109 research outputs found
Supplemental_material – Supplemental material for The effect and optimal parameters of repetitive transcranial magnetic stimulation on motor recovery in stroke patients: a systematic review and meta-analysis of randomized controlled trials
No full textSupplemental material, Supplemental_material for The effect and optimal parameters of repetitive transcranial magnetic stimulation on motor recovery in stroke patients: a systematic review and meta-analysis of randomized controlled trials by Huifang Xiang, Jing Sun, Xiang Tang, Kebin Zeng and Xiushu Wu in Clinical Rehabilitation</p
Graphical Model Debugger Framework for Embedded Systems
No full textModel Driven Software Development has offered a faster way to design and implement embedded real-time software by moving the design to a model level, and by transforming models to code. However, the testing of embedded systems has remained at the code level. This paper presents a Graphical Model Debugger Framework, providing an auxiliary avenue of analysis of system models at runtime by executing generated code and updating models synchronously, which allows embedded developers to focus on the model level. With the model debugger, embedded developers can graphically test their design model and check the running status of the system, which offers a debugging capability on a higher level of abstraction. The framework intends to contribute a tool to the Eclipse society, especially suitable for model-driven development of embedded systems
Logics and Models for Stochastic Analysis Beyond Markov Chains
No full textGennem de sidste 20 år er logik og modeller for stokastisk analyse af informationssystemer blevet studeret bredt både i teori og praksis. De kvantitative egenskaber såsom ydeevne og pålidelighed er blevet bedømt ved hjælp af diskrete og kontinuerte Markovkæder. Denne afhandling overfører stokastiske analyseteknikker for Markovkæder til brug for den mere generelle klasse af Markovprocesser bestående af Fasetype (PH) fordelinger og MatrixEksponentielle (ME) fordelinger. Således kan Markovprocesser med ME fordelte tider ikke formuleres som en Markovproces med hverken endeligt eller tælleligt tilstandsrum.PH fordelinger er kendt for at have mange eksplicitte, analytiske egenskaber. Det medfører, at systemer, der har PH fordelte komponenter, kan formuleres som Markovkæder. Denne afhandling præsenterer flere resultater relateret til PH fordelinger. Vi viser først, hvordan en eksplicit, analytiske form af diskrete PH fordelinger kan bruges som beregningsmæssigt værktøj til at måle ydeevnen af trådløse, parallelle sensornetværk.Ved hjælp af stokastisk procesalgebra, som er et bredt accepteret paradigme, undersøger vi dernæst kompositionaliteten af kontinuerte PH fordelinger. Dette gøres for at redegøre for modellering af parallelle, stokastiske systemer, der bygger på PH repræsentationer. Til slut betragter vi diskrete mærkede punktprocesser med PH fordelte mellemankomsttider. Vi foreslår time-lapse bisimulation, som er en tilstandsbaseret karakterisering af lighedsrelationen mellem punktprocesserne. Vi redegør for, hvordan time-lapse bisimulation er et nyt bidrag til den eksisterende samling af bisimlationer, der beskriver den probabillistiske opførsel over tid af mærkede, diskrete Markovkæder.Selvom klassen af ME fordelinger er strengt større end klassen af PH fordelinger, kan mange af de resultater, der gælder for PH fordelinger, overføres til klassen af ME fordelinger. ME fordelinger har en særlig tiltalende egenskab nemlig minimalitetsegenskaben. Denne betyder, at en ME repræsentation af en PH fordeling generelt vil være af lavere dimension end PH repræsentationen, og der eksisterer altid en ME repræsentation med minimal dimension. Dog tvinger den generelle formulering af ME fordelinger os til at forlade Markovkædernes verden. For at kunne håndtere ME fordelinger med flere udgange introducerer vi en multi–exits ME fordeling. Ved brug af MEME procesalgebra er vi nu i stand til at formulere systemer med samme semantik som en Markov–fornyelsesproces med ME kerne. Fordelen ved dette er, at samtlige komponenter både før og efter kompositioner kan repræsenteres ved en minimal repræsentation. For at kunne udføre kvantitativ efterprøvning præsenterer vi foreslået en stokastisk algoritme til modeltjek for vores problem.Within the last twenty years, logics and models for stochastic analysis of information systems have been widely studied in both theory and practice. The quantitative properties, such as performance and reliability, are evaluated over discrete–time and continuous–time Markov chains. This thesis lifts the stochastic analysis techniques from the class of Markov chains to the more general classes of stochastic processes having PHase–type (PH) distributions and Matrix–Exponential (ME) distributions, such that a Markov renewal process with ME kernels that cannot be formulated as a Markov process with finite or countable state space.PH distributions are known for many explicit analytic properties, such that systems having PH distributed components can still be formulated as Markov chains. This thesis presents several results related to PH distributions. We first show how to use the explicit analytic form of discrete PH distributions as computational vehicle on measuring the performance of concurrent wireless sensor networks. Secondly, choosing stochastic process algebras as a widely accepted formalism, we study the compositionality of