The Three Musketeers, IoT, Cloud, and Big Data: One for All, and All for One 김 대 영 Daeyoung Kim Jan. 24, 2014 교수 Director 카이스트 전산학과 Auto-ID Labs, KAIST http://resl.kaist.ac.kr http://www.autoidlabs.org [email protected]
May 10, 2015
The Three Musketeers, IoT, Cloud, and
Big Data: One for All, and All for One
김 대 영
Daeyoung Kim
Jan. 24, 2014
교수 Director
카이스트 전산학과 Auto-ID Labs, KAIST
http://resl.kaist.ac.kr http://www.autoidlabs.org [email protected]
2
발표 내용
• Internet of Things 의 정의 및 시장 예측
• Expecting New Big Players
• Internet of Things in Practice
• ID(Identification) to Big data services
• KAIST AUTO-ID LABS RESEARCH ACTIVITIES
3
INTERNET 은 INTERNET OF THINGS 시대로 진화중
IOT 의 정의 (1/2)
In 1999, the Internet of Things" was first
coined by Kevin Ashton who cofounded
the Auto-ID Center at the MIT
Business Processes
and Applications
Software and Network
Hardware
4
No concrete consensus! What they commonly say is that “the Internet embraces
everyday objects and lets them available anywhere, anytime.”
People or systems are able to access real-world through the Internet.
INTERNET 은 INTERNET OF THINGS 시대로 진화중
IOT 의 정의 (2/2)
5
IF IOT IS REALIZED?
• Internet상의 상태와 실세계의 상태가 일치
• 실시간으로 실세계의 다양한 상황 정보를 정확하고 세밀하게 제공
4인 3T,
소근소근
Full, 시끄러움
Full, 조용,
몬순 커피
2인2T
테이블 4, 조용
휴업
2인 4T 4인 2T
4인 2T
8인 2T, 조용
4인 3T, 시끄러움
검색: “어은동 카페”
6
EMERGING TECHNOLOGIES HYPE CYCLE
2013
7
IOT 에 관한 통계 및 시장규모 예측
IDC forecasts 15 billion devices by the year 2015"
ABI Research’s study, “Wireless Sensor Networks,” analyzes
IEEE 802.15.4 and which standards, such as ZigBee, Wireless
HART, and ISA100.11a, will influence and drive the market, as
well as emerging technology and competitive solutions such as
Z-Wave, 6LoWPAN, Bluetooth Smart, and Low Power Wi-Fi.
850 Million IEEE 802.15.4 Chipsets to
Ship in 2016, Despite Strong
Competition from Bluetooth
8
사물(THINGS)의 범위 (NO LIMITATION)
수동형 사물 액티브 태그형 사물 무선센서형 사물
스마트형 사물
스마트가전형 사물
지식데이터 사물
9
INTERNET OF THINGS 비전 / ARCHITECTURE
Future IT Eco-System
Human Information
Thing Information
Context-aware
Services
Social Network Service
M2M networking of
Home Devices
Internet Browsing
Entertainment
(Multimedia & Game)
Home Surveillance
Augmented Reality
Big DataAnalysis
Cloud Computing
Internet of Things
Cloud Computing
Internet of Things
User log, Business
activity logs, etc.
Internet contents,
SNS, etc.
Everyday objects Multimedia (video,
audio), etc.
Social Network Service for IoT
One for All, and All for One
10
EXPECTING NEW PLAYERS
11
INTERNET OF THINGS IN PRACTICE
Internet of Things는 새로운 분야이기
때문에 모든 기술과 제품을 새로 만들어야 하나요?
http://www.youtube.com/watch?v=wVOQc5rImqE&feature=share
12
SMART ACCESSARIES (I)
Eye-Fi Nemoto Dice+
Necomimi
13
SMART ACCESSARIES (II)
Instabeat Smart Sock Golfsense
Scanadu Scout
14
INTERNET OF THINGS IN PRACTICE
Internet of Things
어떤 기술(전략)이 필요할까요?
