Interoperability Luca Mottola slides partly by Simon Duquennoy - - PowerPoint PPT Presentation

Politecnico di Milano, Italy and Swedish Institute of Computer Science

Interoperability

Luca Mottola slides partly by Simon Duquennoy

2

Not just stand-alone systems…

3

NES in business processes!

Motivation

WSNs as stand-alone systems, programmed by WSN geeks WSNs integrated with company back-end, programmed by business process experts

Applications

• Smart buildings
• Supply chain management
• Factory automation
• …
• Sensor abstraction layers

are not enough!

• Business processes

must cross gateway boundaries

• distributed in-network control
• disconnected operation

makeSense

Application Model Macroprogramming Abstractions Run-time Support Sensor Node WSN Business Process

Macroprogramming Language

Collective Action Report Action Tell Action Local Action Action Distributed Action Meta-Abstraction Modifier Target Data Operator

<<use>> <<use>> <<use>> <<use>>

1

0..1

Report Collective Tell f(x1 .. xn) Data Operator Target

Application Model

9

Application Model

1 Report Target

1 1

What about the Internet of Things?

CISCO’s Vision

Source: Cisco IBSG, April 2011

“Size doubling every 5.32 years. The IoT was born between 2008 and 2009.”

1 2

IPv6 in the IoT and Sensor Networks

• Connect heterogeneous links
• Getting Wifi, bluethooth, 802.15.4 to talk together
• Connect heterogeneous applications
• For the benefit of the user
• Rely on stable, tested standards
• Benefits of IPv6 over v4
• Scalability
• Stateless autoconfiguration

1 3

The IPv6 Embedded Stack

• Traditional IP stack
• Many of recent IETF standards
• IPv6 as the “narrow waist”
• Runs in system with a few kB memory

1 4

Wifi 802.15.4 PLC Bluetooth IPv6 (+ 6LoWPAN) + RPL TCP UDP HTTP XMPP .. CoAP .. ..

Link Layer

• IP connects a variety of link layers
• Wired or wireless
• Sensor and smart object link layers

are often lossy

1 5

Wifi 802.15.4 PLC Bluetooth IPv6 (+ 6LoWPAN) + RPL TCP UDP HTTP XMPP .. CoAP .. ..

Network Layer

• Narrow waist of the stack
• IPv6 (for scalability and autoconfiguration)
• 6LoWPAN brings IPv6 to 802.15.4
• RPL: routing over low-power and lossy links

1 6

Wifi 802.15.4 PLC Bluetooth IPv6 (+ 6LoWPAN) + RPL TCP UDP HTTP XMPP .. CoAP .. ..

The 6LoWPAN Adaptation Layer

• IPv6
• 128-bit addresses, 40-byte header, minimum MTU of 1280 bytes
• 802.15.4
• Only 127 bytes per packet...
• 6LoWPAN, RFC 4944
• Header compression (down to 2 bytes!), layer 2.5 fragmentation

1 7

The RPL Routing Protocol

• IETF Working Group Roll
• RPL protocol, RFC 6550
• Challenges addressed by Roll:
• Target networks optimized for saving energy
• Traffic pattern is not unicast-only
• Restrications on the frame size
• Efficiency-generality tradeoff
• Routing based on a DODAG topology

1 8

example DODAG

5 ¡ 6 ¡ 4 ¡ 2 ¡ 3 ¡ Root ¡ 7 ¡ 8 ¡

Transport Layer

• Like in the traditional Internet: TCP and UDP
• No specific adaptation (other than 6LoWPAN compression)

1 9

Wifi 802.15.4 PLC Bluetooth IPv6 (+ 6LoWPAN) + RPL TCP UDP HTTP XMPP .. CoAP .. ..

TCP and UDP on Resource-constrained Devices

• UDP is cheap and suits traditional sensing
• TCP more complex and some known issues
• Issues with congestion control
• Overhead of connection establishment
• Overhead of acknowledgments and reordering
• Never say never: IP, and then IPv6, were considered too expansive too!

2

Application Layer

• A variety of application layers
• Borrowed from the Internet (HTTP, XMPP) or not (CoAP)
• Most common goal: enable RESTful interaction

2 1

Wifi 802.15.4 PLC Bluetooth IPv6 (+ 6LoWPAN) + RPL TCP UDP HTTP XMPP .. CoAP .. ..

• Enables direct connection to existing servers and clients
• Lightweight HTTP implementations
• Embeds an HTTP server is systems with < 10kB code memory
• Contiki has Web server and clients: connect small sensor nodes and actuators
• HTTP servers run on Sun SPOT (i.e. embedded HTTP+java solution)

HTTP and the Web of Things

2 2

RESTful Smart Objects

• Protocol
• HTTP as the main/first standard protocol
• CoAP as a more lightweight alternative (detailed later)
• Interface example
• GET ¡/sensors/temperature ¡
• POST ¡/light?intensity=40 ¡
• An API to the world!
• Programmable with any scripting language
• Browsable from any device (computer, tabled, phone)
• Often referred to as the “Web of Things”
• Check out www.webofthings.com ¡!

2 3

CoAP and the Web of Things

• From the IETF CoRE working group
• Brings REST to the smallest devices
• Uses UDP instead of TCP
• Reduces HTTP header overhead
• Comes with new features
• Built-in server-push capability
• Resource description (CoRE) link format
• Group communication (multicast)

2 4

2 5

The cloud in the loop!

Cloud Computing

• Storage and applications as a service
• Took off insanely quickly (quicker than IoT ;))

2 6

app app app app app

A Cloud-centric IoT

2 7

Social Aspects

• Connect your world to your social networks!
• Discover friend’s resources, devices and data
• Create new interactions

2 8

Example: Participatory Sensing

• An initiative of Deborah Estrin at UCLA
• Smartphone-based sensing
• Data posted to the cloud by users
• Feedback helps user
• Ex: Noisetub, in Paris
• Monitoring noise pollution

2 9

Xively, an Online Platform for the IoT

• Formerly Cosm and Pachube
• Stores and shows sensing data
• Allows users to set triggers
• Supports applications
• Visualize from your Computer, phone, tablet

3

Some other existing platforms: sen.se, evrythng, thingsquare, …

SicsthSense: an Open Sensor Cloud Platform

• Connect Contiki, other sensors,

smartphones, internet services

• Perform in-Cloud management,

processing and triggering

• Dumb devices, smartness in the cloud!
• It’s open
• Open source (BSD)
• Freely usable at sense.sics.se ¡

3 1

More on SicsthSense

• Devices may post or be polled
• Both HTTP and CoAP, soon XMPP
• Manage access rights
• Organize your resources as a drive

3 2

• Feel free to start using J