Consider a time-dependent optimal control problem, where the state evolution is described by an initial value problem. There are a variety of numerical methods to solve these problems. The so-called indirect approach is considered detailed in this thesis. The indirect methods solve decoupled boundary value problems resulting from the necessary conditions for the optimal control problem. The so-called Pantoja method describes a computationally efficient stage-wise construction of the Newton direction for the discrete-time optimal control problem. There are many relationships between multiple shooting techniques and Pantoja method, which are investigated in this thesis. In this context, the equivalence of Pantoja method and multiple shooting method of Riccati type is shown. Moreover, Pantoja method is extended to the case where the state equations are discretised using one of implicit numerical methods. Furthermore, the concept of symplecticness and Hamiltonian systems is introduced. In this regard, a suitable numerical method is presented, which can be applied to unconstrained optimal control problems. It is proved that this method is a symplectic one. The iterative solution of optimal control problems in ordinary differential equations by Pantoja or Riccati equivalent methods leads to a succession of triple sweeps through the discretised time interval. The second (adjoint) sweep relies on information from the first (original) sweep, and the third (final) sweep depends on both of them. Typically, the steps on the adjoint sweep involve more operations and require more storage than the other two. The key difficulty is given by the enormous amount of memory required for the implementation of these methods if all states throughout forward and adjoint sweeps are stored. One of goals of this thesis is to present checkpointing techniques for memory reduced implementation of these methods. For this purpose, the well known aspect of checkpointing has to be extended to a `nested checkpointing` for multiple transversals. The proposed nested reversal schedules drastically reduce the required spatial complexity. The schedules are designed to minimise the overall execution time given a certain total amount of storage for the checkpoints. The proposed scheduling schemes are applied to the memory reduced implementation of the optimal control problem of laser surface hardening and other optimal control problems. / Es wird ein Problem der optimalen Steuerung betrachtet. Die dazugehoerigen Zustandsgleichungen sind mit einer Anfangswertaufgabe definiert. Es existieren zahlreiche numerische Methoden, um Probleme der optimalen Steuerung zu loesen. Der so genannte indirekte Ansatz wird in diesen Thesen detailliert betrachtet. Die indirekten Methoden loesen das aus den Notwendigkeitsbedingungen resultierende Randwertproblem. Das so genannte Pantoja Verfahren beschreibt eine zeiteffiziente schrittweise Berechnung der Newton Richtung fuer diskrete Probleme der optimalen Steuerung. Es gibt mehrere Beziehungen zwischen den unterschiedlichen Mehrzielmethoden und dem Pantoja Verfahren, die in diesen Thesen detailliert zu untersuchen sind. In diesem Zusammenhang wird die aequivalence zwischen dem Pantoja Verfahren und der Mehrzielmethode vom Riccati Typ gezeigt. Ausserdem wird das herkoemlige Pantoja Verfahren dahingehend erweitert, dass die Zustandsgleichungen mit Hilfe einer impliziten numerischen Methode diskretisiert sind. Weiterhin wird das Symplektische Konzept eingefuehrt. In diesem Zusammenhang wird eine geeignete numerische Methode praesentiert, die fuer ein unrestringiertes Problem der optimalen Steuerung angewendet werden kann. In diesen Thesen wird bewiesen, dass diese Methode symplectisch ist. Das iterative Loesen eines Problems der optimalen Steuerung in gewoenlichen Differentialgleichungen mit Hilfe von Pantoja oder Riccati aequivalenten Verfahren fuehrt auf eine Aufeinanderfolge der Durchlaeufetripeln in einem diskretisierten Zeitintervall. Der zweite (adjungierte) Lauf haengt von der Information des ersten (primalen) Laufes, und der dritte (finale) Lauf haeng von den beiden vorherigen ab. Ueblicherweise beinhalten Schritte und Zustaende des adjungierten Laufes wesentlich mehr Operationen und benoetigen auch wesentlich mehr Speicherplatzkapazitaet als Schritte und Zustaende der anderen zwei Durchlaeufe. Das Grundproblem besteht in einer enormen Speicherplatzkapazitaet, die fuer die Implementierung dieser Methoden benutzt wird, falls alle Zustaende des primalen und des adjungierten Durchlaufes zu speichern sind. Ein Ziel dieser Thesen besteht darin, Checkpointing Strategien zu praesentieren, um diese Methoden speichereffizient zu implementieren. Diese geschachtelten Umkehrschemata sind so konstruiert, dass fuer einen gegebenen Speicherplatz die gesamte Laufzeit zur Abarbeitung des Umkehrschemas minimiert wird. Die aufgestellten Umkehrschemata wurden fuer eine speichereffiziente Implementierung von Problemen der optimalen Steuerung angewendet. Insbesondere betrifft dies das Problem einer Oberflaechenabhaertung mit Laserbehandlung.