15
ID(IDENTIFICATION) TO BIG DATA SERVICES
ID & 네트워크
• IDENTIFICATION
• RFID/NFC/WSN/CR
• Visual Sensor Networks
• Future Internet
시스템
소프트웨어
• Small&Cheap H/W
• Operating Systems
• Smart Objects/Middleware
• Web of Things
• Tracking Technology
플랫폼
• Distributed File Systems and Database
• IoT Platform
• Cloud Computing
• Big Data
• Open APIs
서비스
• Eco System
• IOT Services
16
ID(IDENTIFICATION) 체계 (1/2)
How to reach a single instance?
How to name all of them?
Outnumbered? Global access!
Heterogeneity Ambiguity
17
• 다양한 현재의 ID 체계
• Uniform Resource Identifier (URI): http://myflowerpot.iot.kr/temperature
• Universally Unique Identifier (UUID): 550e8400-e29b-41d4-a716-446655440000
• Digital Object Identifier (DOI): 10.1000/182
• Global Trade Item Number (GTIN): urn:epc:id:sgtin:0614141.112345.400
• IDENFITICATION SYSTEM for the Internet of Things?
ID is registered to Identity provider
and is used with associated party
ID(IDENTIFICATION) 체계 (2/2)
ID on the WEB
18
Enterprise
OS Consumer
OS Mini
OS
Mid-level
Consumer electronics Enterprise
Tiny footprint
Consumer goods
OS for IoT
• 소비자 가전, 서버시스템은 Linux를 중심으로 표준화 되는 추세
• 적은 메모리, 낮은 수행능력을 가진 소형 Smart Things를 위한 운영체제
및 개발환경 필요 가능성 있음 / FreeRTOS, uCOS-II등 기존 OS
OPERATING SYSTEMS/SDK FOR SMART
RESOURCE CONSTRAINED THINGS
19
• IoT 에서 IPv6의 필요성
• 사물의 주소 역할
• 충분한 주소 공간, Plug &
Play, Mobile IPv6, IPSec ,
QoS
• 이 기종 네트워크 통합
• 이 기종 네트워크 간의
Seamless한 통신을 위해
네트워크 통합 방법 필요
• 표준화 된 IPv6 를 이용하여 이 기종 네트워크 통합
가능성
INTERNET OF THINGS를 위한 네트워크 기술
(FUTURE INTERNET 포함)
[ IP기반의 유무선 IOT 통합 네트워크 (글로벌 액세스) ]
20
“MIDDLEWARE” SOFTWARE FOR
THE INTERNET OF THINGS
• a software platform defined as middleware, fundamentally providing
abstraction to applications from the things, and offering multiple services
[인용] Role of middleware for internet of things: A study
S Bandyopadhyay, M Sengupta, S Maiti, S Dutta - airccse.org
21
• IoT에서는 사물이 생성하는 데이터를 저장하고 사물을 위한 computation power, 플랫폼, 그리고 소프트웨어 서비스 제공
• IoT + Cloud 융합 시도- Pachube
• 전 세계의 센서 데이터를 수집하는 웹 기반 온라인 데이터베이스 서비스
• 사물, 디바이스, 빌딩 등으로부터의 에너지 및 환경 데이터를 수집
(1) 실시간 데이터 그래프를 생성 (2) Historical한 데이터를 가공하거나
(3) 사용자에게 alert을 전송하는 기능
CLOUD COMPUTING FOR THE INTERNET OF
THINGS
22
• Big Data의 효율적 저장/관리/처리를
위하여 Massive infrastructure 가 필요
IOT GENERATES BIG DATA
Everyday objects in the Internet of Things
“Machine-generated sensor data will be
become a far larger portion of the Big Data
world, according to a recent report by
IDC. The research report, “The Digital
Universe in 2020,” published in December,
traces data trends from 2005-20. One of its
forecasts is that machine-generated
data will increase to 42 percent of
all data by 2020, up from 11
percent in 2005.”
23
WEB OF THINGS
A Web of Things Application Architecture, Dominique Guinard
24
IOT-BASED COMPANIES:
OPEN API + CLOUD + CHEAP HARDWARE + APPS
Xively Cloud Services™ The world’s first IoT Public Cloud
25
유럽 동향 : EU-FP7 & IERC
(IOT EUROPEAN RESEARCH CLUSTER)
Ideas
Capaci
ties
People Cooper
ation
Euratom
JRC
ETC.
Coordination and Support Action
for Global RFID-related Activities
and Standardisation.
Universal Integration of the Internet of Things through an IPv6-
based Service Oriented Architecture enabling heterogeneous
components interoperability.
