The thesis presents a contribution to the analysis and visualization of computational performance based on event traces with a particular focus on parallel programs and High Performance Computing (HPC). Event traces contain detailed information about specified incidents (events) during run-time of programs and allow minute investigation of dynamic program behavior, various performance metrics, and possible causes of performance flaws. Due to long running and highly parallel programs and very fine detail resolutions, event traces can accumulate huge amounts of data which become a challenge for interactive as well as automatic analysis and visualization tools. The thesis proposes a method of exploiting redundancy in the event traces in order to reduce the memory requirements and the computational complexity of event trace analysis. The sources of redundancy are repeated segments of the original program, either through iterative or recursive algorithms or through SPMD-style parallel programs, which produce equal or similar repeated event sequences. The data reduction technique is based on the novel Complete Call Graph (CCG) data structure which allows domain specific data compression for event traces in a combination of lossless and lossy methods. All deviations due to lossy data compression can be controlled by constant bounds. The compression of the CCG data structure is incorporated in the construction process, such that at no point substantial uncompressed parts have to be stored. Experiments with real-world example traces reveal the potential for very high data compression. The results range from factors of 3 to 15 for small scale compression with minimum deviation of the data to factors > 100 for large scale compression with moderate deviation. Based on the CCG data structure, new algorithms for the most common evaluation and analysis methods for event traces are presented, which require no explicit decompression. By avoiding repeated evaluation of formerly redundant event sequences, the computational effort of the new algorithms can be reduced in the same extent as memory consumption. The thesis includes a comprehensive discussion of the state-of-the-art and related work, a detailed presentation of the design of the CCG data structure, an elaborate description of algorithms for construction, compression, and analysis of CCGs, and an extensive experimental validation of all components. / Diese Dissertation stellt einen neuartigen Ansatz für die Analyse und Visualisierung der Berechnungs-Performance vor, der auf dem Ereignis-Tracing basiert und insbesondere auf parallele Programme und das Hochleistungsrechnen (High Performance Computing, HPC) zugeschnitten ist. Ereignis-Traces (Ereignis-Spuren) enthalten detaillierte Informationen über spezifizierte Ereignisse während der Laufzeit eines Programms und erlauben eine sehr genaue Untersuchung des dynamischen Verhaltens, verschiedener Performance-Metriken und potentieller Performance-Probleme. Aufgrund lang laufender und hoch paralleler Anwendungen und dem hohen Detailgrad kann das Ereignis-Tracing sehr große Datenmengen produzieren. Diese stellen ihrerseits eine Herausforderung für interaktive und automatische Analyse- und Visualisierungswerkzeuge dar. Die vorliegende Arbeit präsentiert eine Methode, die Redundanzen in den Ereignis-Traces ausnutzt, um sowohl die Speicheranforderungen als auch die Laufzeitkomplexität der Trace-Analyse zu reduzieren. Die Ursachen für Redundanzen sind wiederholt ausgeführte Programmabschnitte, entweder durch iterative oder rekursive Algorithmen oder durch SPMD-Parallelisierung, die gleiche oder ähnliche Ereignis-Sequenzen erzeugen. Die Datenreduktion basiert auf der neuartigen Datenstruktur der "Vollständigen Aufruf-Graphen" (Complete Call Graph, CCG) und erlaubt eine Kombination von verlustfreier und verlustbehafteter Datenkompression. Dabei können konstante Grenzen für alle Abweichungen durch verlustbehaftete Kompression vorgegeben werden. Die Datenkompression ist in den Aufbau der Datenstruktur integriert, so dass keine umfangreichen unkomprimierten Teile vor der Kompression im Hauptspeicher gehalten werden müssen. Das enorme Kompressionsvermögen des neuen Ansatzes wird anhand einer Reihe von Beispielen aus realen Anwendungsszenarien nachgewiesen. Die dabei erzielten Resultate reichen von Kompressionsfaktoren von 3 bis 5 mit nur minimalen Abweichungen aufgrund der verlustbehafteten Kompression bis zu Faktoren > 100 für hochgradige Kompression. Basierend auf der CCG_Datenstruktur werden außerdem neue Auswertungs- und Analyseverfahren für Ereignis-Traces vorgestellt, die ohne explizite Dekompression auskommen. Damit kann die Laufzeitkomplexität der Analyse im selben Maß gesenkt werden wie der Hauptspeicherbedarf, indem komprimierte Ereignis-Sequenzen nicht mehrmals analysiert werden. Die vorliegende Dissertation enthält eine ausführliche Vorstellung des Stands der Technik und verwandter Arbeiten in diesem Bereich, eine detaillierte Herleitung der neu eingeführten Daten-strukturen, der Konstruktions-, Kompressions- und Analysealgorithmen sowie eine umfangreiche experimentelle Auswertung und Validierung aller Bestandteile.

