Upload
lytruc
View
228
Download
0
Embed Size (px)
Citation preview
2017. 3. 10
Edge
Analytics for
IoT Device
Intelligence
1. IoT Trends
2. IoT Analytics
3. Edge Analytics Platform:
Kanga
4. Future Direction
14 1
IoT Trends - Business/Technology (1/3)
Google Cloud Platform
Google : IoT Solution Business using Google Could Platform
MS : Artificial Intelligence IoT Ecosystem based on Azure
GE : Predix Platform Solution & Service Business for Industrial Internet
Microsoft IoT Platform Services
14 2
IoT Trends – IoT Analytics (2/3)
IoT business models will exploit the information collected by “things” in many
ways - for example, to understand customer behavior, to deliver services,
to improve product, and to identify and intercept business moments.
However, IoT demands new analytic approaches: [Gartner, Jan 2016]
■ Tens of billions of connected “things”
The volume of data to be analyzed will increase dramatically
Demanding technologies such as machine learning to identify patterns
.■ Traditional IT approach to data analytics is not always desirable or practical in IoT context, new analytics architectures are emerging
To much data to store, so analysis of data stream must be conducted on the fly
Data filtering and analysis may sometimes distributed in gateways at the edge of the network or
in the “things” themselves to minimize communication over slow networks or when communications
have undesirable side effects, such as increasing battery consumption in sensory nodes.
■ New data types and analysis algorithms for “things” data
Time series data is very common, demanding filters and Fourier transforms
More data are location-aware, demanding geographic information processing
14 3
Cisco: IoT Reference Model & Edge Computing Architecture Suggestion (’14.10)
Intel: IoT Gateway Commercial Product for Real time Analytics ’15.4)
IBM: Edge Analytic Open Source Project Quarks(’16.2)
IBM-Cisco: A Global Partnership with Watson IoT & Cisco Edge Analytics(‘16.6)
IoT Trends - Business/Technology (2/3)
Cisco Edge Analytics Solution Concept Apache Open Source Project Intel Gateway Solutions for IoT
14 4
IoT Analytics – Model
Smart Health
Smart Home Smart Retail Smart
Education
SmartThings Sensors Samsung Gear Samsung Appliance 3rd Party Things
Samsung Hub Devices
14 5
IoT Analytics – Model
Things
Analyze self data of the thing. Things
Analytics
Answer,
Assisting,
Processed Knowledge
External Open Data/Knowledge/IoT Services
Analyze data of all related user/things and service, And provide the
optimal service in which it fits for the overall situation.
Cloud & Big Data Platform
User Query,
Context Information,
Filtered Log
Collect and analyze data of the
user/things which is in the inside of
specific space (ex. Home).
Samsung Hub Devices
Messaging Middleware (IoTivity)
14 6
IoT Analytics – Service Level
Samsung Hub Devices Smart car
Hub
Oven Printer Vehicle
Kitchen Room Classroom Garage
Home Office Factory School Car
Smart
Factory Hub
▶A Room
- Living room
- Kitchen
▶A Place
- Office
- Home
Wearable
Smart watch
3rd Party Service
Connectivity
Connectivity
▶Public
- People, Citizen
- Colleagues
▶A user & Family
- Me
- family
▶Private Things
- Wearable things
- Appliances
Rule market Samsung
Account Intelligence
Geographic Level Service target Level
Samsung IoT Service Platform
Big data
Analytics
▶Town & Country
- Nation, State
- Company
Components
Manufacturing
lines
Assembly
Office
Air Con. Black
Board
Mobile
User
Small brain Big brain Micro brain
14 7
IoT analytics using edge analytics F/W & Big Data Platform
Supporting user profiles/device knowledge for intelligence services
IoT Analytics – Platform & Benefit
Devices on IoT Cloud Storage & Broker
Customers
3rd-party services
Data
In/Out
Analytics
Model
Deploy
Web
/mobile
Data &
Knowledge
API
Analytics API
Edge Analytics F/W
Collect and analyze data of the
user/things which is in the inside of
specific space(Ex. Home). Analyze data of all related user/things and service,
and provide the optimal service in which it fits for the overall situation.