Internet of Things at Work.
Internet of Things Architecture.
Internet of Things Initiative.
Internet Connected Objects for
Reconfigurable Ecosystems.
Internet of Things Environment
for Service Creation and Testing.
26
IOT 관련 표준화 단체
IoT
3GPP MTC (Machine-
Type Communication)
ETSI TC M2M Architecture
for ICT
IPv6 over Low power
WPAN (6lowpan)
NFC(Near Field
Communication)
standards
ZigBee(IEEE 802.15.4)
standards
RFID(Radio Frequency
Identification) standards
Internet Protocol (IP) for Smart Object
communications standards
BOSS(Business Operation
Support System) standards
Electronical Tech.
standards
Information and
communications technology
(ICT) standards
M2M Service Layer standards
IP(Internet protocol)
standard communication protocols
and interoperable systems
standards
Telecommunication standards
27
INTERNET OF THINGS AT KAIST
Internet of Things 연구 소개
28
GS1 DIGITAL – ID ON THE WEB
(AUTO-ID LABS)
29
GS1 DIGITAL – ID ON THE WEB
30
IP-WSN IN THE INTERNET OF THINGS
• The IP-WSN technologies breathe
life into the things by enabling IP
connectivity between humans and
things as well as between things
themselves.
• 6LoWPAN technology is one of such
upcoming networking technologies
which enable IP connectivity over
resource-limited, low-power, and low-
rate networks.
Internet of Things
SNAIL Border Router (6LBR)
SNAIL Node (6LN)
SNAIL Node (6LN)
SNAIL Node (6LN)
SNAIL Node (6LN)
IEEE 802.15.4
Btle
IEEE 802.15.4
Btle
Entertainment & Social Net. ServiceDevice Browsing & Mashup
Big Data AnalysisUser Experience with IoT Service
31
OPEN STANDARD FOR IP-BASED WSN
• Standardization Activities for IP-based Wireless Sensor Networks
• IETF 6LoWPAN WG : IPv6 over Low power WPAN
• IETF ROLL WG: Routing Over Low power and Lossy networks
• IETF CoRE WG: Constrained RESTful Environments
Formed to adapt IPv6 technology over
IEEE802.15.4 networks
- RFC 4944: “Transmission of IPv6 Packets over
IEEE 802.15.4 networks”
- RFC 4919: “6LoWPANs: Overview, Assumptions,
Problem Statement, and Goals”
- RFC 6282: “Compression Format for IPv6
Datagrams over IEEE 802.15.4-Based Networks”
Formed to define IP layer routing protocol
independent of layer 2
- RFC 5867: “building Automation Routing
Requirements in Low-Power and Lossy Networks”
- RFC 5826: “Home Automation Routing
Requirements in Lower-Power and Lossy Networks”
- RFC 5678: “Industrial Routing Requirements in
Low-Power and Lossy Networks”
- RFC 6206: “The Trickle Algorithm”
- RFC 5548: “Routing Requirements for Urban
Low-Power and Lossy Networks”
Formed to realize the REST architecture in
a suitable form for the most constrained
nodes and networks
- draft-ietf-core-coap-07: “Constrained
Application Protocol (CoAP)”
- draft-ietf-core-block-04: “Blockwise
transfers in CoAP”
- draft-ietf-core-link-format-07: “CoRE Link
Format”
32
SNAIL (SENSOR NETWORK FOR AN ALL-IP WORLD)
SINCE 2007
• Supported Protocols
• Interoperability between IPv4/v6 domains and the IEEE 802.15.4
• Lightweight IPv6, ICMPv6, MIPv6, NEMO, UDP, TCP, SSL
• Dual-Mode gateway and SNAIL Adaptor
• HTML5, Web browsing (HTTP/TCP)
• Mesh routing in adaptation layer, RPL, Hierarchical Addressing
• Fast and Seamless Mobility management, Global Time Synchronization, Security
• Web Browsing architecture, Pretty Cloud Service
33
MULTI-GW BASED LOAD BALANCING
SCHEME
• MLEq virtually model 3D-terrain with reflecting traffic load, hop distance from Gateway, link quality,
and capacity.
• All the node (gateways and routers) dynamically and in distributed way update their virtual height level (VL).