In the past, we could rely on technology scaling and new micro-architectural techniques to improve the performance of processors. Nowadays, both of these methods are reaching their limits. The primary concern in future architectures with billions of transistors on a chip and limited power budgets is power/energy efficiency. Full-custom design of application-specific cores can yield up to two orders of magnitude better power efficiency over conventional general-purpose cores. However, a tremendous design effort is required in integrating a new accelerator for each new application. In this dissertation, we present the design of specialized compute fabrics that maintain the efficiency of full custom hardware while providing enough flexibility to execute a whole class of coarse-grain operations. The broad vision is to develop integrated and specialized hardware/software solutions that are co-optimized and co-designed across all layers ranging from the basic hardware foundations all the way to the application programming support through standard linear algebra libraries. We try to address these issues specifically in the context of dense linear algebra applications. In the process, we pursue the main questions that architects will face while designing such accelerators. How broad is this class of applications that the accelerator can support? What are the limiting factors that prevent utilization of these accelerators on the chip? What is the maximum achievable performance/efficiency? Answering these questions requires expertise and careful codesign of the algorithms and the architecture to select the best possible components, datapaths, and data movement patterns resulting in a more efficient hardware-software codesign. In some cases, codesign reduces complexities that are imposed on the algorithm side due to the initial limitations in the architectures. We design a specialized Linear Algebra Processor (LAP) architecture and discuss the details of mapping of matrix-matrix multiplication onto it. We further verify the flexibility of our design for computing a broad class of linear algebra kernels. We conclude that this architecture can perform a broad range of matrix-matrix operations as complex as matrix factorizations, and even Fast Fourier Transforms (FFTs), while maintaining its ASIC level efficiency. We present a power-performance model that compares state-of-the-art CPUs and GPUs with our design. Our power-performance model reveals sources of inefficiencies in CPUs and GPUs. We demonstrate how to overcome such inefficiencies in the process of designing our LAP. As we progress through this dissertation, we introduce modifications of the original matrix-matrix multiplication engine to facilitate the mapping of more complex operations. We observe the resulting performance and efficiencies on the modified engine using our power estimation methodology. When compared to other conventional architectures for linear algebra applications and FFT, our LAP is over an order of magnitude better in terms of power efficiency. Based on our estimations, up to 55 and 25 GFLOPS/W single- and double-precision efficiencies are achievable on a single chip in standard 45nm technology. / text