A transaction-consistent global checkpoint of a database records a state of the database which reflects the effect of only completed transactions and not the re- sults of any partially executed transactions. This thesis establishes the necessary and sufficient conditions for a checkpoint of a data item (or the checkpoints of a set of data items) to be part of a transaction-consistent global checkpoint of the database. This result would be useful for constructing transaction-consistent global checkpoints incrementally from the checkpoints of each individual data item of a database. By applying this condition, we can start from any useful checkpoint of any data item and then incrementally add checkpoints of other data items until we get a transaction- consistent global checkpoint of the database. This result can also help in designing non-intrusive checkpointing protocols for database systems. Based on the intuition gained from the development of the necessary and sufficient conditions, we also de- veloped a non-intrusive low-overhead checkpointing protocol for distributed database systems. Checkpointing and rollback recovery are also established techniques for achiev- ing fault-tolerance in distributed systems. Communication-induced checkpointing algorithms allow processes involved in a distributed computation take checkpoints independently while at the same time force processes to take additional checkpoints to make each checkpoint to be part of a consistent global checkpoint. This thesis develops a low-overhead communication-induced checkpointing protocol and presents a performance evaluation of the protocol.

The emergence of the Internet of Things (IoT) has increased the demand for networked, resource-constrained devices tremendously. Many of the devices used for IoT applications are designed to be resource-constrained, as they typically must be small, inexpensive, and powered by batteries. In this dissertation, we consider a number of challenges pertaining to these constraints: system support for energy efficiency; flash-based storage systems; programming, testing, and debugging; and safe and secure application execution. The contributions of this dissertation are made through five research papers addressing these challenges. Firstly, to enhance the system support for energy-efficient storage in resource-constrained devices, we present the design, implementation, and evaluation of the Coffee file system and the Antelope DBMS. Coffee provides a sequential write throughput that is over 92% of the attainable flash driver throughput, and has a constant memory footprint for open files. Antelope is the first full-fledged relational DBMS for sensor networks, and it provides two novel indexing algorithms to enable fast and energy-efficient database queries. Secondly, we contribute a framework that extends the functionality and increases the performance of sensornet checkpointing, a debugging and testing technique. Furthermore, we evaluate how different data compression algorithms can be used to decrease the energy consumption and data dissemination time when reprogramming sensor networks. Lastly, we present Velox, a virtual machine for IoT applications. Velox can enforce application-specific resource policies. Through its policy framework and its support for high-level programming languages, Velox helps to secure IoT applications. Our experiments show that Velox monitors applications' resource usage and enforces policies with an energy overhead below 3%. The experimental systems research conducted in this dissertation has had a substantial impact both in the academic community and the open-source software community. Several of the produced software systems and components are included in Contiki, one of the premier open-source operating systems for the IoT and sensor networks, and they are being used both in research projects and commercial products.