Internet
GW
Gateway bottleneckInternet
GW2
GW1
GW3
Single Gateway Network Multi-Gateway Network w/o load balancing
Imbalanced Data Traffic without load balancing Internet
GW2
GW1
GW3
Multi-Gateway Network w/ load balancing
Fairly distributed traffic load
High portion of traffic is focused on a few Gateways
InternetInternet
4
2
2
1
2
111
3
2
2 2
3 3
2
3
2
3
0
GW1
GW20
6
4
2
1
2
331
3
2
4 4
3 5
2
5
4
3
0
Routers
(6LRs)
Intersection
NodeIntersection
area
GW1 GW22
Higher Traffic load
Previous
Intersection areaGW1's Service domain GW2's Service domain GW1's Service domain GW2's Service domain
Number: VL
Lower Traffic load Balanced Traffic load
Gateways4
GW
MR
Level: 0
Level: 1
Level: 2
2
2
3
2
0
5
6
6 5
5
4
3
1
0
3
34
MULTI-GW BASED LOAD BALANCING
SCHEME
• Performance Evaluation (compared with RPL) – ns-2 Simulation
35
MOBILITY MANAGEMENT PROTOCOL
• MARIO includes movement detection, handoff management, and
location management schemes.
Data Req.Poll Req.
ACK
MACNET
Poll confirm
Data Req.Poll Req.
POLL
Interval
Retransmissions {Poll fail
# of Poll Req.
Fail : 1
Retransmissions {Poll fail
# of Poll Req.
Fail : 2
Retransmissions {Poll fail
# of Poll Req.
Fail : 3
Data Req.
Data Req.
Movement Detection Total 12 data requests
are transmitted to
detect MN's movement
MN MR
Time t0
Time t1
Time t2
Timeline
Poll Req.
Poll Req.
MAC
MN
MR1
MR2
MR3
MR4
Movement Detection MN
Candidate MRs={MR1, MR2, MR3, MR4}
RSSI from MNMR2 > MR3 > MR1 > MR4
Strong < - > Weak
① Send Orphan notification
② Each MR
calculates τslot
MacResponseWaitTime
τslot
Nslot
MR1MR2 MR3 MR4
0
Calculated Time to send realignment command
Signal Strength
③ Each MR sends realignment command
in its own τslot
④ The MN performs handoff to the MR which sent realignment command first.
MRA MRB MRC
MRD
MRE
MN MN MN MN
Initial K=0 MPFS successK=1
MPFS successK=2
MPFS successK=4
AMR IMR1 IMR2
IMR3
IMR4
Trajectory of MN
Forwarding Pointer
Reachability Test
LUReq
Success
Fail
DistanceMRA<->MRB=1MRB<->MRC=1MRC<->MRD=1MRD<->MRE=1MRC<->MRE=2
Movement Detection Handoff Management Location Management
36
WEB-BASED VISUALIZATION
• Smart devices and consumer electronics are equipped with web/CoAP servers that can
response directly to requests from the Internet
• Presentation Cloud provides rich web contents to support those embedded web servers
• Sensing data and Actuation commands/results are retrieved directly from web browser and
display on top of rich web interface, either in numbers or in graphs
Internet
Presentation Cloud which
stores rich web interface
Consumer Electronics Smart Metering Devices
Pricing
information
Rich Web interface for user-
friendly VisualizationDevice
Control
Power
Consumption
information
Web-based Interface
37
SNAIL ARCHITECTURE FOR IPV6 CONNECTION
38
[DEMO]
IOT INTERACTION WITH SMART DEVICES
39
• Slow Hopping MAC (SH-MAC) protocol
• A coordinator-based architecture
• Improved aggregate throughput by concurrent medium access
COGNITIVE RADIO & SDR
System model Operation of SH-MAC
Evaluation Result (without PU) Evaluation Result (with PU)
40
• SH-MAC implementation in high-performance CR/SDR platform
• SH-MAC demonstration in CR/SDR platform which aims to achieve
• Improved computing power by using embedded multi-processors
• Improved data bandwidth between baseband processor and front end H/W
by using USB 3.0 to Gigabit serial I/F
COGNITIVE RADIO & SDR
GNU radio
Ubuntu
ARM Cortex-A15
MCs
Protocols PU
base-band processor RF front-end/ IF processor
SU PU
SU SUC
Block diagram of CR/SDR platform
SH-MAC demo scenario in CR/SDR platform SH-MAC basic functions test
41
GROCERIES TRACE FRAMEWORK
• Goal
• To develop a cloud-based RFID framework for smart fridges.
