluzionale concatenato in serie e di un decoder analogico per Trellis Coded Modulation (TCM) per la correzione degli errori in memorie Flash multi-livello. Il decodificatore iterattivo rappresenta un grosso passo avanti nell'evoluzione dei decodificatori analogici in quanto e' possibile riconfigurarne sia la lunghezza di blocco che il rate del codice. Per di piu', con un'efficienza di 2.1nJ/bit, migliora fino a 50 volte le prestazioni in termini di efficienza dei decodificatori digitali con la stessa lunghezza di blocco. Le potenziali prestazioni e le limitazioni dell'approccio analogico per un decodificatore per TCM sono state investigate considerando due diversi decodificatori, uno a 4 stati ed uno ad 8 stati, entrambi sviluppati in un processo CMOS standard con una lunghezza di canale di 0.18um. Nella seconda parte della tesi viene presentato il design di un transciver per una radio ad impulsi a banda larga (UWB-IR), con particolare enfasi sulla progettazione del trasmettitore. Il trasmettitore utilizza una nuova combinazione di mixer e amplificatore di potenza per generare un impulso gaussiano con una larghezza di banda di 1.25GHz ed una frequenza centrale di 7.875GHz. Il nuovo circuito, inoltre, include un trasformatore monolitico in modo tale da generare una tensione di uscita di 3.2Vpp, necessaria per garantire la distanza di connessione richiesta di almeno 10 metri. Il trasformatore e' stato progettato in modo da massimizzare l'efficienza in termini di potenza e, allo stesso tempo, realizzare un filtro ladder del quarto ordine al fine di ridurre le emissioni fuori banda del trasmettitore stesso. Confrontando l'efficienza di questo design con trasmettori per UWB-IR allo stato dell'arte si e' visto come la soluzione da noi proposta porti ad un miglioramento dell'efficienza del trasmettitore di un fattore pari a 10.

Entendre l'origen i la funció de l'activitat de poblacions neuronals, i com aquesta activitat es relaciona amb els estímuls sensorials, les decisions o les accions motores és un gran repte per les neurociències. En aquest treball hem analitzat l'activitat de desenes de neurones enregistrades a l'escorça visual primària de micos mentre se'ls presentaven escletxes sinusoïdals en diferents orientacions. Hem trobat que les fluctuacions globals de la xarxa mesurades mitjançant l'activitat de la població modulen la selectivitat de les neurones de forma multiplicativa i additiva. A més, l'activitat de la població també afecta la informació present en grups petits de neurones, depenent de la modulació que ha provocat a la selectivitat d'aquestes. La informació de la població sencera, però, no canvia amb aquestes fluctuacions. A la segona part hem desenvolupat un mètode per mesurar 'correlacions diferencials' amb dades limitades. En aplicar-ho a les dades experimentals hem aconseguit la primera estimació preliminar de la grandària d'aquestes correlacions que limiten la informació. Els nostres resultats contribueixen a l'avenç en la comprensió de la codi ficació d'informació en poblacions neuronals, i alhora generen noves preguntes sobre com aquestes processen i transmeten informació.
Entender el origen y la función de la actividad de poblaciones neuronales, y cómo esta actividad se relaciona con los estímulos sensoriales, las decisiones o las acciones motoras es un gran desafio en neurociencia. En este trabajo hemos analizado la actividad de decenas de neuronas registradas en la corteza visual primaria de monos mientras rejillas sinusoidales en diferentes orientaciones eran presentadas. Hemos encontrado que las fluctuaciones globales de la red medidas mediante la actividad de la población modulan la selectividad de las neuronas de manera multiplicativa y aditiva. Además, la actividad de la población también afecta a la información presente en grupos pequeños de neuronas, dependiendo de la modulación que ha provocado en la selectividad de estas neuronas. La información en la población completa, sin embargo, no varía con estas fluctuaciones. En la segunda parte hemos desarrollado un método para medir 'correlaciones diferenciales' con datos limitados. Al aplicarlo a los datos experimentales hemos obtenido la primera estimación preliminar del tamaño de estas correlaciones que limitan la información. Nuestros resultados contribuyen al avance del entendimiento sobre la codi ficación de la información en poblaciones neuronales, y al mismo tiempo generan más preguntas sobre cómo éstas procesan y transmiten información.
Understanding the sources and the role of the spiking activity of neural populations, and how this activity is related to sensory stimuli, decisions or motor actions is a crucial challenge in neuroscience. In this work, we analyzed the spiking activity of tens of neurons recorded in the primary visual cortex of macaque monkeys while drifting sinusoidal gratings were presented in di erent orientations. We found that global uctuations of the network measured by the population activity a ect the tuning of individual neurons both multiplicatively and additively. Population activity also has an impact in the information of small ensembles, which depends on the kind of modulation that the tuning of those neurons undergoes. Interestingly, the total information of the network is not altered by these uctuations. In the second part, we developed a method to measure 'di erential correlations' from limited amount of data, and obtained the rst, although preliminary, estimate in experimental data. Our results have important implications for information coding, and they open new questions about the way information is processed and transmitted by the spiking activity of neural populations.

