Donald Bren School of Information & Computer Sciences Towards End-to-end Data Exchange in the IoT Georgios Bouloukakis Joint work with Nikolaos Georgantas, Valerie Issarny, Pierre-Guillaume Raverdy, Patient Ntumba, Andrew Chio & Nalini Venkatasubramanian CPS & IoT seminar @ USC, Sept 2019 UC Irvine, Computer Science Department, USA
Enabling Data Exchange in IoT Smart Spaces Dynamic Heterogeneous “what is the occupancy of the room 2065? “ decrease the temperature of those rooms with occupancy above 50% of their capacity ?” Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 2
Most recent research IoT application Handling heterogeneity at the middleware-layer Handling heterogeneity at the application-layer Wrappers/ Functional Data model Dev. Code DeXMS Mediators Virtual Sensors semantics Automated placement of artifacts at the Edge Enabling a semantic vision of IoT spaces IoT Space Interaction Optimized User Execution Translation Model Constraints Placement Requests Plans Perf. evaluation of heterogeneous interactions Enabling Efficient Event Prioritization over SDN QoS analysis Analytical models Prioritization Statistical QoS Situational Analysis Semancis Awareness Simulated models Band. Allocation SDN Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 3
IoT heterogeneity at multiple layers Application layer Application layer Middleware layer Middleware layer Transport Layer Transport Layer Network layer Network layer Data Link layer (MAC & LLC) Data Link layer (MAC & LLC) Physical layer Physical layer IP CoAP Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 4
Middleware protocols in the mobile IoT DPWS CoAP MQTT ZeroMQ WebSockets …. Pub/sub Client-server Streaming …. Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 5
Heterogeneous interconnections in the mobile IoT Yannis Bluetooth beacon Providing common Providing common API abstractions API abstractions Bridging Relying on a middleware service bus Protocol Y Protocol X protocols Convergence to a Convergence to a single protocol single protocol How to enable interconnections in the mobile IoT ? Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 6
Our proposed solution Yannis Bluetooth beacon systematic solution to systematic solution to interoperability deployment Protocol X Protocol Y e.g. MQTT subscriber e.g. CoAP server • • pub/sub push-based functional semantics • • resource topic Automated synthesis of interoperability artifacts : • enables functional middleware-layer interoperability Automated placement and deployment at the Edge : • enables the deployment of interoperability artifacts at the Edge Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 7
Models for core interaction paradigms one-way Client – Service (CS) client server Tight Time & Space Coupling two-way sync or async one-way subscriber Publish-Subscribe (PS) publisher broker Time & Space Decoupling subscriber two-way stream Data Streaming (DS) consumer producer two-way stream Tight Time & Space Coupling one-way reader Tuple Space (TS) writer tspace Time & Space Decoupling taker two-way sync Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 8
Data eXchange (DeX) connector model Our generic connector defines 4 basic interaction types: one-way each interaction is represented as combination of post and get primitives two-way async two-way sync post and get primitives abstract CS, PS, DS and TS primitives two-way stream We rely on the DeX abstraction to introduce our middleware protocol interoperability solution Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 9
Our middleware protocol interoperability solution Data eXchange Mediator Synthesizer (DeXMS) 1,2 bluetooth Mediator 1 Mediator 2 mobile app beacon Convertion Convertion PUBLISH logic logic ACCEPT event data PS Protocol X common protocol DS Protocol Y common protocol Protocol X Protocol Y DeX connector A DeX connector B DeX connector C Mediator architecture: relies on DeX for automated Mediator synthesis Primitives & data conversion between the common protocol and the Things’ protocols A universal way to describe the Things ’ I/O required 1 G. Bouloukakis et al., FGCS, 2019 2 G. Bouloukakis et al., ICSOC, 2016 Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 10