The increasing complexity of heterogeneous cellular networks (HetNets) due to the irregular deployment of small cells demands significant rethinking in the way cellular networks are perceived, modeled and analyzed. In addition to threatening the relevance of classical models, this new network paradigm also raises questions regarding the feasibility of state-of-the-art simulation-based approach for system design. This dissertation proposes a fundamentally new approach based on random spatial models that is not only tractable but also captures current deployment trends fairly accurately. First, this dissertation presents a general baseline model for HetNets consisting of K different types of base stations (BSs) that may differ in terms of transmit power, deployment density and target rate. Modeling the locations of each class of BSs as an independent Poisson Point Process (PPP) allows the derivation of surprisingly simple expressions for coverage probability and average rate. One interpretation of these results is that adding more BSs or tiers does not necessarily change the coverage probability, which indicates that fears of "interference overload" in HetNets are probably overblown. Second, a flexible notion of BS load is incorporated by introducing a new idea of conditionally thinning the interference field. For this generalized model, the coverage probability is shown to increase when lightly loaded small cells are added to the existing macrocellular networks. This is due to the fact that owing to the smaller loads, small cells typically transmit less often than macrocells, thus contributing less to the interference power. The same idea of conditional thinning is also shown to be useful in modeling the non-uniform user distributions, especially when the users lie closer to the BSs. Third, the baseline model is extended to study multi-antenna HetNets, where BSs across tiers may additionally differ in terms of the number of transmit antennas, number of users served and the multi-antenna transmission strategy. Using novel tools from stochastic orders, a tractable framework is developed to compare the performance of various multi-antenna transmission strategies for a fairly general spatial model, where the BSs may follow any general stationary distribution. The analysis shows that for a given total number of transmit antennas in the network, it is preferable to spread them across many single-antenna BSs vs. fewer multi-antenna BSs. Fourth, accounting for the load on the serving BS, downlink rate distribution is derived for a generalized cell selection model, where shadowing, following any general distribution, impacts cell selection while fading does not. This generalizes the baseline model and all its extensions, which either ignore the impact of channel randomness on cell selection or lumps all the sources of randomness into a single random variable. As an application of these results, it is shown that in certain regimes, shadowing naturally balances load across various tiers and hence reduces the need for artificial cell selection bias. Fifth and last, a slightly futuristic scenario of self-powered HetNets is considered, where each BS is powered solely by a self-contained energy harvesting module that may differ across tiers in terms of the energy harvesting rate and energy storage capacity. Since a BS may not always have sufficient energy, it may not always be available to serve users. This leads to a notion of availability region, which characterizes the fraction of time each type of BS can be made available under variety of strategies. One interpretation of this result is that the self-powered BSs do not suffer performance degradation due to the unreliability associated with energy harvesting if the availability vector corresponding to the optimal system performance lies in the availability region. / text

