How Dataflow is opening the Internet Frontier

How Dataflow is opening the Internet FrontierDataflow-to-synthesis Retrospective

David E. CullerUC BerkeleyArch Rock

Dataflow-to-synthesis Retrospective

David E. CullerUC BerkeleyArch Rock

2

The Internet

3

The Internet Frontier

0 2 4 6 8 10 12 14 16 18

-1

-0.5

0

0.5

1

Low resolution Sensor, Test4, Increasing frequency

Time (sec)

Acc

eler

atio

n (g

)

4

Why “Real” Information is so Important?

Enable New Knowledge

Improve Productivity

Protect HealthHigh-Confidence Transport

Enhance Safety & Security

Improve Food & H20

Save Resources

Preventing Failures

IncreaseComfort

5

…but what has Dataflow got to do with it?

6

all I ever learned …

Arvind’s Law: Value derived from concurrent activities is inverse to the cost of integrating communication and computation - so tag it and do it all!

Culler’s Corollary: but storage increases with concurrency - so do it with bounded resources

7

all I ever learned …

8

so at the public school …

Time

Log

of S

om

eth

ing

Threaded Abstract Machine - two level scheduling hierarchy - expose network to the compiler - no allocation on arrival

TAMActive Messages

Split-C

NOW

Ninja Staged Event Driven Arch.

for massive internet servers

9

Enabling Technology

Microcontroller RadioCommunication

FlashStorage

Sensors

IEEE 802.15.4

10

Enabling Technology

PhysicalWorld

Not your 3M’s

• MIPs of proc.

• KB of RAM / Prog

• kbps of bandwith

• uW of power

11

Enabling Research

PhysicalWorld

SiliconWorld StorageWireless Processing Sensors

Build an approximation of the future that you imagine and

study it to learn what “is true” in that possible world.

Berkeley Mote / TinyOS Platform

12

WSN Research Phenomenon…

SmartDustWeC Sensit / Expeditions

ReneNEST Mica

Intel/UCBdot

InteliMOTE

XBOWcc-dot

XBOWmica2

Intelrene’

XBOWrene2

Intelcf-mica

Boschcc-mica

Dust Incblue cc-TI

1/0211/00 3/03

digital sunrain-mica

XBOWmica

zeevo BT

nest6/02

Telos

XBOWmicaZ

5/04

IntelMOTE2

Eyes

BTNode TIP

trio

Sensicast

13

Berkeley…and beyond

StorageWireless Processing

Sensors

SmartDustNEST

Wireless Sensor Networks

14

Storage ProcessingWireless SensorsWSN mote platform

The Systems Challenge

Radio Serial

Flash ADC, Sensor I/F

MCU, Timers, Bus,…

Link

NetworkProtocols Blocks,

Logs, FilesScheduling,

ManagementStreaming

drivers

Over-the-air Programming

Applications and Services

Communication CentricResource-ConstrainedEvent-driven Execution

Tin

yOS

2.0

15

The Solution• Basically a dataflow engine in disguise

– Two-level Scheduling Hierarchy• Asynchronous events integrate external world• Computational Tasks

– Collections of Tasks and Events + Bounded State form a Component.– Never, never, never wait. Do work or go to sleep!!!!!!!!!!

• Recognition of reality– Zillions of unattended devices embedded in a noisy physical world require an

unprecedented level of robustness and SIMPLICITY– Application = Structured Graph of Components connected by Well-Designed

Interfaces.

• Escape Velocity– Our System and Network Abstractions (ca. Unix + Sockets + TCP) have

been essentially unchanged for decades and are deeply steeped in the 3M design point!

=> Provide a framework for defining boundaries and let the new layers fall where they may.

16

What we mean by “Low Power”• 2 AA => 1.5 amp hours (~4 watt hours)• Cell => 1 amp hour (3.5 watt hours)

Cell: 500 -1000 mW => few hours active

WiFi: 300 - 500 mW => several hours

GPS: 50 – 100 mW => couple days

WSN: 50 mW active, 20 uW passive

450 uW => one year

45 uW => ~10 years

Ave Power = fact * Pact + fsleep * Psleep + fwaking * Pwaking

* System design

* Leakage (~RAM)

* Nobody fools mother nature

17

Wireless Client vs Wireless Server

• Wireless Client – needs to last a day or two – has a human to keep them working properly– mostly formats specific incoming data for display

• Wireless server– needs to last for long periods– must be self-managing, adaptive, robust– Generates meaningful data for many uses– Often mobile (!!!)

