fun with networks: social, sensor, and shape-shifting phillip b. gibbons intel research pittsburgh...

43
Fun with Networks: Social, Sensor, and Shape- Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except those borrowed from colleagues) are © Phillip B. Gibbons

Upload: angelina-neal

Post on 17-Dec-2015

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Fun with Networks:Social, Sensor, and Shape-Shifting

Phillip B. GibbonsIntel Research Pittsburgh

DISC’08 / Graal’08September 24, 2008

Slides (except those borrowed from colleagues) are © Phillip B. Gibbons

Page 2: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’083

Fun with Networks

Social Networks– SybilLimit: Defending against Sybil Attacks in P2P

Sensor Networks– Synopsis Diffusion: Robust in-network aggregation

Shape-Shifting Networks– Claytronics: Aggregation in programmable matter

Page 3: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’084

Background: Sybil Attack

Sybil attack:Single user assumes many fake/sybil identities– Already observed in real-world

p2p systems

Sybil identities can become a large fractionof all identities – “Out-vote” honest users in

collaborative tasks

launchsybilattack

honest

malicious

Page 4: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’085

Background:Defending Against Sybil Attack

Using trusted central authority (TCA) – Ties identities to human beings

– Not always desirable: who to trust, privacy, etc.– Practice: Gmail accounts

Much harder without a TCA [Douceur’02]– Resource challenges not sufficient

– IP address-based approach not sufficient– Practice: Wikipedia IP blocking

Widely considered real & challenging– 40 papers on sybil attacks, no distributed solution

Page 5: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’086

SybilGuard/SybilLimit Basic Insight: Leveraging Social Networks

Nodes = identities Undirected edges =

strong mutual trust – E.g., colleagues,

relatives in real-world

– Not online friends !

SybilGuard [SIGCOMM’06, TON 2008], SybilLimit [Oakland’08]

(with Haifeng Yu*, Michael Kaminsky)

First to leverage social networks for thwarting sybil attacks with provable guarantees

* Primary author

Page 6: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’087

Attack Model

malicioususers

honestnodes

Observation: Adversary cannot create extra attack edges

attack edges

n honest users: One identity/node each Malicious users: Multiple identities each (sybil nodes)

sybil nodes

sybil nodes may collude – the adversary

Attack edge: edge between honest node& sybil node

Page 7: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’088

SybilGuard/SybilLimit Basic Insight

honest nodes sybil nodes

Dis-proportionally small cut disconnecting a large number of identities

But cannot search brute-force…attack

edges

Page 8: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’089

SybilLimit End Guarantees

Completely decentralized

Enables any given verifier node to decide whether to accept any given suspect node– Accept: Provide service to / receive service from

– Ideally: Accept and only accept honest nodes – unfortunately not possible

Bounds # of accepted sybil nodes (w.h.p.)

Accepts (1- )n honest nodes (w.h.p.)

nnO log/ per attack edge [up to)(logn attack edges]

We also prove that SybilLimit is away from optimal)(lognO

Page 9: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0810

Example Application Scenarios

If # of sybil nodes accepted is

Then applications can do

< n/2 byzantine consensus

< n majority voting

< n/c for some constant c secure DHT [Awerbuch’06,

Castro’02, Fiat’05]

… …

Page 10: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0811

Identity Registration

Each node (honest or sybil) has a locally generated public/private key pair– “Identity”: V accepts S means

V accepts S’s public key KS– We do not assume/need PKI

Every suspect S “registers” KS on some other nodes

Page 11: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0812

Registration Goals

Ensure that sybil nodes (collectively) register only on limited number of honest nodes

– Still provide enough “registration opportunities” for honest nodes

sybil regionhonest region

K: registered keys of sybil nodes

K K

K

KK

K

K K

K

K

K

K

K

KK K

K: registered keys of honest nodes

Page 12: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0813

Acceptance Criteria

Accept S only if KS is register on sufficiently many honest nodes

– Without knowing where the honest region is !

– Circular design? We can use small cut against adversary

K K

K

KK

K

K K

K

K

K

K

K

KK K

sybil regionhonest region

K: registered keys of sybil nodes

K: registered keys of honest nodes

Page 13: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0814

)(logn Take random “walks” of w= hops– Honest nodes: likely to remain in honest region*

– Sybil nodes: must cross an attack edge to reach honest region

Key Idea

sybil regionhonest region

K K

K

KK

K

K K

K

K

K

K

K

KK K

• Register key at last hop of “walk”

* w = Social network’s mixing time End up at ~random edge in honest region

Page 14: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0815

Random 1 to 1 mapping between incoming edge and outgoing edge

Random Route: Convergence

a db ac bd c

d ee df f

a

b

c

d e

f

randomized

routing table

Using routing table gives Convergence Property:

Routes merge if crossing the same edge

Page 15: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0816

Implication of Convergence

Claim: There are at most w K’s per attack edge– Proof: By the Convergence property

– Regardless of whether sybil nodes follow protocol

honestnodes

sybilnodes

attack edgeK

K KK

Route length w

Use independent instances of random routing m

Page 16: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0817

4. Is KS registered?

Verification Procedure

VS

1. request S’s set of tails AB

CDEF

F

2. I have three tails

AB; CD; EF

3.common tail: EF

5. Yes. 4 messages involved

V accepts S Tails intersect + key registered

Earlier: Each node registers at tails m

Page 17: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0818

Further Details in Paper

Birthday paradox V & honest S share a common tail w.h.p.

