DFG Research Center MATHEON
Mathematics for key technologies



Dear Colleagues,

we would like to draw your attention to the

MINI WORKSHOP

"Analysis of performance parameters in telecommunication networks"


taking place at TU Berlin (MA 415) on November 21, 2008.

The programme is as follows:
  • 9:00-9:40 Modeling and Performance Evaluation of Optical Buffers
    Ahmad Rostami (Telecommunication Networks Group, TU Berlin)
    Optical buffers are crucial to realization of optical switch/routers, which are considered to be one of the main building block of future Internet. Optical buffers differ from their electronic counterparts, in that they cannot offer random access property. Accordingly, conventional queueing theory techniques cannot be easily applied to this new kind of queues. In this talk, I will first give an introduction about operation of optical buffers, then introduce relevant analytical models and discuss open issues in this field.

  • 9:50-10:30 Achieving Reliable Communications with ARQ Protocols: Optimization and Performance Trade-offs
    Anastasios Giovanidis (Fraunhofer German-Sino Lab Mobile Communications (MCI))
    The lack of reliability in node to node communications due to the stochastic nature of the wireless channel can be resolved by the use of Automatic Retransmission Request (ARQ) Protocols, which repeat transmission of erroneous packets until success. The high costs of reliability include a waste of resources e.g. power and spectrum for erroneous transmissions, as well as a noteworthy increase in packet delay. The talk aims at presenting several dynamic optimization problems which highlight the tradeoffs between delay-dropping-power in single user scenarios as well as network problems where policies, which maximize the average arrival rate that can be reliably served, are derived based on stochastic optimization techniques.

  • 10:40-11:10 Probabilities in Networks
    Silke Meiner (TU Berlin)
    In this talk I will give an introduction on probabilities in stochastic queueing networks \& how to approximate probabilities for rare events, like: large queue sizes in a stable network or an unexpected evolution of queue sizes in a network with known initial queue sizes. Knowing such probabilities is important in performance analysis and resource planning for packet switched communication networks.
    Models for networks in this talk are such that the number of nodes and the topology is arbitrary and fixed. The queue sizes are discrete and evolve continuously over time. Examples are Jackson networks and some generalisations there of.

  • 11:10-11:30 coffee break

  • 11:30-12:00 Network Planning and Capacity Management in Multi-Cell Interference-Limited Wireless Systems
    Hans-Florian Geerdes (Matheon-Project B4)
    The wireless group at ZIB focuses on multi-cell interference-limited radio networks. In the talk, we will introduce our modeling approach, which aims at understanding multi-cell networks at cell level rather than at the individual user level. We will present models and optimization algorithms for UMTS radio network planning. In addition, we will outline some recent work on post-UMTS technologies such as HSDPA and LTE, and pose open questions, especially regarding the stochastic modeling of interference.

  • 12:10-12:40 On the performance of some access networks
    Frank Aurzada (Matheon Project B7)
    We investigate queueing systems that are of importance in optical network technologies. The talk focusses exemplarily on a particular model for access networks, namely, the so-called EPON, where several LANs (representing queues) communicate with a node in a larger network (that can be seen as the server) via a shared optical channel. The main feature of the system is a time lag in the communication that complicates the scheduling of the different queues by the server.
    After a short survey of the background from telecommunication, we describe our model assumptions and the quantities under investigation. Then we discuss the methods that are used in order to calculate or approximate these quantities. The single channel EPON system is by now understood rather well; and we describe several open problems related to the multi-channel case.

We look forward to seeing you.

Frank Aurzada and Michael Scheutzow, Project B7