18

Self-Organized Mesh Routing

0

112

2

2

22

19

Sensor Network “Networking”

RadioMetrixRFM

CC1000Bluetooth 802.15.4

eyesnordic

WooMacSMAC

TMACWiseMAC

FPS

MintRoute

ReORg

PAMAS

CGSR

DBF

MMRP

TBRPF

BMAC

DSDV

ARADSR

TORA

GSR GPSR GRAD

Ascent

SPIN

SPAN

Arrive

AODV

GAFResynch

Yao

Diffusion

Deluge Trickle Drip

RegionsHood

EnviroTrackTinyDB

PC

TTDD

Pico

FTSP

Phy

Link

Topology

Routing

Transport

Appln

Scheduling

20

What WSNs really look like

Field Tools

Client ToolsExternal Tools

Embedded Network

Gateway

Excel, MatlabEnshare, etc.

DeployQueryCommandVisualize

InternetGUI

LegacyData analysis

21

THE Question

If Wireless Sensor Networks represent a future of “billions of information devices embedded in the physical world,”

why don’t they run THE standard internetworking protocol?

22

THE Question

RFM,CC10k,…,802.15.4Sonet 802.11Ethernet

XML / RPC / REST / SOAP / OSGI

IP

Web Services

TCP / UDP

HTTP / FTP / SNMP

Serial

Plugs and PeopleSelf-Contained

Enet 10MEnet 100MEnet 1GEnet 10G

802.11a

GPRS802.11b

802.11g

23

The Answer

They should

• Substantially advances the state-of-the-art in both domains.• Implementing IP requires tackling the general case, not just a specific

operational slice– Interoperability with all other potential IP network links

– Potential to name and route to any IP-enabled device within security domain

– Robust operation despite external factors• Coexistence, interference, errant devices, ...

• While meeting the critical embedded wireless requirements– High reliability and adaptability

– Long lifetime on limited energy

– Manageability of many devices

– Within highly constrained resources

24

LoWPAN - 802.15.4 the low-power wireless IP Link

802.15.4, …802.11Ethernet Sonet

XML / RPC / REST / SOAP / OSGI

IP

Web Services

TCP / UDP

HTTP / FTP / SNMP

• 1% of 802.11 power, easier to embed, as easy to use.

• 8-16 bit MCUs with KBs, not MBs.

• Off 99% of the time

25

Making sensor nets make sense

802.15.4, …802.11Ethernet Sonet

XML / RPC / REST / SOAP / OSGI

IP

IETF 6lowpan

Web Services

TCP / UDP

HTTP / FTP / SNMP Pro

xy

/ G

ate

wa

y

26

Challenges for IP over 802.15.4• Header

– Standard IPv6 header is 40 bytes [RFC 2460]– Entire 802.15.4 MTU is 127 bytes [IEEE ]– Often data payload is small

• Fragmentation– Interoperability means that applications need not know the constraints of

physical links that might carry their packets– IP packets may be large, compared to 802.15.4 max frame size– IPv6 requires all links support 1280 byte packets [RFC 2460]

• Allow link-layer mesh routing under IP topology– 802.15.4 subnets may utilize multiple radio hops per IP hop– Similar to LAN switching within IP routing domain in Ethernet

• Allow IP routing over a mesh of 802.15.4 nodes– Options and capabilities already well-defines– Various protocols to establish routing tables

27

A little example

WiFiArchRockDemo

WiFiArchRockDemo Router

App Server andPresentation

WiFiWiFi

InternetInternet

Web ServicesProxy Server

6LoWPANRouter(6to4)

10.97.0.xxxFD97::xxx

192.168.1.xxx

192.168.1.109

29

dsp

mho

p

HC

1

frag

6LoWPAN Format Design• Orthogonal stackable header format• Almost no overhead for the ability to interoperate and scale.• Pay for only what you use

