NDN DN A Applica cations NDNComm 2014 ICN Tutorial Dry Run - - PowerPoint PPT Presentation

NDN DN A Applica cations NDNComm 2014 – ICN Tutorial Dry Run September 3, 2014 jburke@ucla.edu

1 ¡

NSF F FIA 2 2010-p

• present

Video s streaming, l live ch chat, f file s sharing, l lighting co control, s sensing NSF F FIA “ “Next P Phase” ( (2014-2

• 2016)

En Enterprise B Building A Automation & & M Management Op Open M Mobile H Health Mobile M Multimedia A Applica cations Ot Other o

• ngoing w

work ( (2014-2

• 2016)

Internet o

• f T

Things, R Raspberry P y PI s support Climate M Modeling ( (Colorado S State) Information M Maximizing N Networks ( (UIUC) Vehicu cular N Networking ( (UPMC, U UCLA)

NDN: Application-motivated Approach

Support from NSF, Cisco, Comcast, Toyota, Qualcomm, Verisign, and others.

Ch ChronoCh Chat

3 ¡

What is a group text chat application?

Alice Bob Ted Name Content ... ... /ca/alice/15 Hello, guys /ca/alice/16 Do you guys need coffee? Name Content ... ... /ca/ted/3 Hello, Alice. Name Content ... ... /ca/bob/31 Yup.

4 ¡ Synchroniza/on ¡of ¡distributed ¡chat ¡room ¡dataset ¡(set ¡of ¡sequences ¡of ¡chat ¡messages) ¡ among ¡the ¡par/cipants ¡ Sequence ¡of ¡ Alice’s ¡ messages ¡ Sequence ¡of ¡ Ted’s ¡ messages ¡ Sequence ¡of ¡ Bob’s ¡ messages ¡

ChronoChat

Many I y Internet a applica cations a are co collaborative b by n y nature

• group t

text ch chat

• file s

sharing

• audio/video co

conferenci cing

Ke Key p y piece ce i in t these a applica cations

• distributed s

state s syn ynch chronization

• ch

chat r room m messages

• files a

and f folders i in t the s shared f folder

• voice

ce/video s streams f from e each ch p partici cipant

Based o

• n ChronoSyn

ync

• syn

ynch chronize t the k knowledge o

• f t

the ch chat m message s set

Two t trust m models h have b been e experimented

• hierarch

chica cal

• authentica

cate u users t through a a ce certifica cate ch chain t that s strict ctly f y follow t the n naming hierarch chy y

• web-o
• of-t
• trust
• authentica

cate u users t through e endorsements m made b by e y each ch o

• ther’s d

direct ctly/ y/indirect ctly y trusted u users

5 ¡

NDN DNVideo

6 ¡

ACQUIRE ¡ A/V ¡ ENCODE ¡ (Future: ¡SVC) ¡ STORE ¡ TO ¡REPO ¡

VIDEO ¡ AUDIO ¡ SENSORS ¡ CLOCK ¡ CLOCK ¡

STORE ¡ TO ¡REPO ¡

EDGE ¡ROUTER ¡ W/CONTENT ¡STORE ¡ Interest ¡for ¡ 00:01:40.20 ¡

• r ¡

Segment ¡2981 ¡ “REPO” ¡ “REPO” ¡

NDNVideo: Sessionless, Scalable Live, Pre-recorded Streaming

• D. Kulinski and J. Burke. NDNVideo: Live and Prerecorded Streaming over NDN.

Technical Report NDN-0007, September, 2012.

Timeco code ( (frame-b

• based) n

names make r random a acce ccess s straightforward.

Design App Namespace to Leverage Architecture

• D. Kulinski and J. Burke. NDNVideo: Live and Prerecorded Streaming over NDN.

Technical Report NDN-0007, September, 2012. Session-l

• less a

approach ch f for s sca calability. y.

ndnrt ndnrtc

9 ¡

Browser

NDN-RTC Web App (Discovery, UI) NDN-RTC Add-on JavaScript interface C++ engine NDN-RTC NDN-CPP library WebRTC library NDN

NDN Real Time Conferencing Tool Go Goals ls: :

• Real-t
• time a

audio/video/text ch chat l library w y which ch a allows m many-t y-to-m

• many

y co conferenci cing o

• ver t

the N NDN DN n network a and r requires n no d direct ct co communica cation b between p peers

• Traffic g

c generator f for t the te testbe tbed

• Start p

point f for N NDN DN t traffic co c congestion co control a algorithm r research ch

