Scalable Differential Analysis of Large Process Algebra Models

http://www.dcs.ed.ac.uk/home/jeh/

http://www.pst.ifi.lmu.de/Personen/team/tribastone

This tutorial is concerned with the performance evaluation of hardware/software systems using ordinary differential equations which approximate large-scale continuous-time Markov processes derived from models described with the stochastic process algebra PEPA.

The tutorial is divided into three parts. The first part illustrates the main theoretical results. The second part gives an overview of a software tool – the PEPA Eclipse Plug-in – which supports the differential analysis of PEPA. In the last part, this approach is related to other efficient analysis techniques in the literature. In particular, a comparison against layered queues is presented.

Reliability Models

Performance modeling researchers typically use Markov models associated with their analysis of computer systems. This tutorial introduces additional probabilistic models (e.g., competing risks, proportional hazards) and the associated statistical models that are commonly used in reliability engineering and biostatisics to complement the Markov models.

http://www.math.wm.edu/~leemis/

From Systems to Components: Constructive Methods for Product-Form Solutions

Product-form models are a class of Markovian models whose steady-state solution can be computed efficiently thanks to a separable equilibrium probability distribution. They first appeared in queuing theory with the well-know results of Jackson, Gordon Newell, and BCMP theorem. Many research efforts have been devoted to the identification of new product-form models both for queueing networks (e.g. G-networks) and other formalisms (e.g. for Markovian process algebra or for stochastic Petri nets etc.).

In this tutorial, we present recent results in which the product-form solutions are derived efficiently via analysis of the underlying Markov process and its reversed. This methodology has successfully been applied to prove both well-known product-form solutions and new ones. Since the results are formulated purely in Markovian setting they may be applied also for non-queueing models. The tutorial will be of great interest to an ample audience, well beyond researchers working on queuing theory and it aims to give both a theoretical overview on the results mentioned above and their practical applications to modeling.

http://www.doc.ic.ac.uk/~mgv98/

http://www.dsi.unive.it/~marin/

Quantitative Evaluation of Security Metrics

Making sound security decisions when designing, operating, and maintaining a complex system is a challenging task. Analysts need to be able to understand and predict how different factors affect overall system security. During system design, security analysts want to compare the security of multiple proposed system architectures. After a system is deployed, analysts want to determine where security enhancement should be focused by examining how the system is most likely to be successfully penetrated. And when several security enhancement options are being considered, analysts would like to evaluate the relative merits of each.

In each of these scenarios, quantitative security metrics should provide insight on system security and aid security decisions. This tutorial provides a survey of existing quantitative security evaluation techniques, and describes new work we at Illinois are doing in this regard.

http://www.crhc.illinois.edu/Faculty/whs.html