Afanasov14flynet slides
- 1. Context-Oriented Adaptation in Cyber-Physical Systems Mikhail Afanasov*, Luca Mottola*† and Carlo Ghezzi* *Politecnico di Milano (Italy), †SICS Swedish ICT
- 3. Problem • CPSs are intimately tied to the real world – Multiple environment dimensions at hand – Continuously changing environment • The software must adapt • Missing design and programming support
- 5. Problem module ReportLogs{ uses interface Collection; uses interface DataStore; } implementation { int base_station_reachable = 0; event msg_t Beacon.receive(msg_t msg) { if (!acceleromenter_detects_activity()) return; if (call Battery.energy() <= THRESHOLD) return; base_station_reachable = 1; call GPS.stop() call BaseStationReset.stop(); call BaseStationReset.startOneShot(TIMEOUT); } event void BaseStationReset.fired() { base_station_reachable = 0; } event void ReportPeriod.fired() { switch (base_station_reachable){ case 0: call DataStore.deposit(msg); case 1: call Collection.send(msg);} } }
- 6. Solution • Context • Context-Oriented Programming (COP) • Language-independent design concepts • Programming support
- 7. Solution context Unreachable { uses interface DataStore; } implementation { layered command void report(msg_t msg){ call DataStore.deposit(msg);} } context group BaseStationG { layered command void report(msg_t msg); } implementation { contexts Reachable, Unreachable is default, MyErrorC is error; components Routing, Logging; Reachable.Collection -> Routing; Unreachable.DataStore -> Logging; } context Reachable { uses interface Collection; } implementation { layered command void report(msg_t msg){ call Collection.send(msg);} } Functionality are decoupled; implementations are individually simpler
- 8. Context-oriented Programming • Layered functions change the behavior depending on the context
- 9. Context-oriented Programming Java Python Erlang Lisp Objective-C • Layered functions change the behavior depending on the context
- 10. Design Concepts • Context represents a single environmental situation • Context group as collection of environmental situations sharing the same characteristics Context group Context Context
- 11. Patterns: Behavior control Provides a context- dependent functionality send(Msg* msg)
- 14. Verification: Initial Model • Simple transition: S0={…, …, …, i, …} -> S1={…, …, …, j, …} • Dependency: S0={…, n, … , i, …} -> S1={…, n, … , j, …}
- 16. ConesC • Context-Oriented extension of nesC • Enable layered functions (commands) in nesC • Redefine component and configuration Core features: • Define context group • Layered function • Implementation variations in contexts https://github.com/muxanasov/ConesC
- 17. Evaluation: Coupling W. Stevens et al. Classics in software engineering. Chapter Structured Design. 1979. Strongest Weakest
- 19. Evaluation: Run-time Overhead 0 5 10 15 20 25 30 Wildlife tracking Adaptive stack Smart-home MCUcycles Context transition overhead Function call overhead MCU Overhead Memory Overhead 0 0.5 1 1.5 2 2.5 3 Wildlife tracking Adaptive stack Smart-home % Binary overhead RAM overhead Turning an LED on is 8 MCU cycles Max 2.5%!
- 20. Conclusions and Ongoing Work • Context as a CPS programming concept: – Language independent design – Programming support: ConesC, IDE, Model-checker • Key results: – Easier to maintain and to understand – Verification against environment evolutions – Negligible performance overhead • Current work: – Extending on micro & nano aerial drones
- 21. Other Activities @POLIMI • Programming Systems for Coordinating Drones • Example: – L. Mottola et al. “Team-level programming of drone sensor networks”, in ACM SENSYS 2014 – Real-deployment for aerial photogrammetry in archaeological sites (Aquileia, Italy) – Currently being extended to indoor scenarios for tiny devices “Domus dei putti danzanti” youtu.be/PPDGO-jc0It
- 22. Other Activities @POLIMI • Integration and remote control – Mission-level service-oriented interfaces (REST) – Remote-control service oriented interfaces (CoAP) • Flight control loops – Reactive programming techniques – Testing and verification
- 23. Questions? 0 5 10 15 20 25 30 Wildlife tracking Adaptive stack Smart-home MCUcycles Context transition overhead Function call overhead
Editor's Notes
- #3: Mobile or not, battery level
- #4: Mobile or not, battery level
- #5: Atteched to collars of animals, what makes them mobile as drones. They are buttery operated as well, and, with respect to drones, situations are changed even fastly then in this example.
- #6: nesC is component-based language, endeed, but there are too much dimentions, changing independently and the functionaluty is relies on these dimentions
- #9: Enable it on the r-c platforms, limited multythreading and no memory protection… as in high langs
- #11: We claim that our design concepts inspired by COP can significantly simplify the software developing process for WSN. It is worth noticing, that these concepts are language independent and can be implemented in any of them.
- #20: Layered function call
- #22: The team-level programming model provides a middle ground between programming individual devices and swarm programming, enabling the specification of coordinated actions based on global states (unlike swarm programming), but still without resorting to individual addressing (like when programming individual devices)
- #23: We built abstractions and prototypes allowing the integration of aerail drones into larger processes, also using graphical interfaces. The picture shows an example of a composition of individual processing blocks that instruct the drone to take pictures in a given area and then post-process the pictures looking for a certain pattern as well as uploading them on Flickr.