Limit on sybil Ks in honest region V &sybil S don’t share a common tail w.h.p.– Unless V has a tail in sybil region: Handled in paper

How to estimate parameters: w & m

Evaluation w/ real-world social networks– Friendster, LiveJournal, DBLP (Added sybil nodes)

Page 18: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0819

Conclusions (to Part I)

Sybil attack:– Widely considered a real & challenging problem

SybilLimit: Fully decentralized defense protocol based on social networks– Provable near-optimal guarantees

– Experimental validation on real social networks

Open Problem (in SybilLimit & Politics):

Honest users not voting

Page 19: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0820

Fun with Networks

Social Networks– SybilLimit: Defending against Sybil Attacks in P2P

Sensor Networks– Synopsis Diffusion: Robust in-network aggregation

Shape-Shifting Networks– Claytronics: Aggregation in programmable matter

Page 20: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0821

Wireless Sensor Network Aggregation

Aggregate in-network over a tree– Each node sends 1 short message (saves energy)

10

0

10

20

30

40

50

60

70

0 10 20 30 40 50

Time

% N

od

es In

clu

ded

3

1 1

31

1

37

1

2 1

Page 21: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0822

The Problem and the Goal

Tree topology used to avoid double-counting

Aggregation and routing are tightly coupled

Our goal: Decouple the two components– They can be independently optimized

– Robust multi-path routing can be used

– Can exploit the broadcast medium

11 1

1

3 13

7

1 1

3 45

12

In contrast, a gossip approach requires point-to-point messages & explicit acks

Page 22: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0823

Synopsis Diffusion

Each node generates a small synopsis of its readings (SG)

Starting with outer ring, each node broadcasts its synopsis

Synopsis Fusion (SF): Each node in next ring combines all synopses it hears into its own synopsis

SF must be order- and duplicate- insensitive (ODI)

ExampleTopology:

Rings

e.g., Compute count or sum using Flajolet-Martin’s e.g., Compute count or sum using Flajolet-Martin’s distinct-values countingdistinct-values counting [Considine et al, ICDE’04] [Considine et al, ICDE’04]

[with Suman Nath*, Srini Seshan, Zach Anderson, SenSys’04, TOSN 2008]

* Primary author

Page 23: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0824

SD Example: Uniform Sample of Size K

SG(): Each node selects a random r in [0,1], and creates a synopsis (r, id, val)

SF(s,s’): Output the K (r,id,val) triples from s U s’ with maximum r-values

SE(s): Output the K val’s in s

K=2: (.4,1,v1), (.7,2,v2), (.3,3,v3), (.8,4,v4)

{(.4,1,v1),(.7,2,v2)}

{(.7,2,v2), (.4,1,v1)}

{(.7,2,v2),(.8,4,v4)}{v2,v4}

{(.7,2,v2),(.3,3,v3)}{(.3,3,v3),(.8,4,v4)}

Page 24: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0825

Key Challenge & A Solution

ODI Goal: S1 is always the same

SF SF SF

SG

r1

SG

r2

SG

r3

SG

r4

SG

r5

SFSF SF

SF

SE

S1

Result

Aggregation Topology

SF SF SF

SFSF SF

SF

SF SF SF

SFSF SF

SF

SF SF SF

SFSF SF

SF

Potentially large unknown

set of combinations!

Key Result:Give 4 simple,locally testableproperties for ODI correctness(necessary & sufficient)

Makes topologyindependence tractable

Page 25: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0826

Order- & Duplicate-Insensitive Synopses

Necessary & sufficient conditions1. SF is commutative

2. SF is associative

3. SF is same-synopsis idempotent: SF(s,s) = s

4. If readings r and r’ are “duplicates”, then SG(r) = SG(r’)

E.g., suppose use SF(s1,s2) = (s1+s2)/2, which of P1-P3 fails?

P2: SF(2,SF(6,30)) = 10 but SF(SF(2,6),30) = 17

Page 26: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0827

Implications SF forms a semi-lattice Lattice property can tell if another ODI synopsis accounts for my synopsis

E.g., SF is bitwise-OR00101

10111

Implicit acks (Listen to what parent sends to know if your message was “received”)

Efficient adaptation to dynamic message loss, even when asymmetric links

More robust routing More accurate answers

4

6Not true for

non-ODI e.g., sum

Page 27: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0828

ODI-Correct Algorithms

Count, Count Distinct, Sum, Average, Standard deviation, Second moment, Uniform sample, k’th statistical moment, Quantiles, Frequent items, Range aggregates, Inner product queries

