internet of things (iot)를위한애널리틱개발및적용...6 iot analytics devices analytics...

28
1

Upload: others

Post on 11-Aug-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

1

Page 2: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

2© 2017 The MathWorks, Inc.

Internet of Things (IoT)를위한애널리틱개발및적용

성호현차장

Senior Application Engineer

The MathWorks Korea

Page 3: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

3

What is IoT?

Devices InsightAnalytics

Page 4: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

4

Do hamsters run a marathon every day?

Fun Example from Michelle Leonhart

Page 5: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

5

ChallengeMeasure and evaluate the energy efficiency of

residential homes and factories based on data from

onsite sensors

SolutionCadmus used ThingSpeak to collect up-to-the-minute

temperature, humidity, and power usage data and use

MATLAB to analyze and visualize the data

Results Market opportunity seized

Development effort cut by two-thirds

Sensor networks quickly deployed

Commercial Example from Cadmus

Page 6: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

6

IoT Analytics

Devices InsightAnalytics

Preprocess Data

Data Reduction/

Transformation

Feature Extraction

Develop Predictive

Models

Model Creation

Model Validation

Access and

Explore Data

Business Data

Sensor Data

Page 7: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

7

Algorithms are Key to IoT Systems and MATLAB Can Help

Real data is messy and

needs to be cleaned up.

Features need to be

detected and classified

Missing data points

need to be handled

Predictions need to be

made

Page 8: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

8

IoT Analytics Framework

Smart Connected Devices

• Storage

• On-Line analytics

• Visualization & reporting

• Communication

• Embedded Sensor Analytics

• Data Reduction

Analytic IoT Platform

Page 9: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

9

IoT Analytics Challenges

How do I collect enough data to build my algorithm?

How do I develop my algorithms?

How do I deploy my algorithms on a smart device?

How do I deploy my algorithms to the cloud?

Page 10: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

10

IoT Analytics Challenges

Algorithm Development

Sensor Analytics

Data Aggregation

& AnalyticsCommunications

Network

Deploy analytics

to server/cloud

Deploy algorithms to nodes/devices

How do I collect

enough data to

build my

algorithm?

Smart Connected

Devices

How do I

develop my

algorithms?

Page 11: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

11

What Is ThingSpeak?

Web Site For People Web Service for Devices

Page 12: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

12

ThingSpeak

New MathWorks web service

hosted on AWS

Lets you collect, analyze and act

on data from “things”

Over 130,000 users worldwide

It has MATLAB for IoT Analytics

It’s free to get started

https://thingspeak.com

Page 13: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

13

Custom Visualizations with ThingSpeak- Weather Station Example

Page 14: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

14

Predictive Analytics Example with ThingSpeak

Page 15: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

15

IoT Analytics Challenges

Algorithm Development

Sensor Analytics

Data Aggregation

& AnalyticsCommunications

Network

Deploy analytics

to server/cloud

Deploy algorithms to nodes/devices

Smart Connected

DevicesHow do I deploy my

algorithms on a

smart device?

How do I collect

enough data to

build my

algorithm?

How do I

develop my

algorithms?

Page 16: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

16

Sensor Analytics and Development of Smart Connected Devices

Machine

LearningConnect and Acquire

Signal

Processing

Embedded

Implementation

Page 17: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

17

Example: Human Activity Analysis and Classification

ClassificationFeature

Extraction

Page 18: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

18

Objectives

Measure, explore, discover traffic patterns

Provide live local traffic information service

Solution

RaspberryPi + webcam

Automated deployment of vision

algorithms on embedded sensor

Full example available at

makerzone.mathworks.com

Example: Partitioning Algorithms Between Device and Cloud

Page 19: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

19

When should I go t

o work?

From Data to Insight

Page 20: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

20

Designing Smart Connected Devices

Gather data from sensors using I2C/SPI and other interfaces

Use pre-built libraries for signal processing , computer vision, m

achine learning and more

Automatically generate C / C++ and HDL code

Embedded targeting packages for a wide variety of hardware

Smart Connected

Devices

Sensor Analytics

Algorithm

Development

Deploy algorithms to device(s)

Retrieve data from sensor(s)

C, C++ HDL PLC

Page 21: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

21

IoT Analytics Challenges

Algorithm Development

Sensor Analytics

Data Aggregation

& AnalyticsCommunications

Network

Deploy analytics

to server/cloud

Deploy algorithms to nodes/devices

Smart Connected

Devices

How do I deploy

my algorithms to

the cloud?

How do I collect

enough data to

build my

algorithm?

How do I

develop my

algorithms?

How do I deploy

my algorithms on

a smart device?

Page 22: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

22

ThingSpeak for Small Scale Deployment

External Data & Business

Systems

Store

Deploy analytics

To cloudComputeIngest

Algorithm Development

Sensor Analytics

Smart

Connected Devices

Page 23: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

23

Integrating MATLAB with Third Party

IoT Cloud Platforms

External Data & Business

Systems

Store

Algorithm Development

Deploy analytics

To cloudComputeIngest

IoT

Platform

MATLAB

Production

Server

Gateway

Smart Connected

DevicesAlgorithm Development

Sensor Analytics

Page 24: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

24

Databases

Cloud

Storage

IoT

Visualization

Web

Custom App

Public Cloud Private Cloud

Integrating MATLAB in Large Scale Production Systems

Platform

Data Business System

MATLAB

Production

Server

Analytics

Request

Broker

Azure

Blob

Azure

SQL

Page 25: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

25

MathWorks Addresses IoT Analytics Challenges

Quickly collect and analyze IoT data with ThingSpeak and MATLAB

Develop analytics algorithms using MATLAB and toolboxes

Deploy on smart devices using code generation and embedded target support

Deploy on cloud using ThingSpeak and MATLAB Production Server

Page 26: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

26

What You Can Do to Learn More

Log-in to ThingSpeak with you MathWork

s account and explore

View a webinar on Machine Learning wit

h MATLAB

Read a Technical Article on Forecasting

Tides with MATLAB

Read a tutorial on how to send data to Th

ingSpeak over MQTT

Page 27: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

27

Using the Corporate Template

Avoid manually formatting whenever possible. Instead, use built-in styles, templates, layouts, and col

ors.

When creating new presentations, select the slide layout that best suits your needs from the built-in t

heme, then add content.

When creating custom shapes, text boxes, and other elements, start from scratch rather than reforma

tting template shapes.

When applying the new template to existing presentations, review your presentation carefully and ma

nually adjust any formatting issues that have occurred. For additional help, contact Creative Services.

V17.0 Delete this slide before finalizing your presentation.

Page 28: Internet of Things (IoT)를위한애널리틱개발및적용...6 IoT Analytics Devices Analytics Insight Preprocess Data Data Reduction/ Transformation Feature Extraction Develop Predictive

28

감사합니다.