• Test N

NDN DN-C

• CPP l

library a y and T TLV p pack cket f format

9/9/14 ¡ 10 ¡

Deadline-based pipeline for driving interest retransmission

B1 ¡>= ¡RTT, ¡B2 ¡>= ¡RTT ¡ Minimal ¡buffer ¡size ¡>= ¡2*RTT ¡milliseconds ¡ 9/9/14 ¡ 11 ¡

frames move playback pointer re-transmit pointer interests for segments data segments interests for missing segments missing data segments should be assembled,

• therwise -

considered missed

B2 B1

new frame - no segments fetched yet fully fetched frame frame being assembled (some segments fetched)

Vehicu cular n networking

12 ¡

Applying NDN to vehicle networking Question v via t the n network: Collisions a ahead o

• n I

I-8

• 80?

/toyota/collision/by_road/us/california/I-80 /toyota/collision/recent

Research ch g goal: a a single f framework for v vehicl cles

• to f

fully u y utilize a all a available p phys ysica cal ch channels

• to co

communica cate i in a an i infrastruct cture-f

• free m

manner

• to co

communica cate w with i infrastruct cture s servers

• to p

provide d delay-t y-tolerant d delivery y No n need t to d deal w with h host a addresses; still m must d design a applica cation n namespace ce.

13 ¡

UCLA & UIUC

Ad-h

• hoc /

c / M Mule Infrastruct cture

Pau, Zhang, et al. with Toyota

(Mobile) d data g gets t to ( (mobile) co consumers a any w y way i y it ca can

Climate M Modeling

15 ¡

Publisher Announcement

l Server 1

1 a advertises J January p y prefix

l Server2 a

advertises F February p y prefix

l Routing p

protoco col p propagates announce cements

l Servers a

answer a at a appropriate granularity f y for t the a applica cation

l Client w

wants J Jan 3 30 – – F Feb 0 02

l NDN

DN r routes J Jan r requests t to Server1 a and F Feb r requests t to Server2 Server2 Consumer Request

NDN Approach

/cmmap/precipitation/ GCRM/GridZ/ <horiz_resolution>/ <field>/<date>/<time>/ Colorado State

Naming in Climate Modeling Application

CMIP5: “ “.. a a co common n naming s sys ystem t to b be u used i in f files, d direct ctories, m metadata, and U URLs t to i identify d y datasets w wherever t they m y might b be l loca cated w within t the distributed C CMIP5 a arch

• chive. I

It d defines co controlled v voca cabularies f for m many o y of t the co components co comprising t the d data r reference ce s syn yntax ( (DR DRS).

17 ¡

CMIP5 DRS is NDN-compliant Hierarchical Clearly defined name components Well-defined vocabularies Organizes components from less specific to more specific Utilities to translate into DRS (CMOR) Produces virtually ready-to-use NDN names

Di Direct ctory e y enco coding:

• /<act

ctivity> y>/<product ct>/<institute>/<model>/<experiment>/ < <frequency> cy>/ <modeling r realm>/ < <variable n name>/<ensemble m member>

• Ex

Example: /CMIP5/output/MOH

OHC/HadCM3/deca cadal1990/day/ y/at atmos mos/ta tas/r /r3i2 i2p1/

/

/coupled /control /CMMAP /r3i1p1 /spcesm-ctrl /pop /1M /1891-01/

spcesm-ctrl.pop.h.1891-01.nc Contents of file

activity/sub-activity/organization/ensemble/experiment/model/ granularity/start-time

Our Translator Architecture

NP N Network En Environments

19 ¡

En Enterprise b building a automation and m management s sys ystems

20 ¡

Two r research ch t threads

• En

Enterprise-l

• level B

BAS/BMS i in co collaboration w with U UCLA F Faci cilities Management, b based o

• n w

work s started i in a an EA EAGER ER.

• De

Device ce-s

• side IoT

IoT, m motivated b by co y consumer e experience ce / / h home environment f for n

• now. (

(Ex External s support f from Q Qualco comm a and potentially H y Huawei.)

21 ¡

Home e environment

22 ¡

• W. ¡Shang ¡et ¡al. ¡

EB EBAMS: F Focu cusing o

• n “

“Management” / / “ “Automation” l levels

23 ¡

Variable Speed Drives Power Meters LonWorks Devices BACnet OPC SQL ODBC Web

Management- Level Automation- Level Field- Level

BACnet Chillers Boilers Fire/Life Safety Security/Access Lighting Power Monitoring PLCs Particle Measuring Other BAS

Management Level Network Building Level Network Floor Level Network

Siemens ¡Apogee ¡System ¡

Two Two te testbe tbeds ds a at U UCLA – – o

• ne s

shared a and o

• ne u

unique.

