Weak State Routing for Large Scale Dynamic Networks Utku Gnay Acer, - - PowerPoint PPT Presentation

Weak State Routing for Large Scale Dynamic Networks

Utku Günay Acer, Shivkumar Kalyanaraman, Alhussein A. Abouzeid

Rensselaer Polytechnic Institute

Department of Electrical, Computer & Systems Engineering

q Which area should we NOT be working

• n in MOBICOM anymore?

q Ans: Routing !

• Victor Bahl, Mobicom 2007 panel

Routing in Large-scale Dynamic Networks

q Routing table entries: “state” = indirections from persistent names (ID) to locators q Due to dynamism, such indirections break q Problematic on two dimensions

þ Dynamism/mobility => frequent update of state þ Dynamism + large scale => very high overhead, hard to maintain structure

q We propose constructing routing table indirections using probabilistic and more stable state: WEAK STATE

Number of Nodes Node Mobility

A new class of State

q Strong State

þ Deterministic þ Requires control traffic to refresh þ Rapidly invalidated in dynamic environments

q Weak State

þ Probabilistic hints þ Updated locally þ Exhibits persistence

STATE B STATE B

Hard, Soft and Weak State

a b INSTALL

STATE A STATE A

REMOVE

Hard State Soft State

UPDATE Time elapsed since state installed/ refreshed

Weak State

Confidence in state information

Weak State is natural generalization of soft state

Random Directional Walks

q Both used to announce location information (“put”) and forward packets (“get”)

Outline

q Our Weak State Realization q Disseminating Information and Forwarding Packets q Simulation Results q Discussion & Conclusion

An Instance of Weak State

q The uncertainty in the mappings is captured by locally weakening/decaying the state q Other realizations are possible

þ Prophet, EDBF etc…

{a,b,c,d,e,f} Probabilistic in terms of membership SetofIDs GeoRegion Probabilistic in terms of scope

q Consider a node a q Where is node a?

þ (i): It is in region AB with probability 1 þ (ii) It is in region B with probability 2 (1 · 2)

Example: Weak State

128.113. 128.113.62. 128.113.50.

xn

1

x

2

x x

2

θ

1

θ x

How to “Weaken” State?

Larger Geo-Region Aggregation SetofIDs -> GeoRegion

Aggregation: setofIDs

q setofIDs: We use a Bloom filter, denoted by B.

m1 m2

…. 0 1 0 0 0 0 1 1 1 1 1 1 u hj(m1)

…. ….

k k hi(m1) hj(m2) hi(m2)

Decaying/Weakening the setofIDs

q Randomly reset 1’s to 0. Same as EDBF [Kumar et al. Infocom’05] q Let (m)=i=1

m B(hi(m))

q Large (m) ! fresh mapping q (m)/k is a rough measure of P{xm 2 A}

…. 0 1 0 0 0 0 1 1 1 1 1 1 hj(m) hk(m) h1(m) hi(m) …. 0 1 0 0 0 0 1 1 1 1 1 …. 0 1 0 0 0 0 1 1 1 1 1 1 …. 0 1 0 0 0 0 1 1 1 1 1 …. 0 1 0 0 0 1 1 1 1

Weakening State (Contd)

xn

2 u B ≤ setofIDs small, time passes: Decay GeoRegion

xn

2 u B ≥

xn

Either setofIDs large OR GeoRegion Large => Decay SetofIDs

Random Directional Walks

q Both used to announce location information (“put”) and forward packets (“get”)

Dissemination/Proactive Phase: (put)

q When a node receives a location announcement, it

þ creates a ID-to- location mapping þ aggregates this mapping with previously created mappings if possible

A B C

Confidence 0.71 Confidence 0.84 Confidence 1.0 Confidence 0.71 Confidence 0.84 Confidence 0.71 Confidence 0.84 Confidence 0.71 Confidence 0.71

Forwarding Packets (get)

Forwarding decision: similar to longest-prefix-match. “strongest semantics match” to decide how to bias the random walk.

Confidence 0.71 Confidence 0.84 Confidence 1.0 1.0 Confidence 0.71 Confidence 0.84 Confidence 1.0 1.0 Confidence 0.71 Confidence 0.84 Confidence 1.0

S D A B C E

WSR involves unstructured, flat, but scalable routing ; no flooding !

Simulation Objectives

q How does WSR perform?

þ Large-scale þ High Mobility

q Comparisons:

þ DSR: works well for small scale, high mobility þ GLS+GPSR: works well for large scale, low mobility

q Short answer: 98%+ packet delivery despite large scale AND high mobility. q Tradeoffs: longer path lengths, (N3/2) state

• verhead

Simulation Setup

q Benchmarks

þ DSR and GLS-GPSR

q Random waypoint mobility model with vmin=5 m/s and vmax=10 m/s

þ WSR is robust against dynamism (please see the paper for details)

q Performance Metrics

þ Packet delivery ratio þ Control packet overhead þ Number of states maintained þ Normalized path length þ End-to-end Delay

400 500 600 700 800 900 1000

• 0.2

0.2 0.4 0.6 0.8 1 1.2 Number of Nodes Packet Devivery Ratio WSR GLS-GPSR DSR

Packet Delivery Ratio

GLS breaks down due to

• verheads

DSR only delivers a small fraction of packets WSR achieves high delivery ratio

Control Packet Overhead

400 500 600 700 800 900 1000 1000 2000 3000 4000 5000 6000 7000 Number of Nodes Total Overhead per Second (Number of Packets) WSR GLS-GPSR

Maintaining structure requires superlinearly increasing overhead in GLS

400 500 600 700 800 900 1000 0.5 1 1.5 2 2.5 3 3.5 4 x 10

4

Number of Nodes Total Number of Mappings/Database Entries Maintained WSR GLS-GPSR

Number of States Maintained

The total state stored in the network increases as (N3/2) instead of (NlogN)

Per-Node State Dynamics

100 200 300 400 500 600 700 800 900 1000 5 10 15 20 25 30 35 40 45 time (seconds) Number of States Maintained

State generation rate matches state removal rate.

Distribution of Per-Node State in the Network

The states are well distributed with a C.o.V 0.101

20 25 30 35 40 45 50 55 20 40 60 80 100 120 Number of States Number of Occurrences

Normalized Path Length

400 500 600 700 800 900 1000 1 1.5 2 2.5 3 3.5 4 4.5 Number of Nodes Normalized Path Length WSR GLS-GPSR DSR

GLS sends packets only to destinations that are successfully located Packets take longer paths with WSR

400 500 600 700 800 900 1000 10 20 30 40 50 60 70 Number of Nodes End to End Delay (s) WSR GLS-GPSR DSR

But, E2E Delay is Lower!

WSR uses random walks for discovery & dissemination => end- to-end delay is smaller

Discussion/Future Work

q Weak State Routing also relates to

þ DTN routing þ Unstructured rare object recall in P2P networks þ Distributed Hashing

q Future work:

þ Such extensions (especially DTNs) þ Theoretical analysis

Conclusion

q Weak state is soft, updated locally, probabilistic and captures uncertainty q Random directional walks both for location advertisement and data forwarding. q WSR provides scalable routing in large, dynamic MANETs q WSR achieves high delivery ratio with scalable overhead at the cost of increased path length

Thank you

Questions?