IEEE 802.15.4 Frame Format

IETF 6LoWPAN FormatIP UDPH

C1

Header compression

dsp

Dispatch: coexistence

preamble

SF

D

Len FCF

DS

N

Dst16 Src16

D pan Dst EUID 64 S pan Src EUID 64

Fchk

Network Header Application Data

Max 127 bytes

Mesh (L2) routing

HC

1

mho

p

dsp

dspfrag

Message > Frame fragmentation

HC

1

31

A new internet citizen

Router

ProxyGateway

High ReliabilityTriply RedundantUltra-low powerHighly ResponsiveAES128 SecuredIP-based Mesh Network

TCP/UDP IP

IP - layer 7

Sensor & Mgmt ServicesHTTPmSystat, Netstat, EchoPing, Traceroute, DHCPReboot

WiFi GPRS EtherNet LoWPAN

nc, telnet, ping, traceroute…

Rich Web View per NodeWeb Services / WSDLSNMP, Ganglia, EmailAdaptersData Warehouse

Browser,Enterprise,Controller

36

Low-Power Schedule-Free Routing

• Extends preamble sampling [DARPA packet radio, 1984] and low-power listening [UC Berkeley, 2000] to enable battery powered routers.

– Simplicity of an “always on” network at low-duty cycle by shifting effort from (frequent) listening to (infrequent) transmission.

– Flexible and configurable– Overlay time synchronization service for correlated sampling

and scheduling optimizations.

• Eliminates need for connectivity / interference survey• Eliminates schedule incompatibility across clusters• Eliminates costly scan for schedule on join / rejoin

37

Energy Cost of Packet Communication vs. Data Size

0 20 40 60 80 100 120

Bytes of Payload

uA

s p

er

Pa

ck

et

RCV 6LoWPAN Local <= Global

RCV 6LoWPAN Local <= Local

RCV Raw 802.15.4

TX 6LoWPAN Local => Global

TX 6LoWPAN Local => Local

TX Raw 802.15.4

Low Impact of 6LoWPAN on Lifetime –Comparison to *Raw* 802.15.4 Frame

MaxPayload

Energy Δ forfixed payload

**

* fully compressed header

* additional 16-byte IPv6 address

38

Power of IP Connectivity

LoWPAN

WiFiEthernet

GPRSsend (IP_addr, port, UDP, &data, len)

FirewallFirewall

40

Still a Role for Proxies

Router

ProxyGateway

High ReliabilityTriply RedundantUltra-low powerHighly ResponsiveAES128 SecuredIP-based Mesh Network

TCP/UDP IP

IP - layer 7

Sensor & Mgmt ServicesHTTPmSystat, Netstat, EchoPing, Traceroute, DHCPReboot

WiFi GPRS EtherNet LoWPAN

nc, telnet, ping, traceroute…

Rich Web View per NodeWeb Services / WSDLSNMP, Ganglia, EmailAdaptersData Warehouse

Browser,Enterprise,Controller

41

Service Oriented Architecture

• Service Description => interface & implementation– Operations supported, input/output objects– Bindings to network and data encoding schemes– Network address where service can be invoked

• Enough that client can generate code to access the service well

ServiceProvider

ServiceRequestor

ServiceRegistry

ServiceDescription

publishfind

bind

42

Embedded Web Services

802.15.4

<value> source=library time=12:31 temp=25.1<\value>

0 2 4 6 8 10 12 14 16 18

-1

-0.5

0

0.5

1

Low resolution Sensor, Test4, Increasing frequency

Time (sec)

Acc

eler

atio

n (g

)

Physical Signal

Wireless Packets

11 010010001Sampled Value010010001int temp;

<request service>

ServiceDescription

< get temp … set sample_rate set alarm … >

Web Services www.weather.com

XML information

<value> source=library time=12:53 temp=26.7<\value>

11 010110111

43

A new WSN world

tier1

tier2

client

server

tier3 SensorNetGW/Proxy

tier4

Sensor

SensorNetmote

physical info net

Embedded Services

PerlPython

Excel

NetWeaver

AquaLogic

C#

44

The Next Tier • Today: we can connect essentially

everybody

• Tomorrow: we will be able to connect and observe essentially everything of value

– physical spaces, objects, and their interactions

– physical information, not just keystrokes

– and we know dataflow is the technology of tomorrow

1000:1

1:1,000,000

LaptopPDA

Motes

years

Mainframe

Mini

Workstation PC

Phone

Comp:People

1:1

1:1,000

45

46

…and a new IETF working group• mailto: rsn@ietf.org