Claude Duvallet

How to Model a Real-Time Database?

This paper presents a framework for real-time database design that is able to support real-time database requirements such as time-constrained data and time-constrained transactions. It is based upon a real-time object-oriented data model in which each object encapsulates time-constrained data, time-constrained methods and concurrency control mechanisms. Our framework is composed of an UML profile for real-time database, a translator to object-relational model, and an UML CASE Tool.

Structural Model of Real-Time Databases: An Illustration

A real-time database is a database in which both the data and the operations upon the data may have timing constraints. The design of this kind of database requires the introduction of new concepts to modelize both data structures and the dynamic behavior of the database. In this paper, we propose an UML2.0 profile, entitled UML-RTDB, allowing the design of structural model for a real-time database. One of the main advantages of UML-RTDB is its capacity to take into account real-time database properties through specialized concepts in rigourous, easy and expressive manner.

Management of QoS and data freshness in RTDBSs using feedback control scheduling and data versions

These recent years, a lot of real-time applications are becoming increasingly sophisticated in their data needs, resulting in a greater demand for real-time data services. Real-time database systems (RTDBS) can manage these applications, but the workload in these systems is unpredictable, then RTDBS may become overloaded. A lot of work dealing with quality of service (QoS) has been done to control the transient overshoot. They are based on feedback control real-time scheduling theory. In this paper, we propose a work which allows to execute transactions before their deadlines while using the fresh data. We have extended the feedback-based miss ratio control, by using a multi versions data architecture to guarantee a set of requirements on the behavior of RTDBS and we have considered two data management policies. In the first policy, we have limited the maximum number of data versions and this number is the same for all the data items. In the second policy, the number is dynamically adjusted for each data item.

A process to derive domain-specific patterns: application to the real time domain

Domain Specific Design Patterns are sets of objects and components that capture the experts knowledge and that can be used in a specific software domain. They provide for a higher software quality and a reduced development cost. However, their design remains a difficult task due to the generality and variability they must encompass, in order to be instantiated for various applications in the domain. This paper presents a design process that generates domain specific design patterns from a set of concrete application designs. The proposed process defines unification rules that apply a set of comparison criteria on various applications in the pattern domain. In addition, domain requirements, constraints and invariants are extracted and then confronted to the pattern, in order to validate it. The process integrates bottom-up and top-down approaches and assists the designer in the construction of domain specific patterns. Finally, the design process is illustrated and evaluated through the design of a pattern in the real-time domain.

Multi versions data for improvement of QoS in RTDBS

In current research toward the design of more powerful behavior of RTDBS under unpredictable workloads, different research groups focus their work on QoS (Quality of Service) guarantee. Their research is often based on feedback control real-time scheduling theory. In this paper we propose a technique which allows to execute transactions on time using fresh and precise data while taking into account the global size of the database. We have extended the feedback-based miss ratio control by both using multi-versions data, and proposing a data management policy combining (1) limitation of the versions number and (2) dynamic adjustment of this limit according to a maximum database size parameter.

RTO-RTDB: A Real-Time Object-Oriented Database Model

In recent years search for proper extensions of the object model to suit real-time database community has become a critical research issue. In this paper, we present such an extension of the object model, called RTO-RTDB. A concrete static and dynamic views of the model are presented. A major attraction of this model is in that it is effective not only in the use of the feedback control real-time scheduling theory but also in the accurate representation of realtime databases properties. It is used to demonstrate how to separate structural and concurrency concerns ensuring also high-level abstraction for parallelism modeling. In order to improve and to facilitate the RTO-RTDB object model specification, we have used the MARTE and UML-RTDB profiles as UML extensions to describe real-time aspects on one hand, and to express real-time database features on the other hand.

Des systèmes multiagents anytime pour la conception de systèmes d'aide à la décision

RESUME. Dans un systeme d’aide a la decision (S.A.D.), un systeme informatique doit permettre au decideur de prendre la meilleure decision possible, souvent avant une echeance donnee. Nous considerons dans ce papier les S.A.D. qui reposent sur une modelisation multiagent. L’objectif de notre travail est d’etendre la notion de systeme multiagent a un environnement contraint par le temps, c’est-a-dire de prendre en compte les aspects temps reel des applications d’aide a la decision. La voie que nous avons choisie pour concevoir de tels systemes est l’utilisation de techniques anytime. Les techniques anytime sont des techniques dont l’objectif est de doter les systemes informatiques de possibilites de fournir des resultats dont la qualite et la precision evoluent en fonction du temps qui leur est imparti. Dans cet article, nous proposons ANYMAS (ANYtime Multi-Agents System), un modele de conception de systemes multiagents temps reel fonde sur des techniques anytime. Nous validons notre travail par la conception d’une application d’aide a la gestion d’un marche electronique.

Intelligent and self-adaptive interface

The information systems for decision-making must provide basic elements to the decision-maker in a synthetic and simple way. So, it is necessary to build Computer Human Interface (CHI) adapted to different users perceptions as well as possible. Therefore, the necessity to design the application data processing providing an intelligent and self-adaptive CHI seems to be more and more necessary. An essential and necessary characteristic of this kind of CHI is the capacity to adapt it to the environment and to the users behavior, and to permit the addition of components without putting back in question the design of this CHI. In order to have these characteristics in an application, a modelling using intelligent agents seems to be well adapted because it permits to take into account the complexe interaction present in the CHI.

A new query processing model for maintaining data temporal consistency in Wireless Sensor Networks

Maintaining data temporal consistency in a Wireless Sensor Network (WSN) has attracted the attention of many researchers. Nowadays, there are many applications based on WSN which require consistent and valid measures to reflect the current state of the environment being monitored, or to quickly respond to some critical situations. In this paper, we propose a new query processing mechanism, giving each node the possibility to control the temporal validity of data in order to reduce the amount of invalid data delivered by WSN. We also provide a new query model, by introducing new SQL clauses, that allows the user setting the data validity interval and assigning a deadline to each query. Finally, we discuss the performances of our query processing approach and its capacity to maintain the data temporal consistency through two scenarios. The first scenario consists of injecting a single query in the network, then receiving periodic responses from sensors, during a period set by the user. In the second scenario, the user sends periodic queries to sensors according to an epoch, thereafter, it waits for query responses. We concluded by discussing the ability of network to maintain the data temporal consistency, based on the average of query responses time and the amount of valid data transmitted across the network.

Modeling Real-Time Design Patterns with the UML-RTDP Profile

The use of design patterns in the real-time (RT) domain could bring many benefits. RT design patterns capture domain knowledge and design expertise. They improve the quality of RT software. These patterns contain a set of common elements and a set of variable elements known as variation points. Yet one of the main difficulties for representing RT design patterns is variability management. Thus, many questions arise: how to express variations? How to ensure the consistency of various views and avoid conflicts?