Advanced digital signal processing systems require specialized high-performance embedded computer architectures. The term high-performance translates to large amounts of data and computations per time unit. The term embedded further implies requirements on physical size and power efficiency. Thus the requirements are of both functional and non-functional nature. This thesis addresses the development of high-performance digital signal processing systems relying on manycore technology. We propose building two-level hierarchical computer architectures for this domain of applications. Further, we outline a tool flow based on methods and analysis techniques for automated, multi-objective mapping of such applications on distributed memory manycore processors. In particular, the focus is put on how to provide a means for tunable strategies for mapping of task graphs on array structured distributed memory manycores, with respect to given application constraints. We argue for code mapping strategies based on predicted execution performance, which can be used in an auto-tuning feedback loop or to guide manual tuning directed by the programmer. Automated parallelization, optimisation and mapping to a manycore processor benefits from the use of a concurrent programming model as the starting point. Such a model allows the programmer to express different types and granularities of parallelism as well as computation characteristics of importance in the addressed class of applications. The programming model should also abstract away machine dependent hardware details. The analytical study of WCDMA baseband processing in radio base stations, presented in this thesis, suggests dataflow models as a good match to the characteristics of the application and as execution model abstracting computations on a manycore. Construction of portable tools further requires a manycore machine model and an intermediate representation. The models are needed in order to decouple algorithms, used to transform and map application software, from hardware. We propose a manycore machine model that captures common hardware resources, as well as resource dependent performance metrics for parallel computation and communication. Further, we have developed a multifunctional intermediate representation, which can be used as source for code generation and for dynamic execution analysis. Finally, we demonstrate how we can dynamically analyse execution using abstract interpretation on the intermediate representation. It is shown that the performance predictions can be used to accurately rank different mappings by best throughput or shortest end-to-end computation latency.

Tracheal respiratory sounds analysis has been investigated as a non-invasive method to estimate respiratory flow and upper airway obstruction. However, the flow-sound relationship is highly variable among subjects which makes it challenging to estimate flow in general applications. Therefore, a robust model for acoustical flow estimation in a large group of individuals did not exist before. On the other hand, a major application of acoustical flow estimation is to detect flow limitations in patients with obstructive sleep apnea (OSA) during sleep. However, previously the flow--sound relationship was only investigated during wakefulness among healthy individuals. Therefore, it was necessary to examine the flow-sound relationship during sleep in OSA patients. This thesis takes the above challenges and offers innovative solutions. First, a modified linear flow-sound model was proposed to estimate respiratory flow from tracheal sounds. To remove the individual based calibration process, the statistical correlation between the model parameters and anthropometric features of 93 healthy volunteers was investigated. The results show that gender, height and smoking are the most significant factors that affect the model parameters. Hence, a general acoustical flow estimation model was proposed for people with similar height and gender. Second, flow-sound relationship during sleep and wakefulness was studied among 13 OSA patients. The results show that during sleep and wakefulness, flow-sound relationship follows a power law, but with different parameters. Therefore, for acoustical flow estimation during sleep, the model parameters should be extracted from sleep data to have small errors. The results confirm reliability of the acoustical flow estimation for investigating flow variations during both sleep and wakefulness. Finally, a new method for sleep apnea detection and monitoring was developed, which only requires recording the tracheal sounds and the blood's oxygen saturation level (SaO2) data. It automatically classifies the sound segments into breath, snore and noise. A weighted average of features extracted from sound segments and SaO2 signal was used to detect apnea and hypopnea events. The performance of the proposed approach was evaluated on the data of 66 patients. The results show high correlation (0.96,p < 0.0001) between the outcomes of our system and those of the polysomnography. Also, sensitivity and specificity of the proposed method in differentiating simple snorers from OSA patients were found to be more than 91%. These results are superior or comparable with the existing commercialized sleep apnea portable monitors.

