Jump to main content
Summer School 2015


Summer School on Network Performance Evaluation and Optimization, 20.6.- 23.6.2015






Morning Session (09:30-13:00)

Afternoon Session (14:00-17:30)



Performance Evaluation of Communication Networks

(Prof. A. Timm-Giel, Dr. T. Zinner)

Performance Evaluation of Communication Networks

(Prof. A. Timm-Giel, Dr. T. Zinner)



Performance Evaluation of Communication Networks

(Prof. A. Timm-Giel, Dr. T. Zinner)

Performance Evaluation of Communication Networks

(Prof. A. Timm-Giel, Dr. T. Zinner)



Network Coding

(Prof. F. Fitzek)

Network Reliabilty

(Dr. C. Mas Machuca)



Network Calculus and its Application to Call Acceptance Control

(Prof. U. Killat)

Application of Meta-Heuristics in Real World Network Planning

(Dr. S. Wolf)




(1) Performance Evaluation of Communication Networks

(Prof. Andreas Timm-Giel, TUHH and Dr. Thomas Zinner, Uni Würzburg)



  • Basics of Stochastic

  • Introduction to Queueing Theory

  • Discrete Event Simulations

  • Comparison of simulative and analytical approaches

  • Analysis of results

  • Advanced topics in Queueing Theory

  • Simulation of realistic communication networks


This two day course introduces into performance evaluation of communication networks by Queuing Theory and Discrete Event Simulations. The differences, opportunities and limitations of these approaches are outlined by using common examples. Hands-on experiments are used for further and in-depth understanding of these methods and to identify drawbacks and advantages of the individual methods. The course ends with an outlook and examples on how to apply these methods to analyze more realistic network scenarios.

The course requires a high level of interaction and discussions and includes hands-on programming examples, implemented by the students together with the tutors on the fly. Programming language will be Matlab/Octave.



(2) Network Reliability (Dr. Carmen Mas Machuca, TU München)



  • Introduction

  • Failure classification and examples

  • Failure models

  • Network reliability parameters (e.g. redundancy, reliability, availability) and calculation methods

  • Survivability

    • at the transmission layer

    • at the logical layer

  • Conclusion


The continuous development of new and more bandwidth_hungry services and the fast increase of terminals are driving a fast evolution at any telecommunications network section. This fast evolution includes obviously optical core networks, which have link capacities of more than 100Tbps. The advantage of having so much capacity in a link is counteracted by the huge amount of information that may be lost in case of a failure occurs.

Hence, these networks should be as reliable as possible and offer protection and restoration mechanisms. First, a study of the network reliability should be performed in terms of redundancy, failure probabilities, connection availabilities, etc. Secondly, an efficient network design should be carried out in order to be able to cope with as many connection as possible (to maximize revenues) while coping with the expected reliability and use as less resources as possible (to maximize the number of future connections that can be established in the network). There are different alternatives which will be presented in the course.

This course will consist of two parts. The first part will give an overview of the most important concepts in network reliability as well as different methods for their calculation. Different failure models will be also introduced and the derivation of some reliability parameters expressions will be presented. The second part of the course will cover different survivability mechanisms at the transmission layer and the logical layer.



(3) Network Coding (Prof. Frank Fitzek, TU Dresden)



  • Introduction

  • Theoretical foundations of Network Coding

  • Application to networking problems

    • data gathering in sensor networks

    • routing in wireless mesh networks

    • peer-to-peer networking

    • content distribution

    • streaming applications

  • Implementation aspects


The tutorial provides an introduction to the rapidly growing research area of network coding focusing on use cases such as communication networks and storage. Network coding allows intermediate nodes in a network to manipulate data, for example by sending out packets that are combinations of previously received packets instead of simply forwarding them. For most practical purposes, these manipulations are linear operations over elements of a finite field. The initial theoretical results on network coding were followed by a wealth of applications in a number of different areas that show that the theoretical insights can be translated into practical gains.

The tutorial is divided into three parts. The first part provides the participants with the theoretical tools necessary to understand the field of network coding and focuses on the underlying algebraic principles. We do not assume any prior knowledge of algebra or optimization. This part of the tutorial also introduces distributed randomized network codes and discusses their properties.

The second part of the tutorial gives an overview of the different application areas and discusses the types of networking problems that are amenable to network coding (and those that aren't). In particular, it covers practical algorithms for data gathering in sensor networks, routing in wireless mesh networks, peer-to-peer networking and content distribution, streaming applications, etc. We illustrate how network coding can be used to increase throughput and robustness as well as reduce storage requirements, delay, and energy consumption.

Finally, we discuss implementation aspects in real-world systems. Such systems may range from core network routers all the way down to mobile phones and tiny sensor nodes. The constraints imposed by these devices in terms of available memory and computing power may differ by several orders of magnitude. As a consequence, the encoding and decoding algorithms need to be carefully adapted to the specific problem at hand. As an example, the size of the finite field for the coding operations has an impact on network coding efficiency, but also on the encoding and decoding complexity. Coding operations may be sped up substantially through the use of specialized hardware, as evidenced by the successful implementation of network coding on Graphics Processing Units (GPUs). The energy consumed by the coding operations is of particular importance on mobile devices and needs to be considered to avoid offsetting the energy gains offered by network coding.



(4) Application of Meta-Heuristics in Real World Network Planning (Dr. Steffen Wolf, Detecon Dresden)



  • Network plannning example: Facility Location Problem

  • Initial approach: solution by local search

  • Refined approach: solution by meta-heuristic methods

    • simulated annealing

    • evolutionary/genetic local search

  • Lessons learned: challenges to realize the solution in real world scenarios

  • Latest (pragmatic) approach: user interaction during optimization run


An interesting set of Meta-Heuristic approaches to NP-hard optimization problems is based on local search. In this tutorial the basic outline and the motivation behind Evolutionary Local Search and Simulated Annealing will be presented. A simple facility location problem from fiber network planning will serve as an example problem: Here, one of the first steps is the placement of distribution points and the assignment of the terminals to one distribution point each. When trying to minimize the cost for these connections, the problem quickly turns out to be an NP-hard optimization problem. The application of meta-heuristic methods to tackle this problem will be outlined and some challenges regarding the realization of the obtained solutions in real world scenarios will be adressed. As experienced planners do not like to hand over all decisions to an optimization black box, a pragmatic approach is to give them as much influence on the optimization steps as possible.



(5) Network Calculus and its Application to Call Acceptance Control (Prof. Ulrich Killat, TUHH)



  • Introduction

  • Min-Plus-Algebra

  • Deterministic Network Calculus

  • Stochastic Network Calculus (SNC)

  • Effective Envelope and Deterministic Service Process

  • SNC with Effective Bandwidth and Effective Capacity

  • Call Acceptance Control (1): Single node criterion

  • Call Acceptance Control (2): Application of Martigale Theory to a Chain of Nodes

  • Conclusions


To avoid the complexity resulting from continuously changing arrival processes in a chain of queues, Network Calculus bounds stochastic processes by envelopes thereby simplifying the analysis of backlogs and delays. Unfortunately the envelopes become less tight bounds when moving from one node to the next. To avoid this problem one can link the calculation of envelopes for the arrivals at each node to the concept of effective bandwidth and can then profit from the (approximative) invariance of the latter.

This seminar gives an introduction to the concepts of Network Calculus and then presents its extensions in terms of effective bandwidth, effective capacity and Martingale Theory. The effectiveness of these concepts is demonstated by the results of call acceptance control derived from simulation experiments.


Press Articles