Navigation

Jump to main content
Summer School 2017
Program

Program

Schedule

 

 

Morning Session
(09:30-13:00)

Afternoon Session
(14:00-17:30)

Monday
28.8.2017

Result Analysis for 5G compliant Reliability Assessment
(Maciej Mühleisen, Ericsson, Aachen)
Robust Optimization
(Dr. Fabio D’Andreagiovanni, UMR
CNRS, Compiegne)
Tuesday
29.8.2017
Techno-Economics
(Prof. Sofie Verbrugge, Uni Gent)
Techno-Economics
(Prof. Sofie Verbrugge, Uni Gent)
Wednesday
30.8.2017
Workshop Workshop
Thursday
31.8.2017
Queuing Theory and Performance
Modeling
(Prof. Paul J. Kühn, Uni Stuttgart)

Queuing Theory and Performance
Modeling
(Prof. Paul J. Kühn, Uni Stuttgart)

Friday
1.9.2017
Dependability and Survivability
Quantification
(Prof. Poul E. Heegaard, NTNU
Trondheim)
Dependability and Survivability
Quantification
(Prof. Poul E. Heegaard, NTNU
Trondheim)

 

 

Presentations

 

(1) Result Analysis for 5G compliant Reliability Assessment (Maciej Mühleisen, Ericsson, Aachen)

 

Outline:

  • Definition(s) of Reliability

  • Mathematical Background

  • Confidence Intervals and Best Practice for Simulation and Experiments for them to be applicable

  • Assessing the Confidence of empirical Distribution Functions: the Limited Relative Error (LRE) Algorithm

Abstract:

5G vertical sectors eHealth, Factory-of-the-Future (Industry 4.0), Energy and Automotive all depend on highly reliable data communication. Monetary assets or even lives are at stake when communication fails in those domains. Therefore, those systems must be carefully designed to assure failure free operation for a defined “number of nines”, e.g. 99.999% of the time. For most vertical sectors authorities define certification processes where this degree of reliability must be proven.


Computer simulation, lab- and real-life experiments are, besides mathematical analysis, well established methods to assess and prove system performance. A key question to be answered in this tutorial is how to setup and conduct those simulations and experiments and how to analyze collected result. The LRE-Algorithm allowing to rate the confidence of collected results and providing an estimation on how long experiments need to be executed for a given reliability target will be presented and explored in a hands-on session.

 

 

(2) Robust Optimization (Dr. Fabio D’Andreagiovanni, UMR CNRS, Compiegne)

 

Abstract:

In this lecture, I will first review fundamentals of mathematical optimization for network design, focusing on Integer Linear Programming methods. Then I will proceed to present some remarkable applications of optimal network design arising in the field of telecommunications.

In the second part of the lecture, I will discuss how to deal with the presence of data uncertainty in optimal network design: a central assumption made in classical mathematical optimization is that all data of a problem are known exactly; however, many real-world problems involve uncertain data and neglecting these uncertainties may have dramatic effects (e.g., solutions supposed to be feasible may turn out to be infeasible and thus useless in practice). To tackle data uncertainty in network design, I will discuss the adoption of Robust Optimization (RO), a methodology for optimization under data uncertainty that has known a wide success in the last decade thanks to its accessibility and computational tractability. RO essentially takes into account data uncertainty in the shape of hard constraints that restrict the feasible set and maintain only robust solutions, i.e. solutions that remain feasible even when the values of the input data change. Among all the models proposed for RO, I will in particular focus on the famous Gamma-Robustness model by Bertsimas and Sim (2004) and its generalization called Multiband Robustness (Büsing and D'Andreagiovanni, 2012).

 

 

(3) Techno-Economics (Prof. Sofie Verbrugge, Uni Gent)

 

Outline:

  • Business model concept (supported by business model canvas by A. Osterwalder)

  • Life cycle cost breakdown

  • Cost models using different levels of abstraction

  • Customers segments and basic revenue models

  • Investment analysis

  • Impact of uncertainty

Abstract:

Techno-economic analysis aims at supporting decision makers by translating technological innovation into business opportunities and challenges. It involves detailed cost and benefit modelling and leads to a good insight in the economic viability of a certain engineering project from the perspective of the different stakeholders involved. Within this course will we focus on application of techno-economics with the ICT domain, including network deployment problems. The course consists of both a theoretical part and a hands-on in case study.

 

(4) Queuing Theory and Performance Modeling (Prof. Paul J. Kühn, Uni Stuttgart)

 

Outline:

Part 1: Introduction to Queuing Theory and System Analysis Methodology

  • Probability Laws, Random Variables and their Mathematical Description

  • Stochastic Processes, Renewal Theory, Non-renewal Processes

  • Markovian Queuing Systems and their Performance Analysis

  • Non-Markovian Queuing Systems and Queuing Networks

Part 2: Applications to Traffic Engineering and Performance Studies

  • Resource Dimensioning of Communication Trunk Systems for Circuit/Packet Switching

  • Single and Multi-Server Systems with Various Service and Priority Schedules

  • Communication Protocol Analyses

  • Modeling Network Function Virtualization and Software-Defined Networks

Abstract:

Queuing theory has proved as a powerful method to analyze systems with discrete stochastic arrival events and stochastic service times as, e.g., computer systems, communication networks, or communication protocols for specific operating modes. Modeling of such systems aim at a mathematical abstraction of system resources with respect of their usage by service requests under specific system operations as scheduling, routing, or resource assignments. Results are obtained which characterize the performance of such systems quantitatively as utilization factors, service delays, or blocking probabilities which are most important for an intended service quality defined by a Service Level Agreement (SLA). Applications of queuing theory and performance analyses are in "Traffic Engineering" how to size system resources and how to organize system operations to meet the negotiated or advertised SLAs. The methods of Queuing Theory are based on probabilistic descriptions of arrival and service processes and the analysis of the respective models by stochastic processes. The results can also be used within methods for system optimization, i.e., how to size the number of required resources and how to select and parameterize an appropriate system organization to meet the SLA for a minimum cost of investment or operation (Capex,Opex).

Course Material will be provided. It is based on comprehensive Lecture Notes of the author. Attendees should have some basic knowledge about probabilities. The presentation will focus on the principal understanding without going into details of methodologies but may be used as reference for applications or individual self-learning approaches. The application parts are complemented by some recently published papers. References on prominent standard text books will also be provided.

 

(5) Dependability and Survivability Quantification (Prof. Poul E. Heegaard, NTNU Trondheim)

 

Outline:

  • Dependability concepts and taxonomy

  • Static and dynamic models in dependability quantification

  • From dependability to survivability (incl. performability, (partial)recovery, forced failure)

  • Case studies

  • Survivability quantification exercises

Abstract:

The goal of this tutorial is to provide an introduction to the concept and definition of dependability and survivability, and approaches to model and quantify dependability and survivability in systems and networks. Examples are taken from mobile networks and virtual connection over an IP network as well as from smartgrid. Exercise will also be prepared to provide useful insight and experience with the use of an analytic-numeric software package (e.g., Sharpe), and a discrete event simulator (e.g., DEMOS/SImula).

Press Articles