This research consists of probabilistic modeling of speech audio signals and deep-brain neurological signals in brain-computer interfaces. A significant portion of this research consists of a collaborative effort with Neural Signals Inc., Duluth, GA, and Boston University to develop an intracortical neural prosthetic system for speech restoration in a human subject living with Locked-In Syndrome, i.e., he is paralyzed and unable to speak. The work is carried out in three major phases. We first use kernel-based classifiers to detect evidence of articulation gestures and phonological attributes speech audio signals. We demonstrate that articulatory information can be used to decode speech content in speech audio signals. In the second phase of the research, we use neurological signals collected from a human subject with Locked-In Syndrome to predict intended speech content. The neural data were collected with a microwire electrode surgically implanted in speech motor cortex of the subject's brain, with the implant location chosen to capture extracellular electric potentials related to speech motor activity. The data include extracellular traces, and firing occurrence times for neural clusters in the vicinity of the electrode identified by an expert. We compute continuous firing rate estimates for the ensemble of neural clusters using several rate estimation methods and apply statistical classifiers to the rate estimates to predict intended speech content. We use Gaussian mixture models to classify short frames of data into 5 vowel classes and to discriminate intended speech activity in the data from non-speech. We then perform a series of data collection experiments with the subject designed to test explicitly for several speech articulation gestures, and decode the data offline. Finally, in the third phase of the research we develop an original probabilistic method for the task of spike-sorting in intracortical brain-computer interfaces, i.e., identifying and distinguishing action potential waveforms in extracellular traces. Our method uses both action potential waveforms and their occurrence times to cluster the data. We apply the method to semi-artificial data and partially labeled real data. We then classify neural spike waveforms, modeled with single multivariate Gaussians, using the method of minimum classification error for parameter estimation. Finally, we apply our joint waveforms and occurrence times spike-sorting method to neurological data in the context of a neural prosthesis for speech.

Extracting the Electric Network Frequency (ENF) fluctuations from an audio recording and comparing it to a reference database is a new technology intended to perform forensic digital audio authentication. The objective of this thesis is to implement and design a range of programs and algorithms for capturing and extracting ENF signals. The developed C-program combined with a probe can be used to build the reference database. Our implementation of the Short-Time Fourier Transform method is intended for the ENF extraction of longer signals while our novel proposed use of the Autoregressive parametric method and our implementation of the zero-crossing approach tackle the case of shorter recordings. A Graphical User Interface (GUI) was developed to facilitate the process of extracting the ENF fluctuations. The whole process is tested and evaluated for various scenarios ranging from long to short recordings.

Tracheal respiratory sounds analysis has been investigated as a non-invasive method to estimate respiratory flow and upper airway obstruction. However, the flow-sound relationship is highly variable among subjects which makes it challenging to estimate flow in general applications. Therefore, a robust model for acoustical flow estimation in a large group of individuals did not exist before. On the other hand, a major application of acoustical flow estimation is to detect flow limitations in patients with obstructive sleep apnea (OSA) during sleep. However, previously the flow--sound relationship was only investigated during wakefulness among healthy individuals. Therefore, it was necessary to examine the flow-sound relationship during sleep in OSA patients. This thesis takes the above challenges and offers innovative solutions. First, a modified linear flow-sound model was proposed to estimate respiratory flow from tracheal sounds. To remove the individual based calibration process, the statistical correlation between the model parameters and anthropometric features of 93 healthy volunteers was investigated. The results show that gender, height and smoking are the most significant factors that affect the model parameters. Hence, a general acoustical flow estimation model was proposed for people with similar height and gender. Second, flow-sound relationship during sleep and wakefulness was studied among 13 OSA patients. The results show that during sleep and wakefulness, flow-sound relationship follows a power law, but with different parameters. Therefore, for acoustical flow estimation during sleep, the model parameters should be extracted from sleep data to have small errors. The results confirm reliability of the acoustical flow estimation for investigating flow variations during both sleep and wakefulness. Finally, a new method for sleep apnea detection and monitoring was developed, which only requires recording the tracheal sounds and the blood's oxygen saturation level (SaO2) data. It automatically classifies the sound segments into breath, snore and noise. A weighted average of features extracted from sound segments and SaO2 signal was used to detect apnea and hypopnea events. The performance of the proposed approach was evaluated on the data of 66 patients. The results show high correlation (0.96,p < 0.0001) between the outcomes of our system and those of the polysomnography. Also, sensitivity and specificity of the proposed method in differentiating simple snorers from OSA patients were found to be more than 91%. These results are superior or comparable with the existing commercialized sleep apnea portable monitors.