Στην απαρχή της καινούριας χιλιετίας, έχει γίνει πλέον ξεκάθαρος ο σπουδαίος ρόλος και η χρησιμότητα των ασυρμάτων τηλεπικοινωνιών και των εφαρμογών τους, η εξέλιξη των οποίων επικεντρώνεται σε μεγάλο βαθμό στις Ασύρματες Προσωπικές Τηλεπικοινωνίες (ΑΠΤ). Εξαιτίας της μεγάλης δημοτικότητας που έχουν τα συστήματα αυτά, υπάρχει συνεχής ανάγκη για αύξηση της χωρητικότητας και αναβάθμιση των παρεχόμενων υπηρεσιών ϕωνής και δεδομένων. Μια πολύ σημαντική περιοχή έρευνας για τις ΑΠΤ, που έχει προκύψει από τις παραπάνω απαιτήσεις, είναι η μελέτη και η αντιμετώπιση των προβλημάτων που δημιουργούνται στην επίδοση των δικτύων αυτών από διάφορα ϕαινόμενα που επιδρούν στο ασύρματο κανάλι. Σε αυτήν την ερευνητική περιοχή εστιάζεται και η παρούσα διδακτορική διατριβή, που έχει ως στόχο τη ϑεωρητική μελέτη της επίδοσης ασύρματων ψηφιακών τηλεπικοινωνιακών συστημάτων τα οποία υπόκεινται σε διαλείψεις. Για την καταπολέμηση των αρνητικών συνεπειών των διαλείψεων έχει υιοθετηθεί η τεχνική του διαφορισμού στο δέκτη. Σύμφωνα με αυτήν, ο δέκτης διαθέτει περισσότερες από μια κεραίες λήψης με αποτέλεσμα να λαμβάνει πολλαπλά αντίγραφα του εκπεμπόμενου σήματος, τα οποία συνδυάζει κατάλληλα ανάλογα με τις απαιτήσεις του διαχειριστή του δικτύου. Η διεθνής έρευνα για τη χρήση δεκτών διαφορισμού σε κανάλια Διαλείψεων Μικρής Κλίμακας (ΔΜΙΚ) είναι εκτενής και αφορά κυρίως κανάλια που βασίζονται στις κατανομές Rayleigh, Rice και Nakagamim. Δύο άλλες σημαντικές κατανομές για τη μοντελοποίηση των ΔΜΙΚ είναι η Weibull και, η γενίκευση της, η Γενικευμένη-Γάμα (ΓG), οι οποίες σε σύντομο χρονικό διάστημα έχουν κερδίσει εκτεταμένο επιστημονικό ενδιαφέρον στην περιοχή των ασύρματων κινητών τηλεπικοινωνιών. ΄Ομως, παρά το ενδιαφέρον αυτό, η μελέτη της επίδοσης δεκτών διαφορισμού σε κανάλια που μοντελοποιούνται με τις κατανομές αυτές, δεν έχει λάβει τις διαστάσεις που έχουν λάβει οι αντίστοιχες έρευνες για τις άλλες κατανομές. Αυτό οφείλεται σε μεγάλο βαθμό στη μεγάλη πολυπλοκότητα, από μαϑηματικής άποψης, που παρουσιάζει η στατιστική αναπαράσταση του σήματος της εξόδου πολλών τύπων δεκτών διαφορισμού που σε συνδυασμό με τη σχετικά σύνθετη μαθηματική έκφραση των κατανομών Weibull και ΓG, δημιουργεί δυσεπίλυτα προβλήματα. Στην παρούσα διατριβή, προσπερνώντας αυτές τις δυσκολίες, αρχικά μελετήθηκε η απόδοση μιας ειδικής κατηγορίας δεκτών διαφορισμού, του δέκτη Συνδυασμού Γενικευμένης Επιλογής (ΣΓΕ), ο οποίος λειτουργεί σε Ανεξάρτητα αλλά Μη-Ομοιόμορφα Κατανεμημένα (ΑΜΟΚ) κανάλια διαλείψεων Weibull. Σε αυτήν τη μελέτη εξετάζεται η περίπτωση όπου επιλέγονται οι δύο κλάδοι με το υψηλότερο λόγο σήματος προς ϑόρυβο, Signal to Noise Ratio (SNR), από τους L διαθέσιμους, ΣΓΕ(2, L). Κατά συνέπεια, με την παρουσίαση μιας καινούριας ιδιότητας για το γινόμενο των ροπών ταξινομημένων Τυχαίων Μεταβλητών (ΤΜ) Weibull, γίνεται δυνατή η εξαγωγή εκφράσεων κλειστής μορφής για τις ροπές του SNR στην έξοδο των δεκτών ΣΓΕ(2, L). Χρησιμοποιώντας αυτές τις εκφράσεις δίνονται, σε κλειστή μορφή επίσης, σημαντικά κριτήρια επίδοσης όπως είναι ο Μέσος Λόγος Σήματος προς Θόρυβο (ΜΛΣΘ) και η Ποσότητα των Διαλείψεων (ΠΔ). Επιπλέον, με την εφαρμογή της ϑεωρίας των προσεγγιστών Pad´e και της προσέγγισης μέσω της Ροπο-Γεννήτριας Συνάρτησης (ΡΓΣ), γίνεται δυνατή η μελέτη της Πιθανότητας Διακοπής της Επικοινωνίας (ΠΔΕ) και της Μέσης Πιθανότητας Σφάλματος Bit (ΜΠΣΒ) του συστήματος. Στη συνέχεια, υποθέτοντας Ανεξάρτητα και Ομοιόμορφα Κατανεμημένα (ΑΟΚ) κανάλια διαλείψεων ΓG, παρουσιάστηκε σε κλειστή μορφή έκφραση για τις ροπές του SNR στην έξοδο των δεκτών