Les avancées technologiques ont conduit les grandes organisations telles que les entreprises,les universités et les instituts de recherche à se doter d'intranets constitués de plusieurs serveurs etd'un grand nombre de postes de travail. Cependant dans certaines de ces organisations, les postes detravail sont très peu utilisés pendant la nuit, les week-ends et les périodes de congés, libérant ainsiune grande puissance de calcul disponible et inutilisée.Dans cette thèse, nous étudions l'exploitation de ces temps de jachère afin d'exécuter desapplications de calcul haute performance. A cet effet, nous supposons que les postes acquis sontrebootés et intégrés à des grappes virtuelles constituées dynamiquement. Toutefois, ces temps dejachère ne permettent pas toujours d'exécuter les applications jusqu'à leur terme. Les mécanismes desauvegarde/reprise (checkpointing) sont alors utilisés pour sauvegarder, dans un certain délai, lecontexte d'exécution des applications en vue d'une éventuelle reprise. Il convient de noter que lasauvegarde de tous les processus dans les délais impartis n'est pas toujours possible. Nousproposons un modèle d'ordonnancement des sauvegardes en parallèle, qui tient compte descontraintes temporelles imposées et des contraintes liées aux bandes passantes (réseau et disque),pour maximiser les temps de calcul déjà effectués pour les applications candidates à la sauvegarde. / The technological advances has led major organizations such as enterprises, universities andresearch institutes to acquire intranets consisting of several servers and many workstations.However, in some of these organizations, the resources are rarely used at nights, weekends and onholidays, thus releasing a large computing power available and unused.This thesis discusses the exploitation of the idle period of workstaions in order to run HPCapplications. The workstations retained are restarted and integrated in dynamically formed clusters.However, the idle periods do not always permit the complete carrying out of the computationsallocated to them. The checkpointing mechanisms are then used to save in a certain period, theexecution context of applications for a possible restart. It is worth nothing that checkpointing all theprocesses in the required period is not always possible. We propose a scheduling model ofcheckpointing in parallel, which takes into account the time constraints imposed and the bandwidthconstraints (network and disk) to maximize the computation time already taken for the applicationswhich are to be checkpointed.

<p>In this project we illustrate how queueing simulation may be used to find the optimal interval for checkpointing problems and compare results with theoretical computations for simple systems that may be treated analytically.</p><p>We consider a relatively simple model for an internet banking facility. From time to time, the application server breaks down. The information at the time of the breakdown has to be passed onto the back up server before service may be resumed. To make the change over as efficient as possible, information of the state of user’s accounts is saved at regular intervals. This is known as checkpointing.</p><p>Firstly, we use GPSS (a queueing simulation tool) to find, by simulation, an optimal checkpointing interval, which maximises the efficiency of the server. Two measures of efficiency are considered; the availability of the server and the average time a customer spends in the system. Secondly, we investigate how far the queueing theory can go to providing an analytic solution to the problem and see whether or not this is in line with the results obtained through simulation.</p><p>The analysis shows that checkpointing is not necessary if breakdowns occur frequently and log reading after failure does not take much time. Otherwise, checkpointing is necessary and the analysis shows how GPSS may be used to obtain the optimal checkpointing interval. Relatively complicated systems may be simulated, where there are no analytic tools available. In simple cases, where theoretical methods may be used, the results from our simulations correspond with the theoretical calculations.</p>

In this project we illustrate how queueing simulation may be used to find the optimal interval for checkpointing problems and compare results with theoretical computations for simple systems that may be treated analytically. We consider a relatively simple model for an internet banking facility. From time to time, the application server breaks down. The information at the time of the breakdown has to be passed onto the back up server before service may be resumed. To make the change over as efficient as possible, information of the state of user’s accounts is saved at regular intervals. This is known as checkpointing. Firstly, we use GPSS (a queueing simulation tool) to find, by simulation, an optimal checkpointing interval, which maximises the efficiency of the server. Two measures of efficiency are considered; the availability of the server and the average time a customer spends in the system. Secondly, we investigate how far the queueing theory can go to providing an analytic solution to the problem and see whether or not this is in line with the results obtained through simulation. The analysis shows that checkpointing is not necessary if breakdowns occur frequently and log reading after failure does not take much time. Otherwise, checkpointing is necessary and the analysis shows how GPSS may be used to obtain the optimal checkpointing interval. Relatively complicated systems may be simulated, where there are no analytic tools available. In simple cases, where theoretical methods may be used, the results from our simulations correspond with the theoretical calculations.

