Keynote Speakers
Professor Edoardo Amaldi
website
Politecnico di Milano
Italy
Title: On energy-aware and fair routing in telecommunication networks
)
Abtract
Energy efficiency and fair resource (bandwidth) allocation are important issues in telecommunication networks. We address two relevant problems for IP (Internet Protocol)
networks: routing subject to max-min fair flow allocation and energy-aware traffic engineering.
Max-min fairness (MMF) is a widely used notion of fairness in several fields. Informally, an allocation of flow (bandwidth) in a network is max min fair if there is no way to give more flow to any origin-destination pair without decreasing the allocation to a pair with lesser or equal flow. The MMF paradigm is of substantial interest for IP networks since it approximates TCP flow allocation when the traffic flows are elastic and the rates are adapted based on resource availability. While in previous work the MMF paradigm has been adopted as a routing objective, we take the network operator point of view and consider it as a requirement of a more general traffic engineering problem. Given a network topology with link capacities and a set of origin-destination pairs, we have to select a single path for each origin-destination so as to maximize the network throughput, assuming a MMF flow allocation. We compare mixed integer programming formulations, describe a column generation algorithm and report some computational results.
Internet energy consumption is a critical issue due to the exponential traffic growth and the rapid expansion of communication infrastructures worldwide. Open Shortest Path First (OSPF) is one of the most commonly used intra-domain internet routing protocol. Traffic flow is routed along shortest paths to the destination and the weights of the links can be changed by the network operator. We consider the energy-aware traffic engineering problem of switching off (putting in sleeping mode) network elements (links and routers) and of adjusting the link weights so as to minimize the energy consumption as well as a network congestion measure. We describe a three-phase MILP-based heuristic for tackling this multi-objective problem with priority (first minimize the energy consumption and then the congestion measure), which exploits the IGPWO heuristic by Fortz and Thorup (2004) for optimizing the link weights so as to minimize a network congestion measure. We report and discuss computational results for real network topologies and different types of traffic matrices.
Riadh Zorgati
EDF R&D, Department OSIRIS (Optimization, SImulation, Risk and Statistics) 1, Avenue du Général de Gaulle 92141 Clamart Cedex (France) riadh.zorgati@edf.fr
Title: Some Challenging Problems in Energy Management
)
Abtract
Electricité de France (EDF), the main French electricity board, is one of the European leaders in the
energy sector and a major actor of the electrical nuclear industry in the world. EDF has activities of production, transport and distribution, commerce and trading on a world wide scale. EDF has to face numerous and various real-life problems with very challenging associated scientific problems in different fields. In this presentation, we will focus on energy management and review some open key problems arising in this field, involving operational research and optimization in uncertain setting.
The group EDF has an important, diversified, flexible and efficient set of production (nuclear, fuel, gas, coal, hydro). Energy management of such a portfolio, in coordination with financial assets, aims at taking optimal decisions from a long term to the short term for determining the main characteristics of the production portfolio for the next twenty years, scheduling the outages of nuclear power plants for refuelling and maintenance, managing hydro stock and supply contracts, mastering physical risk of supply shortage and financial risk on markets, and finally producing electricity in order to satisfy the demand at minimal cost.
The underlying Energy Management Optimization Problems (EMOPs) are characterized by challenging key features such as the stochastic nature of data, the nature of the decision variables of the problem (real, integer, binary), and the huge instance sizes. Hence EMOPs cover a wide range of situations ranging from linear or fully quadratic stochastic optimization programs with mixed variables to nonlinear, nonconvex programs of huge size. Resolution of these optimization problems is often very challenging, especially when reasonable on-time solutions compatible with industrial constraints are requested. Currently, none of known approaches in optimization can provide efficient solutions taking into account all of these features and satisfy all industrial requirements.
The presentation will give the industrial context of energy management, will focuse on some open emerging problems in this field and will describes some related challenging scientific problems. A focus on practical problems encountered in optimization in an uncertain setting will be made.