46
Topology Control Murat Demirbas SUNY Buffalo Uses slides from Y.M. Wang and A. Arora

Topology Control

  • Upload
    percy

  • View
    49

  • Download
    0

Embed Size (px)

DESCRIPTION

Topology Control. Murat Demirbas SUNY Buffalo Uses slides from Y.M. Wang and A. Arora. Why Control Communications Topology. High density deployment is common Even with minimal sensor coverage, we get a high density communication network (radio range > typical sensor range) - PowerPoint PPT Presentation

Citation preview

Page 1: Topology Control

Topology Control

Murat Demirbas

SUNY Buffalo

Uses slides from

Y.M. Wang

and A. Arora

Page 2: Topology Control

2

• High density deployment is common

• Even with minimal sensor coverage, we get a high density communication network (radio range > typical sensor range)

• Energy constraints

• When not easily replenished

• High interference

• Many nodes in communication range

We will look at selecting high-quality links as part of routing!

Why Control Communications Topology

Page 3: Topology Control

3

Problem Statement(s)

1. Choose a transmit-power level whereby network is connected

• per node or same for all nodes

• with per node there is the issue of avoiding asymmetric links

• cone-based algorithm:

node u transmits with the minimum power ρu s.t. there is at least one neighbor in every

cone of angle x centered at u

2. Find an MCDS, i.e. a minimum subset of nodes that is both:

Set cover

Connected

Page 4: Topology Control

4

Problem Statement(s)

3. Find a minimum subset of nodes that provides some density

in each geographic region connectivity we’ll look at the examples of SPAN, GAF, CEC

Sub-problems:

• Prune asymmetric links• Tolerate node perturbations• Load balance

Page 5: Topology Control

5

Outline

• Cone-based algorithm

• SPAN

• GAF-CEC

Page 6: Topology Control

Analysis of a Cone-Based Distributed Topology Control Algorithm for Wireless Multi-hop Networks

L. Li, J. Y. HalpernCornell University

P. Bahl, Y. M. Wang, and R. WattenhoferMicrosoft Research, Redmond

Page 7: Topology Control

7

OUTLINE

• Motivation

• Bigger Picture and Related Work

• Basic Cone-Based Algorithm

Summary of Two Main Results

Properties of the Basic Algorithm

• Optimizations

Properties of Asymmetric Edge Removal

• Performance Evaluation

Page 8: Topology Control

8

• Example of No Topology Control with maximum transmission radius R (maximum connected node set)

High energy consumption High interference Low throughput

Motivation for Topology Control

Page 9: Topology Control

9

Network may partition

• Example of No Topology Control with smaller transmission radius

Page 10: Topology Control

10

Global connectivity Low energy consumption Low interference High throughput

• Example of Topology Control

Page 11: Topology Control

11

Bigger Picture and Related Work

Routing

MAC / Power-controlled MAC

SelectiveNode

Shutdown

TopologyControl

Relative Neighborhood Graphs, Gabriel graphs, Sphere-of-Influence graphs, -graphs, etc.

[GAF][Span]

[Hu 1993][Ramanathan & Rosales-Hain 2000][Rodoplu & Meng 1999][Wattenhofer et al. 2001]

ComputationalGeometry

[MBH 01][WTS 00]

Page 12: Topology Control

12

Basic Cone-Based Algorithm (INFOCOM 2001)

• Assumption: receiver can determine the direction of sender

Directional antenna community: Angle of Arrival problem

• Each node u broadcasts “Hello” with increasing power (radius)

• Each discovered neighbor v replies with “Ack”.

Page 13: Topology Control

13

Cone-Based Algorithm with Angle

Need a neighbor in every -cone.

Can I stop?

No! There’s an -gap!

Page 14: Topology Control

14

Notation

• E = { (u,v) V x V: v is a discovered neighbor by node u}

G = (V, E)

E may not be symmetric

(B,A) in E but (A,B) not in E

R A B 70

60

50

= 145

Page 15: Topology Control

15

Two symmetric sets

• E+ = { (u,v): (u,v) E or (v,u) E }

Symmetric closure of E

G+ = (V, E

+ )

• E- = { (u,v): (u,v) E and (v,u) E }

Asymmetric edge removal

G- = (V, E

- )

Page 16: Topology Control

16

Summary of Two Main Results

• Let GR = (V, ER), ER = { (u,v): d(u,v) R }

• Connectivity Theorem

If 150, then G+ preserves the connectivity of GR and the bound is tight.

• Asymmetric Edge Theorem

If 120, then G- preserves the connectivity of GR and the bound is tight.

Page 17: Topology Control

17

The Why-150 Lemma

150 = 90 + 60

Page 18: Topology Control

18

Both circles have max radius R

A

N

B

• Counterexample for = 150 +

Properties of the Basic Algorithm

Page 19: Topology Control

19

Both circles have max radius R

A

W

N

K

J

B

Y

WAN = 150 WAK = 150

• Counterexample for = 150 +

Page 20: Topology Control

20

Both circles have max radius R

A

N

B W

K

J

Y

WAN = 150 WAK = 150 Z

X 150 < WAX < α

d(A,X) < R < d(X,B)

• Counterexample for = 150 +

Page 21: Topology Control

21

For 150 ( 5/6 )

• Connectivity Lemma

if d(A,B) = d R and (A,B) E+, there must be a pair of nodes, one red and one green, with

distance less than d(A,B).

A B W

Y

Z

X

d

Page 22: Topology Control

22

Connectivity Theorem

• Order the edges in ER by length and induction on the rank in the ordering

For every edge in ER, there’s a corresponding path in G+ .

• If 150, then G+ preserves the connectivity of GR and the bound

is tight.