24

Sensing: Electrical Demand & Chilled Water

NDN DN – – S Suitability / y / B Benefits

Massive a addressing s simplifica cation, w with a a p potential f for h huge i impact ct w when s sca caled to t the e

• enterprise. S

Simpler n network i infrastruct cture n needed t to d deploy co y complex monitoring a and a automation. New w way o y of w working w with e edge r resource ces t that d de-e

• emphasizes g

gateway a y addressing while p preserving s support f for t topologica cal h heterogeneity. y. Lighter-w

• weight, d

data-ce

• centric s

c secu curity o y options e easier t to d develop, w with d data verifica cation i intrinsica cally p y part o

• f t

the a arch chitect cture. Cach ching a and s storage i integration m may p y provide s significa cant a advantages i in d distributed storage a at a all l levels o

• f t

the a arch chitect cture, i incr creasing d data a availability w y without p power incr crease. Intrinsic m c multica cast; m many-t y-to-m

• many co

y communica cation e easier t to d deploy. y.

25 ¡

Hierarchical naming already used at application layer

26 ¡

<root-prefix> strathmore melnitz user data kds public alex power hvac

<key-id> <key-id>

<timestamp>

building studio 1 panel data kds voltage J K current

<timestamp> <timestamp> <timestamp>

acl acl acl acl key apl key apl

Forward based on names closely mapped to physical world ¡

Shang, W, et al. "Securing Building Management Systems Using [NDN].” IEEE Network 2014.

Da Data

Signed ACK

• Asym

ymmetric k c keys ys t to w work d direct ctly w y with P PK-b K-based i identities

• Sym

ymmetric k c keys ys a and H HMACs f for f faster ‘ ‘signatures’

• Leverage N

NDN DN t to d distribute k keys ys a and e establish k key r y relationships

• Command p

privacy b cy by e y encr cryp ypting n non-r

• routable p

portion o

• f n

name

Interest Interest signed b by a y app

boelter/3551/lights/fixture/41/rgb-8bit-hex/FAF87F/<state>/<authenticator>

Lighting Lighting Co Contr trol l Ro Routin tine

NDN DN Ne Netw twork

Authenticated, Closed-Loop Control in NDN

Burke, J., et al. "Securing Instrumented Environments over Content-Centric Networking: the Case of Lighting Control." IEEE NOMEN 2013.

Op Open mH mHealt ealth

29 ¡

Op Open mH mHealt ealth: Data rather than System Interoperability

Interoperable, I Internet-i

• inspired d

data exch change a as t the b back ckbone o

• f t

the a applica cation eco cosys ystem Thin w waist o

• f o
• pen d

data i interch change standards t that w will e enable a an e eco cosys ystem o

• f

sensing, s storage, a analys ysis, a and u user i interface ce co components t to s support m medica cal d disco covery y and e evidence ce-b

• based ca

care Market-s

• supported, p

patient-ce

• centered

landsca cape o

• f i

innovative h health a applica cations Patient-co

• controlled, p

privacy-a cy-aware d data exch change a acr cross d device ce, co component, a and applica cation b boundaries 30 ¡

• D. ¡Estrin ¡(Cornell), ¡I. ¡Sim ¡(UCSF) ¡

Mapping t the Op Open mH mHealt ealth A Arch chitect cture t to N NDN DN

31 ¡

hYp://openmhealth.org/ ¡

Request ¡/ ¡Response ¡model ¡ Namespace ¡design ¡ Composability ¡& ¡ Authen/ca/on ¡ Provenance ¡ Access ¡Audit ¡ Storage ¡Design ¡

Same C Challenges, Di Different L Laye yers

For t this a applica cation i in p particu cular, N NDN DN p provides m much ch m more r relevant f funct ctionality a y at the n network l laye yer t than I IP. So s solutions i in N NDN DN h have m much ch m more d direct ct i impact ct o

• n t

the s sca calability, y, s secu curity, y, a and ease o

• f d

development; w we n need n not b build u up a additional l laye yers o

• n I

IP t to g get n near t the app ch challenges.

Namespace ce / / s sch chema d design Repository / y / s storage d design Service ce co composability Authentica cation / / i identity a y assurance ce Da Data p provenance ce Acce ccess a auditing Mobile p publishing Legal r requirements f for s succe ccess

32 ¡

Open ¡mHealth ¡arch. ¡looks ¡a ¡lot ¡like ¡NDN ¡

NDN DN A Applica cations NDNComm 2014 – ICN Tutorial Dry Run September 3, 2014 jburke@ucla.edu

33 ¡