Under CopyrightRichter, G.G.Richter2022-03-0707.08.20021999https://publica.fraunhofer.de/handle/publica/29029010.24406/publica-fhg-290290Die Flip-Tick Architektur (FTA) ist ein Entwurfs- und Konstruktionsmodell für skalierbare verteilte Systeme, die insbesondere ein a priori unbekanntes dynamisches Verhalten zeigen und/oder gravierende Unsicherheiten wie erhebliche Störungen, Ungenauigkeiten und eine nicht modellierte Dynamik aufweisen. Solche Systeme werden typischerweise dort eingesetzt, wo große Datenräume mit unbekannten Eigenschaften zu durchsuchen, komplexe Kontrollaufgaben in offenen Umgebungen durchzuführen oder reale Prozesse zu simulieren sind, für die geeignete mathematische Modelle fehlen. Der Bericht beschreibt die statischen und dynamischen Merkmale der Flip-Tick Architektur und ausgewählte Aspekte ihrer Implementierungen. Unter Verwendung von höheren Petrinetzen erstellte "Blaupausen" modellieren das generelle Verhalten von FTA-Systemen in ausführbarer Weise.The Flip-Tick Architecture (FTA) is a design paradigm for scalable distributed systems that in particular exhibit a priori unknown dynamic characteristics and/or involve severe uncertainties such as considerable disturbances, inaccuracies, and unmodeled dynamics. Such systems typically search large data spaces with unknown properties, carry out complex control tasks in open environments or perform simulations of real-world processes for lack of suitable mathematical models. The Flip-Tick Architecture provides a platform for implementing such systems on distributed hardware. In particular, it is designed for applications that feature numerous mathematical and heuristic algorithms which are embedded in a flexible connectivity structure and are executed in nested, sequential, concurrent or alternative cycles as parts of iterative or interactive processes. The two implementations of FTA aim at different application fields: realACT (actors operating in real environments) provides the basic functionality and infrastructure for designing and implementing distributed control systems, virtACT (actors operating in virtual environments) supports the realization of complex search and simulation systems. The report describes the static and dynamic features of the Flip-Tick Architecture and selected aspects of its implementations. Using high-level Petri nets formal "blueprints" are provided that model the generic behavior of FTA systems in an executable way.Contents S.5-6 1 Introduction S.7-11 2 Basics of operation: control and data flow S.12-21 - 2.1 Overview S.12-13 - 2.2 Functional units S.13-16 - 2.2.1 Actors and tags S.13-14 - 2.2.2 Ensembles and tagboards S.14-15 - 2.2.3 User's view of FTA systems S.15-16 - 2.3 Relationships between functional units S.16-18 - 2.4 System operation S.18-21 3 Cycles and their synchronization S.22-34 - 3.1 Actor and ensemble cycles S.22-24 - 3.2 Tagboard cycles S.24 - 3.3 Observation-based and notification-based synchronization S.24-26 - 3.4 Synchronization of actors S.27 - 3.5 Observation-based synchronization of tagboards and ensembles S.27-28 - 3.6 Notification-based synchronization between tagboards and ensembles S.28-31 - 3.7 Notification-based synchronization between ensembles S.31-34 4 The FTA application model S.35-56 - 4.1 Submodels and subdiagrams S.35-38 - 4.2 Local states S.38-42 - 4.2.1 Colorsets S.38 - 4.2.2 Places S.38-42 - 4.3 System start-up S.43-44 - 4.3.1 Creating an FTA system S.43 - 4.3.2 Initial state S.43-44 - 4.4 Actor cycle: basic operation S.44-46 - 4.5 Actor cycle: system reorganization S.46-48 - 4.6 Ensemble and tagboard operation with notification-based synchronization S.48-49 - 4.7 Ensemble loop with notification-based synchronization S.49-53 - 4.8 Tagboard loop with notification-based synchronization S.53-54 - 4.9 Ensemble loop with observation-based synchronization S.54 - 4.10 Tagboard loop with observation-based synchronization S.55 - 4.11 Combination of observation-based and notification-based synchronization S.55-56 5 The realACT implementation model S.57-76 - 5.1 Main offices and subsidiary offices S.57-59 - 5.2 Submodels and subdiagrams S.59-60 - 5.3 Local states S.60-66 - 5.3.1 Colorsets S.60 - 5.3.2 Places S.60-66 - 5.4 System start-up S.66-68 - 5.4.1 Creating an FTA system S.66-67 - 5.4.2 Initial state S.67-68 - 5.5 Actor cycle: communication and coordination S.68-70 - 5.6 Distributed ensemble loop without synchronization S.70-71 - 5.7 Distributed tagboard loop with synchronization S.71-72 - 5.8 Refinement of SR- boxes S.72-75 - 5.9 Refinement of RQ -boxes S.75-76 6 Conclusion and related work S.77-82 Appendix S.83-102 - A Glossary S.83-88 - B From elementary to high-level Petri nets S.88-93 - B.1 High-level places S.89-90 - B.2 High-level links S.90-91 - B.3 Summary S.91-93 - C Formal definition of Petri nets for condition/event modeling S.93-102 - C.1 Introduction S.93-95 - C.2 Definitions S.95-102 Bibliography S.103-105 - Loose-leaf part S.105enArchitekturAkteurRobotikKontrollsystemsimulationIntelligenzarchitectureagentactorroboticcontrol systemsimulationintelligence004005006report