Post on 29-Nov-2014
description
transcript
Overview: Data-driven IoT™ Platform
Copyright © 2014 Oleg Puzanov. All rights reserved.
About Us: Project Team
• M2M and IoT Technologist, Software and Hardware Geek • Director, Software Engineering @ Cogniance: www.cogniance.com • IoT Primer blog: www.iotprimer.com • 12+ years in ICT domain projects, large experience with US and
EMEA markets for embedded, web and cloud projects.
Oleg Puzanov
• Senior Expert in embedded systems, FPGA/ASIC design • Senior Systems Engineer @ Cogniance: www.cogniance.com • Founder and Managing Director @ Unicore Systems:
www.unicore.co.ua
Oleg Uzenkov
Copyright © 2014 Oleg Puzanov. All rights reserved.
Introduction
Copyright © 2014 Oleg Puzanov. All rights reserved.
Concept: Data-driven IoT™ Platform
Copyright © 2014 Oleg Puzanov. All rights reserved.
Vertical Applications
Connected Fitness/Telecare NGN Telecom Smart Grid and Utilities
Smart ParkingConnected Car/Vehicles Smart Agriculture
Connected Retail Smart Home/Energy Environmental Safety
Copyright © 2014 Oleg Puzanov. All rights reserved.
High-level Architecture
IoT-ACUI
IoT-SCR
IoT-GW
IoT-SCR
IoT-MD
IoT-FGF
IoT-DPF
IoT-DPF
IoT-SCR
IoT-MD
Location Field Networks
RFID WSAN
BluetoothWLAN IP WAN
XMPP
MQTT
XMPP
MQTT
IoT-DPF
IoT-BIG
IoT-WEB
Copyright © 2014 Oleg Puzanov. All rights reserved.
Prototyping Setup: IoT-GW Device v0.1
BeagleBone Black with Android 4.2.2 (reduced)
433 MHz eZ430 Dongle with DASH7
433 MHz TI Chronos Watch with DASH7
2.4 GHz CC2531 Dongle with TI Z-StackCC2531 and
XBee-Pro
DASH7
ZigBee
USB
USB
Copyright © 2014 Oleg Puzanov. All rights reserved.
Problem-Solution Statement
Copyright © 2014 Oleg Puzanov. All rights reserved.
Problem Statement
Poorly specified in chunks across standards • OGC SWE, W3C SSN, IoT-A Reference Architecture,
ETSI M2M - all of them cover some details in a partial way.
• No single complete standard on the state of today!
Poorly addressed by IoT solution vendors • Industry is focused on device-to-cloud connectivity and
Big Data Analytics instead.• IoT industry is young - many gaps and ambiguities,
vendors have hard times in understanding and adopting IoT architectures.
Disconnect between horizontal platforms and vertical applications • “Everything Connected” or “Device Cloud” horizontal
platforms - most of them are too much generic and don’t support the application-level specifics.
• Data models and data protocols - not defined and not implemented in the horizontal platforms.
No solutions combining all critical characteristics • Semantic, spatial, contextual, distributed and managed
IoT Data Layer - any products supporting all of these characteristics today?
• Offline mode and smart data synchronization are very poorly supported - many IoT products are limited to “always on” device-to-cloud integration.
Challenging E2E integration of IoT applications • Heterogenous technologies, data sources and
connectivity interfaces.• Lack of standards and well-established practices for
IoT application architectures.• Legacy approach for application data models and data
integration (or no approaches at all).• Slow IoT adoption across verticals, both B2C and B2B
applications.
IoT paradigms are not adopted, benefits and innovations are not delivered • M2M, SCADA, RFID and WSAN applications have
been around for decades - many IoT applications do not bring anything new into this space.
• User experiences - they’re still far away from the main ideas of IoT. Legacy UI and user-facing features “hide” all innovations of IoT applications.
Lack of system-wide intelligence and smart data processing • “Dumb data pipes” - sending data from sensor
networks to the cloud applications.• Context awareness - still in the early stages for IoT
applications, including contextual data synchronization and personalized context-driven UI.
• Cloud-side intelligence mainly - role of IoT field networks is limited to data acquisition.
IoT Application LayerIoT Data Layer
Copyright © 2014 Oleg Puzanov. All rights reserved.
Solution
Successful IoT = Data-driven Implementation
1
2
3
Data models are defined and implemented.
E2E integration is driven by IoT Data Layer, not technologies.
Enabled system-wide intelligence and smart data processing for field networks, cloud platforms and user interfaces.
Copyright © 2014 Oleg Puzanov. All rights reserved.
Platform Details
Copyright © 2014 Oleg Puzanov. All rights reserved.