På bullriga arbetsplatser använder personal ofta hörselskydd med inbyggda högtalare för att lyssna på exempelvis musik i underhållningssyfte. Om användaren lyssnar på höga ljudnivåer under långa perioder kan bullerskador uppstå i dennes öron. Enligt lagstiftning måste nivån därför begränsas i förebyggande syfte. Bullernivån är ett genomsnitt av de ljudnivåer användaren exponerats för under en arbetsdag. Användaren måste vila öronen om gränsvärdet för bullernivån nås.Om man utnyttjar att det är ett genomsnitt kan användaren tillåtas lyssna på en hög ljudnivå under en begränsad tid för att sedan sänka den. Det går att bevara både säkerheten och lyssningsupplevelsen om en sänkning införs långsamt. Detta arbete beskriver hur en algoritm till en digital signalprocessor kan konstrueras för att reglera ljudnivån.Målsättningen var att algoritmen skulle skydda användarens hörsel utan att försämra lyssningsupplevelsen, och utan att förbruka mer energi än nödvändigt. I algoritmen ingick en prediktor som predikterar mängden buller användaren riskerar att utsättas för, om denne fortsätter lyssna på samma nivå.Långsamma sänkningar av ljudnivån kan då utföras i tid innan gränsvärdet nås. Det visade sig att algoritmen endast behövde ett fåtal samplingar per sekund för att skatta och reglera ljudnivån tillräckligt precist, vilket reducerade energiförbrukningen.Resultatet visar möjligheten att kombinera målen för säkerhet, lyssningsupplevelse och energieffektivitet i hörselskydd. Algoritmen implementerades inte på ett skarpt system.Den hade enbart tillgång till ljudsignalen användaren ämnade lyssna på i underhållningssyfte. / In noisy workplaces the staff are often using hearing protectors with built-in speakers for entertainment purposes. Prolonged exposure to loud sound levels can cause damage to the user’s ears. The legislation requires therefore a limiting mechanism for the speakers. The noise level is defined as the average of the sound levels the user has been exposed to during a working day. If the noise threshold is reached the user has to rest his ears. This definition can be exploited to allow the user to listen to a loud sound level for a limited time and then lowering it. If the sound level is lowered slowly, it is possible to preserve both safety and listening experience. This work describes how an algorithm can be designed for a digital signal processor with the purpose of controlling the sound level. The aim was to protect the user's hearing without spoiling the listening experience, and without consuming more power than necessary. The algorithm design included a predictor that predicts the amount of noise the user risk being subjected to, if he continues to listen at the same level. Slow reduction of the sound level can then be carried out in time before the noise threshold is reached. It turned out that the algorithm only needed a few samples per second to estimate and control the sound level sufficiently precisely, this reduced the power consumption. The results show that it is possible to combine the objectives for safety, listening experience and power consumption in hearing protectors. The algorithm was not implemented in a real system. The algorithm had only access to the audio signal which the user intended to listen to for entertainment purposes.