ΣΓΕ(2, L), ενώ η ΡΓΣ και η Αθροιστική Συνάρτηση Κατανομής (ΑΣΚ) εξήχθησαν υπό τη μορφή απειροσειρών. Χρησιμοποιώντας αυτές τις εκφράσεις απλοποιείται σημαντικά η μελέτη των κριτηρίων επίδοσης της ΠΔΕ και της ΜΠΣΒ. Επιπλέον, παρουσιάστηκε η χωρητικότητα του καναλιού ΓG με τη χρήση ή όχι δεκτών διαφορισμού, με βάση πολύ γνωστές τεχνικές προσαρμογής της εκπομπής. Μια άλλη πολύ σημαντική ερευνητική περιοχή που αφορά τα κανάλια διαλείψεων και επηρεάζει σημαντικά την επίδοση των ασύρματων ψηφιακών τηλεπικοινωνιακών συστημάτων που λειτουργούν σε αυτά, είναι το σύνθετο περιβάλλον διάδοσης που δημιουργείται όταν οι ΔΜΙΚ συνυπάρχουν με Διαλείψεις Μεγάλης Κλίμακας (ΔΜΕΚ). Στην παρούσα διδακτορική διατριβή, αυτό το περιβάλλον διάδοσης μοντελοποιείται με τις σύνθετες κατανομές K και γενικευμένη-K (KG). Αυτές οι κατανομές είναι αρκετά γενικές, αλλά ταυτόχρονα και μαθηματικά ευέλικτες, και έχει διαπιστωθεί ότι είναι κατάλληλες για τη μοντελοποίηση των συνθηκών του καναλιού που προκύπτει όταν οι ΔΜΙΚ συμβαίνουν ταυτόχρονα με ΔΜΕΚ. Στην περίπτωση της κατανομής KG, εξάγονται σημαντικά στατιστικά χαρακτηριστικά του SNR στην έξοδο των δεκτών διαφορισμού Συνδυασμού Μεγίστου Λόγου (ΣΜΛ), Συνδυασμού ΄Ισης Απολαβής (ΣΙΑ), Συνδυασμού Μεταγωγής και Παραμονής (ΣΜΠ) και του δέκτη Διαφορισμού Επιλογής (ΔΕ). Χρησιμοποιώντας αυτές τις εκφράσεις και υποθέτωντας ΑΜΟΚ συνθήκες διαλείψεων του καναλιού KG, εξάγονται σε κλειστή μορφή τα κριτήρια επίδοσης ΜΛΣΘ, ΠΔΕ και ΠΔ. Επιπλέον, ακολουθώντας την προσέγγιση που βασίζεται στη ΡΓΣ για τους δέκτες ΣΜΛ και ΣΜΠ και τους προσεγγιστές Pad´e για τους δέκτες ΣΙΑ και ΔΕ μελετάται η ΜΠΣΒ. Για την περίπτωση της κατανομής K, υποθέτωντας επιπλέον συσχετισμένες ΤΜ, παρουσιάζεται και μελετάται η συσχετισμένη (δι-μεταβαλλόμενη) κατανομή K, η οποία χαρακτηρίζεται από μη ομοιόμορφους παράγοντες μορφοποίησης και κλίμακας. Η σύνθετη αυτή κατανομή είναι βολική για τη μοντελοποίηση του συσχετισμένου περιβάλλοντος διαλείψεων/σκίασης, όταν η συσχέτιση μεταξύ των πλατών των σημάτων και της ισχύος τους είναι διαφορετική. Για αυτήν την κατανομή, εξάγονται γενικές εκφράσεις με απειροσειρές για τη Συνάρτηση Πυκνότητας Πιθανότητας (ΣΠΠ), την ΑΣΚ και τις από κοινού ροπές. Υποθέτοντας ομοιόμορφους παράγοντες κλίμακας και ασυσχέτιστη ισχύ σημάτων, παρέχονται απλούστερες εκφράσεις για τη ΣΠΠ, την ΑΣΚ και τη χαρακτηριστική συνάρτηση, ενώ οι από κοινού ροπές εξάγονται σε κλειστή μορφή. Επιπλέον, βασιζόμενοι σε αυτές τις ϑεωρητικές εκφράσεις της δι-μεταβαλλόμενης κατανομής K, παρουσιάζεται η ανάλυση της επίδοσης σημαντικών δεκτών διαφορισμού, όπως είναι οι δέκτες ΣΜΛ, ΣΙΑ και ΔΕ οι οποίοι λειτουργούν σε συσχετισμένα κανάλια διαλείψεων K. Για το δέκτη ΔΕ μελετάται η ΠΔΕ, ενώ για τους δέκτες ΣΜΛ και ΣΙΑ η ΜΠΣΒ. Γενικά η προτεινόμενη ανάλυση συνοδεύεται από αριθμητικά αποτελέσματα, τα οποία επιδεικνύουν ξεκάθαρα τη χρησιμότητα και την ευελιξία της ϑεωρητικής προσέγγισης. Τα αριθμητικά αυτά αποτελέσματα περιλαμβάνουν σύγκριση της επίδοσης διαφορετικών δεκτών διαφορισμού, συνθηκών καναλιού και τεχνικών διαμόρφωσης. / At the beginning of the new millennium, it becomes totaly clear the important role and usefulness of the wireless communication systems and their applications, which continue to evolute in several areas including Personal Communication Services (PCS). In this context there is a continuing demand for increased capacity and integration of the provided services, such as voice, data and other types of traffic, over radio channels. One of the key technical problem areas that has emerged out is the requirement for