Automated Mediator synthesis Generic Interface Description Language (DeXIDL) & Generic Mediator "protocol": “MQTT”, "operations": { Generic Mediator "operation_1": { "type": "stream", Generic "role": “consumer", one-way “interaction_type”:” one_way ” 2 logic "scope": “ get_occupancy", Mediator "input_data " : “ capacity,room" synthesizer } xmget 1 post } on_get REST (common protocol) DeX connector X DeX connector Y 3 4 Concrete Mediator Data DeX API conversion logic << Protocol Pool >> receive publish CoAP MQTT DPWS … MQTT REST Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 11
The Where and How Problem Where to place mediators: Cloud, Edge, and Fog Computing Obvious solution: The Edge and Fog Things push data to the cloud to be analyzed (e.g., 4k camera) Use artifacts at the Edge/Fog to filter these data Timeliness, data privacy, etc Edge Cloud IoT Work in progress: Systematic solution to automate the deployment of mediators at the Edge Utilize this solution to deploy mediators and other artifacts in the I3 platform Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 12
Mediators at the Edge Docker Ansible • • Delivers software in packages Automation tool to perform called containers. installation, maintenance, or • DeXMS provides mediators as monitoring operations. • Dockerfiles. Used to automate the installation • A Dockerfile produces a Docker of the smart space infrastructure image. as well the DeXMS service. DeXMS Ansible Kubernetes • • Triggers the DeXMS service to Container-orchestration system generate the required Mediator for automated deployment, containers. scaling, and management. • • Can be used to monitor nerworks Supports the automated and services. deployment of mediators. Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 13
The How Problem How to place mediators? Related Problem: Operator Placement Compute a “cost space” 1 to represent Things and Physical Nodes E.g., a smart building with heterogeneous Things Place mediators in an optimized manner M T N T T Criteria : distance, energy, bandwidth, latency, T T M T N availability, etc Optimization techniques 2,3 : constraint programming solvers, heuristics, linear programing, genetic N T T T programming, etc. 1 P. Pietzuch et al., ICDE, 2006 2 V. Issarny et al., ICDCS, 2019 3 A. Chio et al., ARM, 2019 Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 14
DeXMS novelty Bluetooth WiFi AP Mediator Mediator beacon MQTT CoAP CoAP REST MQTT REST MQTT Lightweight architecture Mediators employed only when necessary Any common protocol CoAP Support for any protocol classified under CS, PS, DS & TS Evolution support Camera Automated Mediator synthesis 75-96 % person-hours reduction when using DeXMS Work in progress: enabling application-layer data exchange 1 1 R. Yus et al., Buildsys, 2019 Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 15
Software artifacts and adoption DeXMS is part of the zefxis 1 platform (https://gitlab.inria.fr/zefxis): • Mediator generator: https://gitlab.inria.fr/zefxis/DeXMS • Eclipse plugin for defining Things’ DeXIDLs: https://gitlab.inria.fr/zefxis/DeX-IDL • Web console: https://gitlab.inria.fr/zefxis/IoT-web-console Demos: • Mediator generation: https://youtu.be/UgfM3810RS8 (ICSOC 2016) • Web console installation: https://youtu.be/IGjZ5u3QYOw (ICWE 2018) • Fire Detection scenario: https://youtu.be/SJeiqJkBhls (ICWE 2018) DeXMS is used as a core component in H2020 CHOReVOLUTION, UCI TIPPERS and Inria/UCI MINES projects. 1 zefxis in ancient Greek ( ζεῦξις ) means connection or coupling 16
Publications G. Bouloukakis, N. Georgantas, P. Ntumba, V. Issarny, "Automated Synthesis of Mediators for Middleware-layer Protocol Interoperability in the IoT", FGCS Journal, 2019. R. Yus, G. Bouloukakis, S. Mehrotra, N. Venkatasubramanian , “Abstracting Interactions with IoT Devices Towards a Semantic Vision of Smart Spaces”, ACM Buildsys, November 2019, New York, USA V. Issarny, B. Billet, G. Bouloukakis, D. Florescu, C. Toma , “LATTICE : A Framework for Optimizing IoT System Configurations at the Edge”, ICDCS 2019, July 2019, Dallas, Texas, USA A. Chio, G. Bouloukakis, C.H. Hsu, S. Mehrotra, N. Venkatasubramanian. “Adaptive Mediation for Data Exchange in IoT Systems”, 18th ARM Workshop 2019, Davis, CA, USA Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis 17
Thank you! Towards End-to-end Data Exchange in the IoT – Georgios Bouloukakis
Recommend
More recommend