The efficient high-throughput VLSI implementation of Soft-output MIMO detectors for high-order constellations and large antenna configurations has been a major challenge in the literature. This thesis introduces a novel Soft-output K-Best scheme that improves BER performance and reduces the computational complexity significantly by using three major improvement ideas. It also presents an area and power efficient VLSI implementation of a 4x4 64-QAM Soft K-Best MIMO detector that attains the highest detection throughput of 2 Gbps and second lowest energy/bit reported in the literature, fulfilling the aggressive requirements of emerging 4G standards such as IEEE 802.16m and LTE-Advanced. A low-complexity and highly parallel algorithm for QR Decomposition, an essential channel pre-processing task, is also developed that uses 2D, Householder 3D and 4D Givens Rotations. Test results for the QRD chip, fabricated in 0.13um CMOS, show that it attains the lowest reported latency of 144ns and highest QR Processing Efficiency.

The efficient high-throughput VLSI implementation of Soft-output MIMO detectors for high-order constellations and large antenna configurations has been a major challenge in the literature. This thesis introduces a novel Soft-output K-Best scheme that improves BER performance and reduces the computational complexity significantly by using three major improvement ideas. It also presents an area and power efficient VLSI implementation of a 4x4 64-QAM Soft K-Best MIMO detector that attains the highest detection throughput of 2 Gbps and second lowest energy/bit reported in the literature, fulfilling the aggressive requirements of emerging 4G standards such as IEEE 802.16m and LTE-Advanced. A low-complexity and highly parallel algorithm for QR Decomposition, an essential channel pre-processing task, is also developed that uses 2D, Householder 3D and 4D Givens Rotations. Test results for the QRD chip, fabricated in 0.13um CMOS, show that it attains the lowest reported latency of 144ns and highest QR Processing Efficiency.

Σκοπός της παρούσας διπλωματικής εργασίας είναι η κατασκευή ενός συστήματος που να μπορεί να ξεχωρίζει τα οχήματα από άλλα αντικείμενα με τη χρήση κυματιδίου Haar και φίλτρου Gabor (εξαγωγή χαρακτηριστικών) και SVM, RBF για ταξινόμηση. / The aim of this thesis is the construction of a system that will be able to distiguish vehicles from other objects using Haar and Gabor filter (export characteristic) and SVM, RBF for classification.

Σκοπός της παρούσας διπλωματικής εργασίας είναι η δημιουργία συστήματος που θα διευρύνει τις δυνατότητες των εξ’ αποστάσεως εργαστηρίων προσανατολισμένα σε θέματα ψηφιακής επεξεργασίας σήματος και εικόνας. Η υλοποίηση πραγματοποιήθηκε με το LabVIEW v 8.6 και ονομάστηκε R-DSP Server. Αξιοποιώντας τις δυνατότητες που προσφέρει ο R-DSP Server οι χρήστες μπορούν να αναπτύξουν τα δικά τους γραφικά περιβάλλοντα (Graphical User Interfaces -GUIs) τα οποία ονομάζονται προσαρμοζόμενα γραφικά περιβάλλοντα (Customized GUIs,) για τον απομακρυσμένο έλεγχο DSP εφαρμογών. Για την εύκολη και γρήγορη ανάπτυξη τέτοιων γραφικών εφαρμογών στο περιβάλλων του LabVIEW, αναπτύχθηκε μια σειρά εργαλείων που ονομάστηκε R-DSP LabVIEW Toolkit. Η εργασία ολοκληρώνεται με την παρουσίαση της λειτουργιάς του R-DSP Server αλλά και της χρήσης του R-DSP Toolkit. / The purpose of this work is to present an approach which could expand the features of Remote Laboratories focused on embedded Digital Signal Processing (DSP) systems. The proposed approach is based on a system which is designed and developed with LabVIEW and is called R-DSP Server. Exploiting this system, users are able to develop their own Graphical User Interfaces (GUIs), named Customized GUIs, for the remote control and validation of real-time DSP applications. These GUIs are tailored to the needs of each DSP application and can be implemented in any programming language. The rapid design of Customized GUIs using LabVIEW for the communication with the R-DSP Server is achieved using an implemented set of functions, called R-DSP LabVIEW Toolkit.