The use of autonomous robots in our society is increasing every day and a robot is no longer seen as a tool but as a team member. The robots are now working side by side with us and provide assistance during dangerous operations where humans otherwise are at risk. This development has in turn increased the need of robots with more human-awareness. Therefore, this master thesis aims at contributing to the enhancement of human-aware robotics. Specifically, we are investigating the possibilities of equipping autonomous robots with the capability of assessing and detecting activities in human teams. This capability could, for instance, be used in the robot's reasoning and planning components to create better plans that ultimately would result in improved human-robot teamwork performance. we propose to improve existing teamwork activity recognizers by adding intangible features, such as stress, motivation and focus, originating from human behavior models. Hidden markov models have earlier been proven very efficient for activity recognition and have therefore been utilized in this work as a method for classification of behaviors. In order for a robot to provide effective assistance to a human team it must not only consider spatio-temporal parameters for team members but also the psychological.To assess psychological parameters this master thesis suggests to use the body signals of team members. Body signals such as heart rate and skin conductance. Combined with the body signals we investigate the possibility of using System Dynamics models to interpret the current psychological states of the human team members, thus enhancing the human-awareness of a robot.
Användningen av autonoma robotar i vårt samhälle ökar varje dag och en robot ses inte längre som ett verktyg utan som en gruppmedlem. Robotarna arbetar nu sida vid sida med oss och ger oss stöd under farliga arbeten där människor annars är utsatta för risker. Denna utveckling har i sin tur ökat behovet av robotar med mer människo-medvetenhet. Därför är målet med detta examensarbete att bidra till en stärkt människo-medvetenhet hos robotar. Specifikt undersöker vi möjligheterna att utrusta autonoma robotar med förmågan att bedöma och upptäcka olika beteenden hos mänskliga lag. Denna förmåga skulle till exempel kunna användas i robotens resonemang och planering för att ta beslut och i sin tur förbättra samarbetet mellan människa och robot. Vi föreslår att förbättra befintliga aktivitetsidentifierare genom att tillföra förmågan att tolka immateriella beteenden hos människan, såsom stress, motivation och fokus. Att kunna urskilja lagaktiviteter inom ett mänskligt lag är grundläggande för en robot som ska vara till stöd för laget. Dolda markovmodeller har tidigare visat sig vara mycket effektiva för just aktivitetsidentifiering och har därför använts i detta arbete. För att en robot ska kunna ha möjlighet att ge ett effektivt stöd till ett mänskligtlag måste den inte bara ta hänsyn till rumsliga parametrar hos lagmedlemmarna utan även de psykologiska. För att tyda psykologiska parametrar hos människor förespråkar denna masteravhandling utnyttjandet av mänskliga kroppssignaler. Signaler så som hjärtfrekvens och hudkonduktans. Kombinerat med kroppenssignalerar påvisar vi möjligheten att använda systemdynamiksmodeller för att tolka immateriella beteenden, vilket i sin tur kan stärka människo-medvetenheten hos en robot.

The thesis work was conducted in Stockholm, Kista at the department of Informatics and Aero System at Swedish Defence Research Agency.

博士
國立臺灣科技大學
電機工程系
96
Distributed Classification Fusion using Error-Correcting Codes (DCFECC) has recently been proposed for wireless sensor networks. It adopts the Minimum Hamming Distance (MHD) fusion rule and performs much better than traditional classification approaches when the network has faulty sensors. Different fusion rules were proposed later. One of them is Distributed Classification fusion using Soft-decision Decoding (DCSD). The DCSD fusion rule has a considerably misclassification probability than the MHD fusion rule. However, the probability of misclassification using DCSD approach is high when the detection result is not reliable. Moreover, the transmission channel is highly noisy. Since the sensor makes its local decision without evaluating the reliability of the detection result, it may waste its power to transmit an unreliable local decision. In this work, the performance of the DCSD fusion rule is analyzed. Asymptotic performance approximation of the DCSD fusion rule is derived based on the Central Limit Theorem. Furthermore, an asymptotic upper bound on the misclassification probability is obtained. The numerical simulations are conducted to verify our analysis results. Besides, this work proposes an adaptive redetection scheme to resolve this problem of the unreliable detection results. An unreliable range is set for each sensor. If the detection result of the sensor is not located in the unreliable range, the sensor makes a local decision. Otherwise, the sensor has to make another detection. This work further proposes an adaptive retransmission scheme to reduce the misclassification probability in highly noisy channels. When a final decision made by a fusion center is unreliable, the sensor which has sent the local decision with the lowest channel reliability will be asked to retransmit its local decision by the fusion center. Performance analysis and simulation results show that the misclassification probability can be efficiently reduced through the adaptive redetection and retransmission.