Page 23: Topology Control

23

Optimizations

• Shrink-back operation

“Boundary nodes” can shrink radius as long as not reducing cone coverage

• Asymmetric edge removal

If 120, remove all asymmetric edges

• Pairwise edge removal

If < 60, remove longer edge e2

e1

e2

A B

C

Page 24: Topology Control

24

Properties of Asymmetric Edge Removal

• Counterexample for = 120 +

R A B

60+/3

60

60-/3

Page 25: Topology Control

25

For 120 ( 2/3 )

• Asymmetric Edge Lemma

if d(A,B) R and (A,B) E, there must be a pair of nodes, W or X and node B, with distance less than d(A,B).

A B

W

X

Page 26: Topology Control

26

Asymmetric Edge Theorem

• Two-step inductions on ER and then on E

For every edge in ER , if it becomes an asymmetric edge in G , then there’s a corresponding path consisting of only symmetric edges.

• If 120, then G- preserves the connectivity of GR and the bound

is tight.

Page 27: Topology Control

27

Performance Evaluation

• Simulation Setup

100 nodes randomly placed on a 1500m-by-1500m grid. Each node has a maximum transmission radius 500m.

• Performance Metrics

Average Radius

Average Node Degree

Page 28: Topology Control

28

Average Radius

0

100

200

300

400

500

600

Basic With opt1 Withopt1&2

With allopts

Ave

rag

e ra

diu

s

Max power

150-deg

120-deg

Page 29: Topology Control

29

Average Node Degree

0

5

10

15

20

25

30

Basic With opt1 Withopt1&2

With allopts

Ave

rag

e n

od

e d

egre

e

Max power

150-deg

120-deg

Page 30: Topology Control

30

• In response to mobility, failures, and node additions

• Based on Neighbor Discovery Protocol (NDP) beacons

Joinu(v) event: may allow shrink-back

Leaveu(v) event: may resume “Hello” protocol

AngleChangeu(v) event: may allow shrink-back or resume “Hello” protocol

• Careful selection of beacon power

Reconfiguration

Page 31: Topology Control

31

• Distributed cone-based topology control algorithm that achieves maximum connected node set

If we treat all edges as bi-directional

150-degree tight upper bound If we remove all unidirectional edges

120-degree tight upper bound

• Simulation results show that average radius and node degree can be significantly reduced

Summary

Page 32: Topology Control

32

Outline

• Cone-based algorithm

• SPAN

• GAF-CEC

Page 33: Topology Control

33

SPAN

• Goal: preserve fairness and capacity & still provide energy savings

• SPAN elects “coordinators” from all nodes to create backbone topology

• Other nodes remain in power-saving mode and periodically check if they should

become coordinators

• Tries to minimize # of coordinators while preserving network capacity

• Depends on an ad-hoc routing protocol to get list of neighbors & the

connectivity matrix between them

• Runs above the MAC layer and “alongside” routing

Page 34: Topology Control
Page 35: Topology Control

35

Coordinator Election & Announcement

• Rule: if 2 neighbors of a non-coordinator node cannot reach each other

(either directly or via 1 or 2 coordinators), node becomes coordinator

• Announcement contention is resolved by delaying coordinator

announcements with a randomized backoff delay

• delay = ((1 – Er/Em) + (1 – Ci/(Ni pairs)) + R)*Ni*T

Er/Em: energy remaining/max energy

Ni: number of neighbors for node i

Ci: number of connected nodes through node i

R: Random[0,1]

T: RTT for small packet over wireless link

Page 36: Topology Control

36

Coordinator Withdrawal

• Each coordinator periodically checks if it should withdraw as a coordinator

• A node withdraws as coordinator if each pair of its neighbors can reach each other

directly of via some other coordinators

• To ensure fairness, after a node has been a coordinator for some period of time, it

withdraws if every pair of nodes can reach each other through other neighbors (even

if they are not coordinators)

• After sending a withdraw message, the old coordinator remains active for a “grace

period” to avoid routing loses until the new coordinator is elected

Page 37: Topology Control
Page 38: Topology Control

38

Performance Results

Page 39: Topology Control

39

Outline

• Cone-based algorithm

• SPAN

• GAF-CEC

Page 40: Topology Control

40

GAF/CEC: Geographical Adaptive Fidelity

• Each node uses location information (provided by some orthogonal

mechanism) to associate itself to a virtual grid

• All nodes in a virtual grid must be able to communicate to all nodes

in an adjacent grid

• Assumes a deterministic radio range, a global coordinate system

and global starting point for grid layout

• GAF is independent of the underlying ad-hoc routing protocol

Page 41: Topology Control

41

Virtual Grid Size Determination

• r: grid size, R: deterministic radio range

• r2 + (2r)2 <= R2

• r <= R/sqrt(5)

Page 42: Topology Control

42

Parameters settings

• enat: estimated node active time

• enlt: estimated node lifetime

• Td,Ta, Ts: discovery, active,

and sleep timers

• Ta = enlt/2

• Ts = [enat/2, enat]

• Node ranking:

Active > discovery (only one node active per grid)

Same state, higher enlt --> higher rank (longer expected time first)

Node ids to break ties

Page 43: Topology Control

43

Performance Results

Page 44: Topology Control

44

CEC

• Cluster-based Energy Conservation

• Nodes are organized into overlapping clusters

• A cluster is defined as a subset of nodes that are mutually

reachable in at most 2 hops

Page 45: Topology Control

45

Cluster Formation

• Cluster-head Selection: longest lifetime of all its neighbors

(breaking ties by node id)

• Gateway Node Selection:

primary gateways have higher priority

gateways with more cluster-head neighbors have higher priority

gateways with longer lifetime have higher priority

Page 46: Topology Control

46

Network Lifetime