Browsing my fridge
Third-party
apps
RFID framework over
Cloud infra.
42
Pointer
(e.g., URL)
Data
flow
GROCERIES TRACE FRAMEWORK:
ARCHITECTURE OVERVIEW
Core Business
Vocabulary (CBV)
Tag Data Standard
(TDS)
Tag Data
Translation (TDT)
D-EPCIS (Distribution
EPCIS)
M-EPCIS
(Manufacturer EPCIS) FridgeApp +RFID Readers
BizApp.
Other F&C
Middleware
ALE v1.1.1
implementation
(F&C middleware)
EPCIS (v1.0.1)
LLRP
ONS (v2.0.1)
Cap
turin
g
Ap
ps.
LLRP
EPCIS Query
control interface
Ca
ptu
re
inte
rface
EPCIS Query
callback interface
Other
EPCISs Cassa
ndra
LLRP v1.1 LLRP
Inside of
Cloud Infra Interface
Outside of
Cloud infra
Ca
ptu
re
inte
rface
ALE ALE
Discovery Service D-EPCIS
43
EPC NETWORK ON THE CLOUD
CapApp
RFID reader
EPCN
Admin
RFID reader
RFID reader
F&C
A
2. configure
LRspec
7. V* dispatches tag data to the
corresponding EPC components
according to flow mapping
5. configure CapApp to receive EC
reports from F&C
6. subscribe (EC report)
(flo
w m
ap
pin
g)
(DDNS for cl
oud infra)
EPCIS
A
Accessing
App
EPCDS
A
EPCIS
A Datastore
EPCIS
Datastore EPCDS
A EPCDS
F&C
A F&C CapApp
CapApp
RFID reader F&C (ALE)
external 2 cloud
(incoming)
src ip addr, src port dst private ip, dst port
cloud 2 external dst ip addr, dst port src private ip, src port
44
PERFORMANCE MEASUREMENT
RESULTS (SELECTED)
• EPCIS with Cassandra Datastore
• Improved response time with Cassandra (3-5 times faster)
Contents
EPCIS event capture
performance comparison
(Cassandra vs MySQL)
EPCIS event query
performance comparison
(Cassandra vs MySQL)
45
THE CURRENT SMART THING INFORMATION SERVICES
(STIS) IN IOT6 – EPC SENSOR NETWORK
Accessing App.
Other F&C
Middleware
Filtering and collection (F&C)
middleware
STIS (e.g., Manufacturer)
CoAP/
JSON(oBix)
Captu
ring
Apps.
LLRP
ALE
STIS Query
control interface
Pointer
(e.g., URL)
Ca
ptu
re
inte
rface
STIS Query callback
interface
EPC
standard
Other STISs Cass
andra
Captu
re
inte
rface
ALE CoAP/
JSON(oBix)
CoAP/JSON(
oBIX) 6LoWPAN
Geo-distance discovery service (ElasticSearch) ONS
DS Query
interface
DS Query
interface
Mobile phone oBix device
46
RFID Reader
(e.g., passive
tags)
Sensor Network (e.g., smartphone
phone)
Wireless ID and
Sensor networks
(e.g., active tags)
RFID Filter & Collection
Middleware
Sensor F&C&CEP
middleware
LLRP LLRP Sensor Interface Sensor Interface
Actuator Networks (e.g., consumer
electronics, appliance)
Sensor & Actuator Networks
Actuation interface
Object orchestration and organization management
Domain-specific applications
Sensor Interface
Sensor Interface
Smart Things Information Service
(static and dynamic information)
ALE
Actuation Interface
ObjectObject
Actuation Interface
ONS
Geo-distance
Discovery
Services
Fe
de
rate
d S
ys
tem
sD
ec
en
traliz
ed
Sy
ste
ms
SCALABLE STIS ARCHITECTURE FOR
THE CLOUD OF THINGS
Document-based NoSQL DB
for Geo-distance and text
search, analytics, etc. For time-series data with
frequent insertion
Relational DB for structed
MasterData
Oracle Spatial
and Graph
Graph database organization
relationship management
47
• IoT Mashup as a Service Cloud
• 사물 매쉬업 서비스 모델을 기반으로 런타임 IOT 서비스 생성을 지원하는 클라우드 기반 플랫폼
• 사물 매쉬업 서비스 모델: 사물, 소프트웨어, 컴퓨팅 리소스의 조합으로 이루어진 서비스 모델로서
엔드 유저가 런타임에 세 가지 요소를 선택하여 커스터마이징이 가능함
• 엔드 유저, 매쉬업 서비스 디자이너, 클라우드 제공자, 소프트웨어 개발자, 사물 소유자 등으로 이루어진 에코시스템
Low
Traffic
IOT MASHUP AS A SERVICE CLOUD
Thing
Selection
Processing
Logic
Selection
Computation
Resource
Selection
48
• IoT Mashup as a Service Cloud (Lightweight Cloud on Smartphones, IoT App.)
