Nicolas Markey

Résumé

Energy consumption problems (e.g., electric vehicles charging) are very related to communication problems. The "water-filling" algorithm can be used, but it is not robust w.r.t. the noise of the "non-flexible", i.e. not-controllable, energy consumption forecasting. We propose a robust algorithm using the probabilistic model-checker PRISM, to exploit the idea of discretizing the consumption action (as for a modulation to small constellation) and the dynamic structure of the problem in a Markov-decision-process model.

@inproceedings{gretsi2017-GBBLM,
  author =              {Gonz{\'a}lez, Mauricio and Beaude, Olivier and
                         Bouyer, Patricia and Lasaulce, Samson and Markey,
                         Nicolas},
  title =               {Strat{\'e}gies d'ordonnancement de consommation
                         d'{\'e}nergie en pr{\'e}sence d'information
                         imparfaite de pr{\'e}vision},
  booktitle =           {{A}ctes du 26{\`e}me {C}olloque du {G}roupe
                         d'{\'E}tudes du {T}raitement du {S}ignal et des
                         {I}mages ({GRETSI}'17)},
  acronym =             {{GRETSI}'17},
  year =                {2017},
  month =               sep,
  abstract =            {Energy consumption problems (e.g., electric vehicles
                         charging) are very related to communication
                         problems. The "water-filling" algorithm can be used,
                         but it is not robust w.r.t. the noise of the
                         "non-flexible", i.e. not-controllable, energy
                         consumption forecasting. We~propose a robust
                         algorithm using the probabilistic model-checker
                         PRISM, to exploit the idea of discretizing the
                         consumption action (as~for a modulation to small
                         constellation) and the dynamic structure of the
                         problem in a Markov-decision-process model.},
}