This thesis investigates a graph and information theoretic approach to design and analysis of low-density parity-check (LDPC) codes and wireless networks. In this work, both LDPC codes and wireless networks are considered as random graphs. This work proposes solutions to important theoretic and practical open problems in LDPC coding, and for the first time introduces a framework for analysis of finite wireless networks. LDPC codes are considered to be one of the best classes of error-correcting codes. In this thesis, several problems in this area are studied. First, an improved decoding algorithm for LDPC codes is introduced. Compared to the standard iterative decoding, the proposed decoding algorithm can result in several orders of magnitude lower bit error rates, while having almost the same complexity. Second, this work presents a variety of bounds on the achievable performance of different LDPC coding scenarios. Third, it studies rate-compatible LDPC codes and provides fundamental properties of these codes. It also shows guidelines for optimal design of rate-compatible codes. Finally, it studies non-uniform and unequal error protection using LDPC codes and explores their applications to data storage systems and communication networks. It presents a new error-control scheme for volume holographic memory (VHM) systems and shows that the new method can increase the storage capacity by more than fifty percent compared to previous schemes. This work also investigates the application of random graphs to the design and analysis of wireless ad hoc and sensor networks. It introduces a framework for analysis of finite wireless networks. Such framework was lacking from the literature. Using the framework, different network properties such as capacity, connectivity, coverage, and routing and security algorithms are studied. Finally, connectivity properties of large-scale sensor networks are investigated. It is shown how unreliability of sensors, link failures, and non-uniform distribution of nodes affect the connectivity of sensor networks.

Utilizzo dei sensori TreeTalker per la decodifica di variabili ambientali e conseguenti risposte fisiologiche degli alberi. Creazione pacchetto R per pulizia database e controllo funzionalità dei sensori, calibrazione sensore di sapflow, monitoraggio forestale su un pendio. Using TreeTalker sensors for decoding environmental variables and consequent physiological responses of trees. Creating R package for database cleanup and sensor functionality check, sapflow sensor calibration, forest monitoring on a slope.

An important problem in neuroscience is to understand how the brain encodes information. A hypothesis is that differences in the timing of action potentials, reflecting synchronization changes among neuronal ensembles --often occurring in the context of oscillations-- can be meaningful to downstream neurons detecting coincident input. Several properties, such as active conductances, feedforward inhibition and oscillatory input, could potentially influence whether a neuron acts as a coincidence detector. Although different neural circuits in various animal groups will use different strategies to solve somewhat varying problems, there will also be many powerful solutions to coding problems that will be used repeatedly across diverse processing stages and animal phyla. The insect olfactory system, sharing many design similarities with other systems while having a reduced complexity, provides an excellent model in which to study the functional interactions of all these coding features.

This dissertation focuses on the decoding of olfactory information by the mushroom body (MB), the second relay of the insect olfactory system, which receives oscillating input from the antennal lobe (the first relay, analogous to the vertebrate olfactory bulb). Kenyon cells (KCs), the intrinsic neurons of the MB, are found to respond very specifically to odors. These responses typically consist of one or two reliable action potentials, phase-locked to the global oscillations, over extremely low baseline firing rates. This leads to a dramatic sparsening of the olfactory representation in the MB. Several circuit and intrinsic properties are found to take part in this transformation. Feedforward inhibition contributes to odor specificity and sparseness: blocking inhibitory input to the KCs broadened their odor tuning and abolished their phase-locking, supporting the idea that feedforward inhibition limits the temporal window over which KCs integrate their inputs. Voltage-dependent conductances contribute to a supralinear summation of coincident postsynaptic potentials and a reduction of their half-widths, indicating that KC intrinsic properties further contribute to coincidence detection. Taken together, these results indicate that oscillations serve as a framework on which KCs act as coincidence detectors and sparsen the olfactory representation. Abolishing the input oscillations disrupts KC odor responses, decreasing their specificity and the sparseness in the MB.

