This manuscript describes my thesis work within the CMS experiment ofthe LHC collider. It presents both subjects on which I worked : a detectorsubject and a data analysis subject. The first subject, detector, is situated within the framework of the HL-LHCproject which plans an increase by a factor five the instantaneous luminosity.Such an increase and the preservation of the high level trigger (HLT)system, generate new constraints on the CMS detector. This is particularly true for the electronic trigger (L1) system. The CMS collaboration intends to implant this within tracker detector. Two methods are proposed : the stubmodules method and the associative memorie method. However, associativememorie method require a clever preselection of electric signals to be usable.The cluster width method is jointly proposed by Fabrizio Palla’s team (Pisa) and by Didier Contardo’s team (IPNL) to preselecte electric signals. The optimization of this method, in particular geometrical parameters of modules and thresholds applied for the electric signals preselection, is the result of myown work. It is detailed throughout the part III of this manuscript. The second subject concerns the analysis of data collected by CMS during 2011. This analysis has for objective to look for new particles in the spectreof anti-top top invariant mass. Due to its big mass, close to the electroweak symmetry breaking energy, the top quark plays a important role in numerous extensions of the Standard Model. The observation of narrow resonances in the spectre could be the sign of new particles. The analysis is subdivided into three parts, the selection of top anti-top events, the reconstruction of top anti top invariant mass and a statistical study to quantify the presence of new physics. The results of this work are limits on the cross-section productionof new particles. They are presented in the part IV of this manuscript.