<p>Recovery is a fundamental service in database systems. In this work, we present a new mechanism for diskless real-time recovery in fully replicated distributed real-time database systems. Traditionally, recovery has relied on disk-resident redundant data. Unfortunately, disks cannot always be used in real-time systems since these systems are sometimes used in environments which do not allow the use of disks. Also, minimizing the amount of hardware can save money, especially in mass-produced products. Instead of loading the database from disk, our recovery mechanism enables a restarted node to retrieve a copy of the database from an arbitrary remote node. The recovery mechanism does not violate timeliness during normal processing and, during recovery, all nodes except for the recovering node can guarantee the timeliness of critical transactions. The mechanism uses fuzzy checkpointing to copy the database to the recovering node. Fuzzy checkpointing has been chosen since it copies the database without regard to concurrency control and, thus, does not increase data contention in the database. We conclude that the suggested recovery mechanism is a feasible option for fully replicated distributed real-time database systems.</p>

Recovery is a fundamental service in database systems. In this work, we present a new mechanism for diskless real-time recovery in fully replicated distributed real-time database systems. Traditionally, recovery has relied on disk-resident redundant data. Unfortunately, disks cannot always be used in real-time systems since these systems are sometimes used in environments which do not allow the use of disks. Also, minimizing the amount of hardware can save money, especially in mass-produced products. Instead of loading the database from disk, our recovery mechanism enables a restarted node to retrieve a copy of the database from an arbitrary remote node. The recovery mechanism does not violate timeliness during normal processing and, during recovery, all nodes except for the recovering node can guarantee the timeliness of critical transactions. The mechanism uses fuzzy checkpointing to copy the database to the recovering node. Fuzzy checkpointing has been chosen since it copies the database without regard to concurrency control and, thus, does not increase data contention in the database. We conclude that the suggested recovery mechanism is a feasible option for fully replicated distributed real-time database systems.

Avec l'évolution des machines parallèles, le besoin en protocole de tolérance aux fautes devient de plus en plus important. Les protocoles de tolérance aux fautes existants ne sont pas adaptés à ces architectures car soit ils forcent un redémarrage global (protocoles de sauvegarde de points de reprise coordonnés) soit ils forcent l'enregistrement de tous les messages (protocoles à enregistrement de messages). Nous avons étudié les caractéristiques des protocoles existants. Dans un premier temps, nous avons étudié le déterminisme des applications, étant donné que les protocoles existants supposent des exécutions non déterministes ou déterministes par morceaux. Dans notre étude, nous nous sommes intéressés au modèle par échange de messages, et plus précisément aux applications MPI. Nous avons analysé 26 applications MPI et avons mis avant une nouvelle caractéristique appelée "déterminisme des émissions" qui correspond à la majorité des applications étudiées. Dans un second temps, nous nous sommes intéressés aux schémas de communications des applications afin d'étudier l'existence des groupes de processus dans ces schémas. L'étude a montré que pour la plupart des applications, il est possible de créer des groupes de processus de façon à minimiser la taille des groupes et le volume des messages inter-groupe. A partir de là nous avons proposé deux protocoles de tolérance aux fautes. Le premier est un protocole de sauvegarde de points de reprise non coordonnés pour les applications à émissions déterministes qui évite l'effet domino en n'enregistrant qu'un sous ensemble des messages de l'application. Nous avons également adapté le protocole pour l'utiliser sur des groupes de processus. Par la suite, nous avons proposé HydEE, un protocole hiérarchique fondé sur le déterminisme des émissions et les groupes de processus. Il combine un protocole de sauvegarde de points de reprise coordonnés au sein des groupes à un protocole à enregistrement de messages entre les groupes. / With the evolution of parallel computers, the need for fault tolerance protocols is becoming increasingly important. The existing fault tolerance protocols are not adapted to thèse architectures because they either force a global restard (coordinated checkpointing protocols) or all message logging (message logging protocols). We studied the characteristics of the existing protocols. We first studied the determinism of the applications, since existing protocols assumenon deterministic or piecewise deterministic executions. In our study, we examined the message passing model, and more specifically MPI applications. We have analyzed26 MPI applications and have put forward a new characteristic called "send-determinism" which corresponds to moststudied applications. In a second step, we studied the communication patterns of the applications to study the existence of clusters of processes in these patterns. The study showed that for most applications, it is possible to create clusters of processes to minimize the size of clusters and the volume of inter-cluster messages. Then we designed two fault tolérance protocols. The first one is an uncoordinated checkpointing protocol which is based on the send-deterministic assumption and avoids emissions deterministic domino effect while logging only a subset of the application messages. We have also adapted the protocol to clusters of processes. Then, we proposed HydEE, a hierarchical protocol that is lso based on the send-deterministic assumption and that is used on clusters of processes. It combines coordinated checkpointing protocol inside clusters to a message logging protocol for inter-cluster messages.

No description available.