Hub Devices
Big Data
Platform
IoT Cloud
Connection
Server Registration
Server
Data
Storage Gateway
Server
Samsung
Account
Big
Data
Analytics
Real-time analytics Batch analytics
1. Context aware Home Watcher & Controller
using real-time stream data analytics
2. CE-RM & Remote Checkup using device
maintenance/prediction analytics
3. The Function and Content Recommendation
based on analysis of user life
Intelligent
Services
14 8
. User/Product/Market Profile & User & Product Integrated Profiles
Things-Integrated Profile
IoT Analytics – Data Perspective
14 9
Edge Analytics Platform – Kanga
Real-time
Dashboards Kanga IDE
Query Builder and Browser
Big Data Platform
Storage broker
Privacy Management
Library Store/Community
Data Mining
Analytics
model
Complex event script
(Kanga QL)
Event-In Event-Out
Analytics APIs
Logic
Topology
Edge Analytics
Management
Server
3rd Party
Library
Query Optimizer
On-demand
Event Processor Compiler
Continuous Stream I/O API
Stream Processing Engine Search Engine
Message Broker Platform
Plug-in
Manager Time
Series
Library
Statistical
Library
MLR
Library
Default plug-ins
Edge Analytics Framework (Kanga)
Base Library
Large-scale
data-processing engine
Open Source Site : Kanga Lite v1.0 (‘16. 10) : https://github.com/Samsung/Kanga
14 10
Edge Analytics Platform – Kanga Concept Architecture
Real-time Dashboards Kanga IDE
Query Builder and Browser
Query Optimizer
On-demand Event Processor Compiler
Continuous Stream I/O API
Streaming Query Processor
Storm / Node.js Event Processing Engine
Elasticsearch Search Engine
Kafka / ZeroMQ / MQTT Message Broker Platform
Spark / R / Python Large-scale
data-processing engine
14 11
Job Status/Monitoring
Drag-and-drop query (topology) builder
Data visualizer
Edge Analytics Platform – Kanga IDE & Visualizer
Data visualizer Data visualizer
14 12
Future Direction – Kanga Milestone
Kanga IDE (Topology Design/
Query Builder/
Job Monitoring)
Kanga
Visualizer (Report &
Dashboard)
Kanga Main
Common
Library (Collection/
Store/Processing)
Machine
Learning Library (Classification/Clustering/
Outlier detection)
Kanga (Java) / Kanga Lite(node.js)
Kanga
Marketplace
Kanga Suite (Common)
Kanga v1.0 (2016) Kanga v2.0 (2017)
One device supporting
Kanga IDE (Topology Design/
Query Builder/
Job Monitoring
/Deployment)
Kanga
Visualizer (Report &
Dashboard)
Kanga Main
Common
Library (Collection/
Store/Processing)
Machine
Learning Library (Classification/Clustering/
Outlier detection)
Kanga Cloud(Java) / Kanga Desktop(Java) / Kanga Lite(node.js)
Kanga
Marketplace
Kanga Suite (Common)
Multi device & cloud supporting
Reasoning
Library
(Sematic CEP)
High Speed
I/O Library
(Sensor F/W)
Iotivity
(OCF) Library
Open Source : Kanga Lite v1.0 Open Source : Kanga Lite v2.0 (’17.4Q)
14 13
Future Direction – Kanga v2.0
Sensor HW (Vibration/Temperature,)
High speed IO (Sensor) Framework
Kanga lite framework
Application
Sensor data simulator
Kanga lite framework
Application
Layer 1
Layer 2
Layer 3
Layer 4
Real Layer Test Layer
14 14
. Bringing all the elements together
* strong Industry collaboration to accelerate IoT deployments
Summary – Extract Business Value and Take Action
Intelligent Devices
Intelligent System of Systems
Edge Analytics + + = Customer Value
Copyright © 2017 Samsung SDS Co., Ltd. All rights reserved