# 10th of May, 2016: Network Calculus with Compact Domains

**Speaker**: Kai Lampka, Department of Information Technology, Uppsala University**Title:**Network Calculus with Compact Domains**Abstract:**Real-time Calculus (RTC) aims at tightly bounding end-to-end flow delays and buffer requirements for distributed real-time systems. Workload is modelled per event stream and service is modelled per component. RTC then transforms this system model into a sequence of operations deriving the above performance measures. For accurate analysis results, an accurate system model is required. This can be achieved with piece-wise linear, pseudo-periodic functions. However, these functions impose a high computational effort on RTC. Applying an operation usually results in a function whose periodic tail ranges over the hyperperiod of both input functions; consequently, computational effort increases fast and vastly in the sequence of operations.Finitary RTC counteracts this problem by computing a bound on each function’s prefix, i.e., the finite part that is ultimately required by the RTC analysis. This solves the problem of the unbounded growth of hyperperiods, yet, the current approach has a major drawback. It can only be applied to one specific RTC analysis, namely the Greedy Processing Component (GPC) oriented analysis, which is known to result in rather inaccurate delay bounds.In this paper, we provide an improved finitary RTC that shrinks the prefixes of the GPC-oriented analysis and is applicable to more accurate RTC methods for delay bounding.