IOT MASHUP AS A SERVICE CLOUD
49
IOT SNS PLATFORM
• Lilliput – IOT SNS
• IoT를 위한 소셜 네트워킹 플랫폼
• Smart Thing Middleware + Online Social Network
• 유연한 그래프 구조를 통해 실 세계의 정보와 소셜 정보를 함께 제공
• 온라인 소셜 네트워크와 스마트 디바이스를 다루는 부담을 경감시키고 응용 개발자에게 추상화된
API를 제공하는 미들웨어
• Social 관계에 기반한 Access Control 기능
IoT Social Networks
Internet of Things Online Social Networks
User
Researcher
Application Developer
50
IOT SNS PLATFORM
Two Space Reflection Manager
Online Social Networks
IoTSocial Graph
Lilliput APIs
Lill
ipu
t A
rch
ite
ctu
re
IoTSocial Model
IoT Platforms
ApplicationApplication
Application
Real world contextMetadata, etc
Smart Thing
Personal profilessocial relationships, etc
Web Service Interface
Graph BuilderElement Extractor
Querying Manager
Access Controller
Thing ServiceInvoker
Service Invocatione.g. Air-conditioning
Publish ing messageUpdating relationships
Graph Utilization Manager
Security Manager
Lilliput Reflection Manager
OSN Publisher
IoT-Social Graph Manager
ModificationManager
xAcceptable Request Unacceptable Request
Reasoning Manager
Rules
Smart ThingService Manager
Exte
rnal
Com
po
nent
s
Entity Manager
Relationship Manager
Sync. Manager
Query Engine
Reasoning Engine
Notification Receiver
Chg. Notif. Manager
• Lilliput – IOT SNS
• Sorcerer’s Book – IoT SNS Application, Lilliput API를 이용한 응용 서비스
Lilliput Architecture
Sorcerer’s Book
51
VISUAL SENSOR NETWORKS - SEAHAVEN
비주얼 빅데이터의 분석을 통한 인간의 시각을 뛰어넘는 인지능력 확보
52
SEAHAVEN
53
IPv6Network
6to4
Tunneling
Object Naming Service (ONS)
EPC Information Service(EPCIS)
Filtering and Collection (F&C)
EPC Information Service(EPCIS)
2002:8ff8:6a89::8ff8:6a89
Filtering and Collection (F&C)
2002:8ff8:6a6c::8ff8:6a6c
DNS
Port: 53
Client
ALE Administrator
CoAP Client
Object Naming Service (ONS)
2002:8ff8:6a87::8ff8:6a87
1. Lookup IP address of EPCIS (DNS Protocol)
Accessing
Application
Port: 80
Capturing
Application
Port: 10300ALE report subscription
F&C Port:
8081
/ALEService
/ALELRService
SNAIL Gateway
2001:220:806:22::1
2001:220:806:20::1
CoAP LLRP poll
LLRP Port:
5084
RFID Reader FX7400
143.248.106.220
RFID Tagged
DevicesSensor Nodes
PRETTY Cloud Servers
Seahaven
iPhone
Laptop
Windows Mobile 7
Powerful Server
Beam Projector
Composited Service
(Service Specification)
On-Demand
Service Usage IoT
IOT 테스트베드 AT KAIST
54
IOT CAN MAKE IT HAPPEN
55