In this thesis we present the first measurement in the CMS experiment of the top-antitop production cross section in the tau+jets final state. To perform this measurement we designed a specific trigger requiring the presence of four jets, one of them being identified as an hadronic tau. The performance of this trigger has been studied in this thesis. A dataset of 3.9 fb-1 was collected with this trigger and analyzed. At offline level we needed to apply a sophisticated tau identification technique to identify the tau jets, based on the reconstruction of the intermediate resonances of the hadronic tau decay modes. Another crucial point was the b-jet identification, both to identify the b-jets in the final state and to modelize the background using a data driven technique. The studies done on the b-tag algorithms along the PhD period are also presented with particular attention to the ”Jet Probability” algorithm. It is the algorithm for which I performed the calibration since 2009 as well as the one used to tag the b-jets from the top decays. A neural network has been developed to separate the top-antitop events from the W+jets and multijet backgrounds. A binned likelihood fit to the neural network output distribution is done in order to extract the signal contribution from the background. A detailed estimation of the systematic uncertainties on the cross section measurement is also presented. The result for the cross section measurement, σ(tt) = 156 ± 12 (stat.) ± 33 (syst.) ± 3 (lumi) pb, is in perfect agreement with the standard model expectation.
