Thomas Birkhölzer

Numerical nonlinear regulator design

This paper considers the state regulation problem for nonlinear plants P with initial conditions in a prespecified region G. A computer implementable algorithm is presented, which is theoretically ensured to yield a practically asymptotically stabilizing feedback controller from G, if one exists. Since the plant P is represented by a discrete time mapping, which is only assumed to be continuous, the approach is of wide applicability for plants of moderate order. >

An interactive software management simulator based on the CMMI framework

A software-producing organization is a complex system. An interactive simulation that allows playful exploration and experimenting can enhance the understanding of such complex systems. Therefore, a simulator concept is presented that lets its user act in the role of a software-producing organizations top-level manager. In this role, business measures must be monitored and investment decisions need to be made. The model is developed as a feasibility investigation with the premise of maximal abstraction, using concepts of the Capability Maturity Model Integration (CMMI) and balanced scorecards: 15 process areas deduced from CMMI serve as states and investment fields and 27 business performance indicators form the outputs. Simulations of different scenarios illustrate that the model is capable of reflecting underlying strategies for advancing or maintaining an organizations processes. Moreover, the iterative and interactive investment approach and the graphical and historical result presentation can give valuable insight into complex process dependencies. Therefore, the simulator can serve as a tool to playfully enhance the appreciation of software engineering practices. It can be used by all stakeholders involved in software development to better understand the various aspects of software engineering, their interdependencies, and the resulting complex dynamics of the software-producing organizations operations. Copyright

Deriving a valid process simulation from real world Experiences

This paper presents a systematic approach to develop and configure a process simulation model that relates process capabilities to business parameters in order to support process improvement projects within Siemens. The research work focuses on the systematic set up of a validated and acknowledged model that matches the companys process improvement needs by involving experts to adapt an existing mathematical framework and simulation application. The methodology consists of three complementary steps: An approved conceptual model is used as structural skeleton, quantitative parameters are derived by a prospective expert survey, and final adaptation and customization is facilitated in order to be useable for process experts themselves (instead of model developers).

Knowledge Acquisition and Automated Generation of Bayesian Networks for a Medical Dialogue and Advisory System

Probabilistic models such as Bayesian networks [6] are well suited for medical decision support and are the basis of many successful applications [1],[3],[4],[8],[9],[10]. Bayesian networks provide a rigorous and efficient framework for inference, i.e. for calculating the probability of each stochastic variable given a set of observations. However, knowledge acquisition and generation of the network are still demanding tasks when large medical domains have to be modelled.

SimSWE - a library of reusable components for software process simulation

SimSWE is a library of components for modeling and simulation of software engineering processes. It consists of a generic, implementation independent description of the components and a reference implementation using the MATLAB® / Simulink® environment. By providing ready-to-use building blocks for typical functionality, the library should facilitate and ease the use of simulation to analyze software engineering issues. The goals are to provide a collection of reusable components within an open and evolvable framework to help practitioners develop simulation models faster and enable researchers to implement complex modeling and simulation tasks. Currently, the library contains more than thirty components. It can be used under the LPGL license. The paper presents the current status of SimSWE to encourage its use and participation within the research community.

A Framework for Systematic Evaluation of Process Improvement Priorities

Each process improvement project must prioritize and select the areas or aspects to focus on. Moreover, each such selection needs to be communicated, motivated, and justified to all stakeholders (management as well as participating or affected parties). For both challenges, a conceptual framework and tool set is presented in this paper which is based on a quantitative model of the relations between improvement efforts and outcomes. These relations are represented as weighted dependency graph with extensions for time dynamics as well as uncertainties. The graph can be used for a goal-driven search of strategies as well as for simulation analysis of specific scenarios. The development of the graph basically requires a systematic documentation of the expectations or experiences in the respective context. The framework is outlined at the example of agile process fragments based on a knowledge base of evidential data.

Applications of a generic work-test-rework component for software process simulation

Software process simulation can be a valuable support for process analysis and improvement provided the respective model development can focus on the issues at hand without spending effort on basically modeling everything from scratch. Rather, the modeling groundwork should readily be available as building blocks for (re)use. In the domain of software engineering, the work-test-rework cycle is one of the most important reoccurring patterns - at the level of a individual tasks as well as at the level of process phases or projects in general. Therefore, a generic reusable simulation component, which captures and represents this pattern, has been realized as part of the SimSWE library. This WorkTestRework component is designed such that it provides comprehensive adaptability and flexibility to be used in different simulation scenarios and process contexts. This paper introduces the component and substantiates this claim by modeling typical, but very different process scenarios based on this component.

Customized Predictive Models for Process Improvement Projects

A methodology is presented to quantitatively model the expected relationships between investments in process improvements and improvements in business measures. Such a predictive model can be used as an auxiliary in process improvement planning in addition to established models like CMMI. Different from a generic model like CMMI, the proposed methodology allows for creating a fully customized model focusing on the context or product at hand. To manage the inherent parameter uncertainty of quantitative modelling of software processes a novel approach in this context is used by explicitly handling the parameter variations using interval arithmetic. The paper outlines the methodology and presents results from a study at Siemens.

Goal-driven evaluation of process fragments using weighted dependency graphs

Software process improvement needs guidance in proposing, assessing, and selecting a right set of steps and measurements. Often, a process model or a specific methodology is used for this purpose, e.g. the V-Modell, RUP, or agile process methodologies. Each such model or methodology consists of certain activities, procedures, or methods, i.e. a set of process fragments. Debates usually focus on the choice among these models or methodologies as a whole, but not on the specific, quantitative contributions of the distinct process fragments with respect to the particular goals of the organization. Such a simplification misses important chances for adaptation and motivation in diversified system development environments. Therefore, a conceptual framework and tool set is presented in this paper to search for sets of process fragments, which are able to meet actual objectives in an optimal way.

Handling the Complexity of IT-Environments with Enterprise Architecture

The relation of Enterprise Architecture to the traditional architectural roles of software engineering (platform architect, component architect as defined below) is similar to the relation between “city planning” and “building blueprints” in the building domain. There is a difference in scale, scope, necessary competences and methodologies. It is the intent of this article to stimulate this understanding summarized in the following two theses: 1. Enterprise Architecture is a necessary and distinct architectural role. Successful large-scale system as well as virtual enterprise development requires appreciation and inclusion of this role in the IT-engineering process. 2. Enterprise Architecture means cross-system coordination with similar stakeholders outside the own business ownership. This distinguishes Enterprise Architecture from traditional architectural roles and implies distinct tasks, methodologies, and required skills.