For ODI-correct algorithms:Approximation guarantees

Well-studiedStreaming Model

=same3

52

2

2253

Page 28: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0829

Synopsis Diffusion on Rings

Scheme EnergyTree (TAG) 41.8mjA. Rings 42.1mjFlood 685mj

0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8 1

Loss Rate

RM

S E

rro

r

TAG Rings

Adaptive Rings Flood

More robust than TAGAlmost as energy efficient as TAG

600 sensors in 20x20Count query(tree)

Page 29: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0830

Synopsis Diffusion vs. Tree

Tributary-Delta: run both simultaneously, depending on:

[with Amit Manjhi, Suman Nath, ICDE’05]

SD

Tree

Communication error

1%

60%

Approximationerror

10-15%

0-5%

Number of Packets

1-3

1

Delta

Tributary

• regional loss rate• accumulated aggregation

Page 30: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0831

Conclusions (to Part II)

Synopsis Diffusion– ODI-correct algorithms + any multi-path routing

Open Problems– ODI-correct subtraction

– Use Synopsis Diffusion in other contexts:

– P2P, mobile, etc.

– ODI-correctness requires the same synopsis for all aggregation topologies

– However, too strong: E.g., quantiles – always meets guarantees but answer depends on order

– What is a formal framework for such scenarios?

Page 31: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0832

Fun with Networks

Social Networks– SybilLimit: Defending against Sybil Attacks in P2P

Sensor Networks– Synopsis Diffusion: Robust in-network aggregation

Shape-Shifting Networks– Claytronics: Aggregation in programmable matter

Page 32: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0833

The Vision: A Material That Changes Shape

Large groups of tiny robot modules (106 -109 units), working in unison to form tangible, moving 3D shapes

Not just an illusion of 3D (as with stereo glasses), but real physical objects Both an output device (rendering, haptics) & an input device (sensing)

Page 33: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0834

Suppose Software CouldControl Shape

Video: CMU Entertainment Technology Center

Page 34: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0835

Applications

Product design Medical visualization Adaptive form-factor devices Telepario 3D fax Smart antennas Paramedic-on-demand Entertainment Etc.

Page 35: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0836

Claytronics[PIs: Seth Goldstein, Jason Campbell, Todd Mowry]

Each sub-millimeter module (“catom”) integrates computing & actuation

Key issues: – very high concurrency (106 -109 catoms)– nondeterminism & unreliability– efficient actuators, strong adhesion– power, heat, dirt– complex, dynamic networking (network diameters

≥ 1000, and changing topologies)

Page 36: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0837

Moving Catoms Without Moving Parts:

Two Potential Actuation Methods Magnetic field

Electric field

one coil two assembled magnet rings 2 magnetic-field prototype catoms

electrostatic latch design

completed latch

Page 37: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0838

patterned “flower”,including actuators& control circuitry

arms curl up due to stresses between layers

Making Submillimeter Catoms

[J. Robert Reid, Air Force Research Labs]

[Igal Chertkow & Boaz Weinfeld, Intel]

2 mold wafersbonded around

1 thinned logic wafer

Note: Both areearly attempts

Page 38: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0839

Catom Design Actuation: Roll across each other (using electrostatics) under software control– Planned motion, Reactive motion

Power: Form own power grid– Connected to external power source

Communication: Between physically adjacent modules– Either electrical contact, capacitive-coupled

connections, or free space optics (wire-like)

– Simultaneously with multiple neighbors

Page 39: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0840

Aggregation Goal

In order to self-organize into a desired shape, the catom ensemble must:– Be able to measure key aggregate properties

(e.g., center of mass)

– Coordinate their activities

…in real time

Diameter too large for standard hop-by-hop approach

Ensemble too dense for longer range wireless

Page 40: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0841

Speculative Forwarding[with Casey Helfrich, Todd Mowry, Babu Pillai,

Ben Rister, Srini Seshan]

Standard approach:(regular) gradient

E.g., regular 2D grid

Our approach:• Hierarchical Overlay• Speculative forwarding on the long links

Page 41: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0842

Speculative Forwarding Each catom maintains incoming-to-outgoing link mapping (e.g., last used)

Each bit along incoming wire sent on outgoing wire according to the mapping

When accumulate header, check for miss-speculation

Aggregation deferred to nodes in the overlay

Many issues:• miss-speculations• creating overlay• shape changes

Initial resultsare promising

Page 42: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0843

Conclusions (to Part III)

Shape-Shifting Networks pose a new problem domain for algorithmic research– Details are in flux; realizations years away

– Key issues: scale, dynamics, soft real-time

Open Problems– Much theory work to be done:

Formal modeling, new algorithms, new insights, lower bounds, etc.

– E.g., what is a robust, low-latency communication/aggregation scheme for catom ensembles?

– Ensemble algorithmics: local algsBrownian hole motionGrow/consume holes

Page 43: Fun with Networks: Social, Sensor, and Shape-Shifting Phillip B. Gibbons Intel Research Pittsburgh DISC’08 / Graal’08 September 24, 2008 Slides (except

Phillip B. Gibbons, DISC’08/Graal’0844

Fun with Networks

Social Networks– SybilLimit: Defending against Sybil Attacks in P2P

Sensor Networks– Synopsis Diffusion: Robust in-network aggregation

Shape-Shifting Networks– Claytronics: Aggregation in programmable matter