Augmented Context UI Application (IoT-ACUI)
IoT network view with Augmented Reality features
Android and iOS: tablets, smartphones
Contextual POI presentation in real-time (RFID, WSAN)
Rich spatial data models
Connects to IoT-GW via Bluetooth or directly to the cloud
GeoJSON over XMPP or MQTT for data synchronization
Offline mode by default: local cache of IoT-SCR
Copyright © 2014 Oleg Puzanov. All rights reserved.
IoT Gateway Device (IoT-GW)
• Integrated firmware to run on ARM Cortex-A or Intel x86/64 platforms.
• Linux v3.8.x or Android 4.2.x
• Built on top of OSGi and Java frameworks, easy porting to other operating systems.
Wearables
Home Gateways
In-vehicle Gateways
Body Area Network Gateways
Industrial Gateways
Smart Metering/Utility Gateways
Telecare/Telehealth Gateways
• Includes IoT-SCR, IoT-MD, IoT-FGF, IoT-DPF and system specific modules.
Copyright © 2014 Oleg Puzanov. All rights reserved.
Field Gateway Framework (IoT-FGF)
OSGi bundles running on Embedded Android/Linux stack
RFID, RTLS, Proximity and WSAN controllers
DASH7, ZigBee, Bluetooth, Wi-Fi and OBDII supported initially
Part of IoT-GW responsible for field connectivity and services
No protocol stack implementations, integration only (e.g. TI CC2530 with Z-Stack)
6LoWPAN, Wireless M-Bus, WirelessHART, KNX in the roadmap
Copyright © 2014 Oleg Puzanov. All rights reserved.
IoT Vehicle Gateway Device (based on IoT-GW)
Heavy-duty Trucks
Civil Cars
Industrial Machinery
Agricultural Equipment
Military Vehicles
Construction Equipment
Copyright © 2014 Oleg Puzanov. All rights reserved.
Smart Context Repository (IoT-SCR)
Full Repository Lite Repository
Spatial Search GeoSPARQL
CRUD Batch
Core Repository Functions
Context Processing Functions
Export/Import FunctionsSemantic Spatial Contextual
• Geospatial RDF framework with advanced context-driven data processing features.
GeoJSON GeoRSSRDF
• Graph DB (Neo4j) is used by default for storage engine with H2 DB for Lite version.
Copyright © 2014 Oleg Puzanov. All rights reserved.
Metadata Directory (IoT-MD)
Core Metadata Field Services Metadata
Cloud Services Metadata
Application Metadata
• Hierarchical data model definitions - ontologies, class hierarchy, templates • Covers the common classes and the application-specific data models • RDF/RDFS, OWL, GeoJSON or the native POJO classes • Import/Export into Java (POJO) data models • Every supported vertical application has its own metadata
Copyright © 2014 Oleg Puzanov. All rights reserved.
Data Protocol Framework (IoT-DPF)
Communication Protocols
XMPP MQTT HTTP
Payload Formats
GeoJSON GeoRSS RDF
Data Transfer Functions
Context Sync
PubSub
Data Query
Upload/DownloadCommunication framework for data synchronization and event-driven processing.
Copyright © 2014 Oleg Puzanov. All rights reserved.
Big Data Integration Framework (IoT-BIG)
• Enables Apache Hadoop and Apache Storm for IoT-SCR with all related tooling for Big Data management.
!• Both batch processing and real-time
processing of Big Data.
Real-time Processing Batch Processing Storage Analytics
Storm
Copyright © 2014 Oleg Puzanov. All rights reserved.
Web Portal Demo (IoT-WEB)
API for Web Apps (REST, Java)
Integrated Portlets Framework
+
Integration Middleware
Demo web portal and API to showcase IoT applications development.
Copyright © 2014 Oleg Puzanov. All rights reserved.
Roadmap for Industry Standards and Features
Copyright © 2014 Oleg Puzanov. All rights reserved.
Out of Scope for Data-driven IoT™ Platform
• Field protocol stack implementations (ZigBee, DASH7, 6LoWPAN) - integration with 3rd-party SDK and API only.
• Remote device management - to be handled partially by OSGi remote management functions and commands over XMPP/MQTT protocols.
• Complete web portal UI - demo web portal is provided only. Reference implementation to showcase the platform features.
• Full-featured Big Data analytics and reporting - current implementation relies on the available tooling for Apache Hadoop and Apache Storm.
• On-device firmware for RFID tags or Sensor nodes - we use OpenTag, Contiki OS, Tiny OS, FreeRTOS, TI Z-Stack or other ready-to-use software stacks here. Minor configuration changes are applied only, no major implementations of tag/sensor software.
To be focused on the main scope of Data-driven IoT™ platform the following features are considered out of scope or low priority:
Copyright © 2014 Oleg Puzanov. All rights reserved.