continuous PH distributions in order to support modelling concurrent stochastic systems having PH representations as building blocks. At last, we consider discrete–time point processes having PH distributed interarrival times with multiple marks, we propose time-lapse bisimulation, a state-based characterisation of the equivalence relation between the point processes. We clarify that time-lapse bisimulation is a new contribution to the existing bisimulation family, which captures probabilistic behaviourover time for labelled discrete–time Markov chains.ME distributions is a strictly larger class than PH distributions, such that many results from PH distributions also are valid for ME distributions. ME distributions have a very appealing property, called minimality property: generally a ME representation of a PH distribution will be of lower dimension than PH representations, and one can always find a ME representation with the minimal dimension. However, because of the generality of ME distributions, we have to leave the world of Markov chains. To support ME distributions with multiple exits, we introduce a multi-exits ME distribution together with a process algebra MEME to express the systems having the semantics as Markov renewal processes with ME kernels. The most appealing feature is that all the components before and after compositions are secured to have a minimal state space representation. To support quantitative verification, we also propose stochastic model checking algorithms to our problem
The stochastic quality calculus
No full textWe introduce the Stochastic Quality Calculus in order to model and reason about distributed processes that rely on each other in order to achieve their overall behaviour. The calculus supports broadcast communication in a truly concurrent setting. Generally distributed delays are associated with the outputs and at the same time the inputs impose constraints on the waiting times. Consequently, the expected inputs may not be available when needed and therefore the calculus allows to express the absence of data.The communication delays are expressed by general distributions and the resulting semantics is given in terms of Generalised Semi-Markov Decision Processes. By restricting the distributions to be continuous and by allowing truly concurrent communication we eliminate the non-determinism and arrive at Generalised Semi-Markov Processes (GSMPs); further restriction to exponential distributions gives rise to numerically analysable GSMPs, in particular using techniques from stochastic model checking
Stochastic Model Checking of the Stochastic Quality Calculus
No full textThe Quality Calculus uses quality binders for input to express strategies for continuing the computation even when the desired input has not been received. The Stochastic Quality Calculus adds generally distributed delays for output actions and real-time constraints on the quality binders for input. This gives rise to Generalised Semi-Markov Decision Processes for which few analytical techniques are available. We restrict delays on output actions to be exponentially distributed while still admitting real-time constraints on the quality binders. This facilitates developing analytical techniques based on stochastic model checking and we compute closed form solutions for a number of interesting scenarios. The analyses are applied to the design of an intelligent smart electrical meter of the kind to be installed in European households by 2020
The Stochastic Quality Calculus
No full textInternational audienceWe introduce the Stochastic Quality Calculus in order to model and reason about distributed processes that rely on each other in order to achieve their overall behaviour. The calculus supports broadcast communication in a truly concurrent setting. Generally distributed delays are associated with the outputs and at the same time the inputs impose constraints on the waiting times. Consequently, the expected inputs may not be available when needed and therefore the calculus allows to express the absence of data.The communication delays are expressed by general distributions and the resulting semantics is given in terms of Generalised Semi-Markov Decision Processes. By restricting the distributions to be continuous and by allowing truly concurrent communication we eliminate the non-determinism and arrive at Generalised Semi-Markov Processes (GSMPs); further restriction to exponential distributions gives rise to numerically analysable GSMPs, in particular using techniques from stochastic model checking
Graphical Model Debugger Framework for Embedded Systems
No full textModel Driven Software Development has offered a faster way to design and implement embedded real-time software by moving the design to a model level, and by transforming models to code. However, the testing of embedded systems has remained at the code level. This paper presents a Graphical Model Debugger Framework, providing an auxiliary avenue of analysis of system models at runtime by executing generated code and updating models synchronously, which allows embedded developers to focus on the model level. With the model debugger, embedded developers can graphically test their design model and check the running status of the system, which offers a debugging capability on a higher level of abstraction. The framework intends to contribute a tool to the Eclipse society, especially suitable for model-driven development of embedded systems.<br/
- …