The work in this dissertation describes a mechanism for decoding timing information and indicates that not all spikes are equally relevant to downstream neurons, their specific relevance depending on whether they are correlated, within a specific phase of an oscillation cycle, with other input spikes. These general features can also provide useful insights into neural coding in more complex neural systems, where all the mechanisms described here have been separately observed. This work illustrates how these mechanisms can interact to code sensory information and bring about drastic transformations of neural representations, increasing our understanding of how nervous systems can process information.

In daily activities, humans manipulate objects and do so with great precision. Empirical studies have demonstrated that signals encoded by mechanoreceptors facilitate the precise object manipulation in humans, however, little is known about the underlying mechanisms. Models used in literature to analyze tactile afferent data range from advanced—for example some models account for skin tissue properties—to simple regression fit. These models, however, do not systematically account for factors that influence tactile afferent activity. For instance, it is not yet clear whether the first derivative of force influences the observed tactile afferent spike train patterns. In this study, I use the technique of microneurography—with the help of Dr. Birznieks—to record tactile afferent data from humans. I then implement spike sorting algorithms to identify spike occurrences that pertain to a single cell. For further analyses of the resulting spike trains, I use a Bayesian decoding framework to investigate tactile afferent mechanisms that are responsible for sensorimotor control in humans. The Bayesian decoding framework I implement is a two stage process where in a first stage (encoding model) the relationships between the administered stimuli and the recorded tactile afferent signals is established, and a second stage uses results based on the first stage to make predictions. The goal of encoding model is to increase our understanding of the mechanisms that underlie dexterous object manipulation and, from an engineering perspective, guide the design of algorithms for inferring stimulus from previously unseen tactile afferent data, a process referred to as decoding. Specifically, the objective of the study was to devise quantitative methods that would provide insight into some mechanisms that underlie touch, as well as provide strategies through which real-time biomedical devices can be realized. Tactile afferent data from eight subjects (18 - 30 years) with no known form of neurological disorders were recorded by inserting a needle electrode in the median nerve at the wrist. I was involved in designing experimental protocols, designing mechanisms that were put in place for safety measures, designing and building electronic components as needed, experimental setup, subject recruitment, and data acquisition. Dr. Ingvars Birznieks (performed the actual microneurography procedure by inserting a needle electrode into the nerve and identifying afferent types) and Dr. Heba Khamis provided assistance with the data acquisition and experimental design. The study took place at Neuroscience Research Australia (NeuRA). Once the data were acquired, I analyzed the data recorded from slowly adapting type I tactile afferents (SA-I). The initial stages of data analysis involved writing software routines to spike sort the data (identify action potential waveforms that pertain to individual cells). I analyzed SA-I tactile afferents because they were more numerous (it was difficult to target other types of afferents during experiments). In addition, SA-I tactile afferents respond during both the dynamic and the static phase of a force stimulus. Since they respond during both the dynamic and static phases of the force stimulus, it seemed reasonable to hypothesize that SA-I’s alone could provide sufficient information for predicting the force profile, given spike data. In the first stage, I used an inhomogeneous Poisson process encoding model through which I assessed the relative importance of aspects of the stimuli to observed spike data. In addition I estimated the likelihood for SA-I data given the inhomogeneous Poisson model, which was used during the second stage. The likelihood is formulated by deriving the joint distribution of the data, as a function of the model parameters with the data fixed. In the second stage, I used a recursive nonlinear Bayesian filter to reconstruct the force profile, given the SA-I spike patterns. Moreover, the decoding method implemented in this thesis is feasible for real-time applications such as interfacing with prostheses because it can be realized with readily available electronic components. I also implemented a renewal point process encoding model—as a generalization of the Poisson process encoding model—which can account for some history dependence properties of neural data. I discovered that under my encoding model, the relative contributions of the force and its derivative are 1.26 and 1.02, respectively. This suggests that the force derivative contributes significantly to the spiking behavior of SA-I tactile afferents. This is a novel contribution because it provides a quantitative result to the long standing question of whether the force derivative contributes towards SA-I tactile afferent spiking behavior. As a result, I incorporated the first derivative of force, along with the force, in the encoding models I implemented in this thesis. The decoding model shows that SA-I fibers provide sufficient information for an approximation of the force profile. Furthermore, including fast adapting tactile afferents would provide better information about the first moment of contact and last moment of contact, and thus improved decoding results. Finally I show that a renewal point process encoding model captures interspike time and stimulus features better than an inhomogeneous Poisson point process encoding model. This is useful because it is now possible to generate synthetic data with statistical structure that is similar to real SA-I data: This would enable further investigations of mechanisms that underlie SA-I tactile afferents.