In Chapter 1 we present our contributionto the scheduling of real-time systems on multiprocessor platforms. We start thechapter by providing an exhaustive list of all results we proposed on this topic andthen we detail two contributions: the predictability results and the schedulabilityanalysis. We present the proofs of these results as their details are important.The first result on predictability is a corrected and extended version of previousresults from [38]. The results on schedulability analysis provide a methodology forfeasibility tests based on intervals.In Chapter 2 we present our contribution to the definition and the analysis ofprobabilistic real-time systems. We start the chapter by providing an exhaustivelist of all results we proposed on this topic and then we detail two contributions.The first contribution concerns the proposition of probabilistic worst-case reasoningwithin response time analysis of systems with multiple probabilistic parameters.The second contribution concerns the estimation of probabilistic worst-case executiontimes and its relation to rare events theory. It indicates that the real-timecommunity should use the mathematical results on the tails of the probability distributions.In Chapter 3 we present the perspectives of our results and the vision of howthese results will modify the real-time community that needs to face data delugeand increased connectivity with less critical systems that will coexist within onelarger system built following a mixed-critical model.