determining the fading effects on network performance. This also the research area of the current Philosophy diploma, which aims to the theoretical study of the performance analysis of wireless communication systems, which are subject to fading. The current study has employed diversity reception as a fading mitigation technique. According to this technique, the receiver employs more than one antennae, in order to receive multiple copies of the transmitted signal. Hence, these copies are appropriate combined in order to satisfy network administrator demands. The technical literature concerning diversity reception techniques over fading channels is quite extensive. However, it deals mainly with fading channels, which are based on Rayleigh, Rice, and Nakagami-m distributions. Additionally, important distributions for modeling small scale fading are also the Weibull and, its generalization, the generalized-gamma (ΓG), which have recently gained an increased scientific interest in the wireless mobile communication area. However, the performance analysis of diversity receivers operating over fading channels, modeled by these distributions, has not been thoroughly investigated. Reasons for that include the composite statistical representation of several diversity receivers output in conjunction with the quite complicated mathematical expressions of the Weibull and/or ΓG distributions. In this thesis, by overcoming these difficulties the performance of a class of Generalized-Selection Combining (GSC) receivers operating over independent but non identically distributed Weibull fading channels is studied. The case where the two branches with the largest instantaneous Signal-to-Noise Ratio (SNR) are selected, from a total of L available GSC(2, L), is considered. By introducing a novel property for the product of moments of ordered Weibull Random Variables (RV), convenient closed form expressions for the moments of the GSC(2, L) output SNR are derived. Using these expressions, important performance criteria, such as average output SNR and amount of fading (AF ), are obtained in closed form. Furthermore, employing the Pad´e approximants theory and the Moment-Generating Function (MGF)-based approach, outage and bit-error rate performances are studied. Additionally, by considering independent and identical distributed (iid) ΓG fading channels, the moments of the output SNR of the GSC(2, L) receivers were derived, also, in closed form, while infinite series representations were obtained for the MGF and the Cumulative Distribution Function (CDF). Using these expressions, a simplified study of the Outage Probability (OP) and Average Bit Error Probability (ABEP) performance criteria can be performed. Moreover, the capacity of the ΓG fading channel, using different adaptation policies, is also studied, with and without diversity reception. Another important research issue concerning the performance analysis of wireless digital communications systems operating over fading channels, is the composite propagation environment arising in cases where fading is superimposed on shadowing. In the current thesis this propagation environment is modeled by the K and generalized-K (KG) composite distributions. These distributions are quite general and mathematical versatile, and have been found to be appropriate for accurately modeling the channel conditions when fading occurs simultaneously with shadowing. For the KG distribution, expressions for important statistical metrics of the output SNR for the Selection Diversity (SD) receiver and Maximal Ratio Combining (MRC), Equal Gain Combining (EGC), and Switch and Stay Combining (SSC) diversity receivers are derived. Using these expressions and by considering independent but not necessarily identical distributed fading channel conditions, performance criteria, such as average output SNR, AF and OP are obtained in closed form. Moreover, following the MGFbased approach for MRC and SSC receivers, and the Pad´e approximants theory for SD and EGC receivers, the ABEP is also studied. For the case of the K distribution, assuming correlated RVs, the correlated (bivariate) K distribution, with not necessarily identical shaping and scaling parameters, is introduced and studied. This composite distribution is convenient for modeling multipath/shadowing correlated fading environments when the correlations between the signal envelopes and their powers are different. For this distribution, generic infinite series representations are derived for the Probability Density Function (PDF), the CDF and the joint moments. Assuming identical shaping parameters and uncorrelated signal powers, simpler expressions for the PDF, CDF and the Characteristic Function (CF) are provided, while the joint moments are derived in closed form. Capitalizing on these theoretical expressions for the statistical characteristics of the correlated K distribution, the performance analysis of various diversity reception techniques, such as MRC, EGC and SD, over bivariate K fading channels is presented. For the SD, the outage probability is studied, while for the MRC and EGC the ABEP is obtained. In general the proposed analysis is accompanied by several numerical evaluated results, clearly demonstrating the usefulness and flexibility of the theoretical approach. These numerical results include performance comparisons of several diversity receiver structures, employing various modulation formats and different channel conditions.

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.

Current network architectures, as well as mobility management protocols, are generally deployed in a centralized manner. Moreover, these protocols are designed to be always activated, even when not needed. As the number of mobile users and the volume of their traffic increase, such centralized architectures are expected to encounter scalability issues as well as performance issues. Recently, mobile network operators are experiencing a rapid increase in mobile data traffic. In order to cope with this, a new trend is to flatten networks architectures and hence IP mobility management protocols need to be adapted for such evolution. Therefore, there is a need to define novel mobility management mechanisms that are both distributed and offered dynamically. In order to cope with this context, the thesis concerns designing, analyzing, and evaluating novel IPv6 network architectures and mobility protocols that are distributed and dynamic, and in particular Distributed Mobility Management (DMM). Before proposing any extension, we categorize the existing mobility schemes and carry out a comparative analysis on each category. Then, we propose a new distributed dynamic mobility management scheme based on the Mobile IPv6 (MIPv6) protocol, with an optional extension for the joint use with the Session Initiation Protocol (SIP). After, we carry out a performance analysis in terms of mobility costs, handover, and quality-of-service. Finally, we study the impacts on other aspects such as security considerations and location management.

An examination and exploration of ‘the (syn)aesthetic style’, a particular sensate mode of performance and appreciation that has become prominent in recent years in contemporary arts practice. The (syn)aesthetic performance style fuses disciplines and techniques to create interdisciplinary and intersensual work with emphasis upon; the (syn)aesthetic hybrid; the prioritisation of the body in performance and the visceral-verbal ‘play-text’. ‘(Syn)aesthetics’ is adopted as an original discourse for the analysis of such work, appropriating certain quintessential features of the physiological condition of synaesthesia to clarify the impulse in performance and appreciation which affects a ‘disturbance’ within audience interpretation. Original terms employed attempt to elucidate the complex appreciation strategies integral to this performance experience. These include the double-edged semantic/somatic or making-sense/sense-making process of appreciation, which embraces the individual, immediate and innate, and the ‘corporeal memory’ of the perceiving body. Liveness and the live(d) moment are considered, alongside notions of ritual and transcendence and the primordial and technological. The argument surveys the inheritance that saw to this contemporary style emerging, in Britain in particular, considering female performance practice, intercultural and interdisciplinary ensemble performance and the ‘New Writing’ aesthetic. Critical and performance theorists referred to include Friedrich Nietzsche, the Russian Formalists, Roland Barthes, Julia Kristeva, Hélène Cixous, Luce Irigaray, Antonin Artaud, Valère Novarina, Howard Barker and Susan Broadhurst. Contemporary practitioners highlighted as case studies exemplary of (syn)aesthetic practice are Sara Giddens, Marisa Carnesky, Caryl Churchill and Sarah Kane. Furthermore, documentation of a series of original performance workshops explores the (syn)aesthetic impulse in performance and analysis from the perspectives of writer, performer and audience. (Syn)aesthetics as an interpretative device endeavours to enhance understanding of the intangible areas of performance which are increasingly difficult to articulate, thereby presenting a mode of analysis that extends performance theory for students and practitioners within the arts.

In networks, using large buffers tend to increase end-to-end packet delay and its deviations, conflicting with real-time applications such as online gaming, audio-video services, IPTV, and VoIP. Further, large buffers complicate the design of high speed routers, leading to more power consumption and board space. According to Moore's law, switching speeds double every 18 months while memory access speeds double only every 10 years. Hence, as memory requirements increasingly become a limiting aspect of router design, studying networks in finite-buffer regime seems necessary for network engineers. This work focuses on both practical and theoretical aspects of finite-buffer networks. In Chapters 1-7, we investigate the effects of finite buffer sizes on the throughput and packet delay in different networks. These performance measures are shown to be linked to the stationary distribution of an underlying irreducible Markov chain that exactly models the changes in the network. An iterative scheme is proposed to approximate the steady-state distribution of buffer occupancies by decoupling the exact chain to smaller chains. These approximate solutions are used to analytically characterize network throughput and packet delay, and are also applied to some network performance optimization problems. Further, using simulations, it is confirmed that the proposed framework yields accurate estimates of the throughput and delay performance measures and captures the vital trends and tradeoffs in these networks. In Chapters 8-10, we address the problem of modeling and analysis of the performance of finite-memory random linear network coding in erasure networks. When using random linear network coding, the content of buffers creates dependencies which cannot be captured directly using the classical queueing theoretical models. A careful derivation of the buffer occupancy states and their transition rules are presented as well as decodability conditions when random linear network coding is performed on a stream of arriving packets.

The heavy-tailed traffic from wireless users, caused by the emerging Internet and multimedia applications, induces extremely dynamic and variable network environment, which can fundamentally change the way in which wireless networks are conceived, designed, and operated. This thesis is concerned with modeling, analysis, and optimization of wireless networks in the presence of heavy tails. First, a novel traffic model is proposed, which captures the inherent relationship between the traffic dynamics and the joint effects of the mobility variability of network users and the spatial correlation in their observed physical phenomenon. Next, the asymptotic delay distribution of wireless users is analyzed under different traffic patterns and spectrum conditions, which reveals the critical conditions under which wireless users can experience heavy-tailed delay with significantly degraded QoS performance. Based on the delay analysis, the fundamental impact of heavy-tailed environment on network stability is studied. Specifically, a new network stability criterion, namely moment stability, is introduced to better characterize the QoS performance in the heavy-tailed environment. Accordingly, a throughput-optimal scheduling algorithm is proposed to maximize network throughput while guaranteeing moment stability. Furthermore, the impact of heavy-tailed spectrum on network connectivity is investigated. Towards this, the necessary conditions on the existence of delay-bounded connectivity are derived. To enhance network connectivity, the mobility-assisted data forwarding scheme is exploited, whose important design parameters, such as critical mobility radius, are derived. Moreover, the latency in wireless mobile networks is analyzed, which exhibits asymptotic linearity in the initial distance between mobile users.

No description available.