Aligned Virtual Coordinates for Greedy Routing in WSNs

Ke Liu and Nael Abu-Ghazaleh

Dept. Of Computer Science SUNY Binghamton MASS, October 12, 2006

Outlines
Motivation of Virtual Coordinates System (VCS) Brief introduction to GPSR/GFG (Geographic Routing) Anomalies in VCS Intuition and Design of Aligned VCS Performance evaluation Conclusion
IEEE MASS 2006, October 12, 2006 2

Motivation of VCS
Geographic Routing Efcient for WSNs Stateless: no state information (info of sink and path) Localized Interactions (only info of one-hop neighbors) GR suffers from Voids and Localization Errors Virtual Coordinate Systems (based on connectivity info.) Better? Based on partial connectivity info. We show they suffer their own anomalies Quantization Error is a factor
IEEE MASS 2006, October 12, 2006 3

GPSR/GFG: Greedy Forwarding (GF)

Greedy Forwarding(GF)
IEEE MASS 2006, October 12, 2006 4

GPSR/GFG: GF may fail

Physical Void in GF
IEEE MASS 2006, October 12, 2006 5

Distance Map Show

Physical Distance to Node(21,20)

45 30 15 0 50 40 30 20 25 30 35 40 45 50

10 0 0 5 10

15

20

Distance Map of a physical hole

IEEE MASS 2006, October 12, 2006 6

Virtual Coordinates For Geometric Routing

Several nodes are elected to be anchors: one node per dimension;

Anchors broadcast Virtual Coordinate beacons;

Each other node forwards beacons, incrementing distance;

Each node obtains a VC based on recevied beacon values;

Distance measured in number of hops: integral value;

IEEE MASS 2006, October 12, 2006 7

Argued VCS (VCap)

3 anchors (a 3D VCS) are enough to map the physical coordinates

VC Zone can be avoided if density is high enough

VC Zone: nodes with the same VC values

VC Zones are connected with 3 anchors adapted (3D VCS)

Void (anomaly) ratio is reduced much

IEEE MASS 2006, October 12, 2006 8

Anomalies found in VCS

3D VCS is not enough to map

VC zones may be disconnected in 3D VCS

Anomaly ratio may be increased by VCS More routing (greedy forwarding) anomalies happen

IEEE MASS 2006, October 12, 2006 9

Anomalies in 3D VCS
6 5 21 (4 4 2) 16 (3 4 2) 11 (2 4 2) (1 4 3) 1 (0 4 4) 22 (4 4 1) 17 (3 3 1) 12 (2 3 2) (1 3 3) 2 (1 3 4) 23

24 (4 4 1) 19 (3 3 1) 14 (3 2 2) 9 (3 1 3) 4 (3 1 4)

(4 4 0) 18 (3 3 1) 13 (2 2 2) 8 (2 2 3) 3 (2 2 4)

25 (4 4 2) 20 (4 3 2) 15 (4 2 2) (4 1 3) 5 (4 0 4)Y

10

2 Units > Radio Range > 1.414 Units

0 1 2 3 4 5 6

Extended & Disconnected VC Zone Problems

Details can be found in previous work

IEEE MASS 2006, October 12, 2006 10

Anomalies in 3D VCS (virtual voids)

Virtual Distance (3D) to Node (1, 50)

Virtual Voids
60 40 20 0 50 45 10 40 35 30 20 25 20 30 15 10 40 5 0 50 0

Radio Range 1.5 Unit

Virtual voids even without physical void

IEEE MASS 2006, October 12, 2006 11

4D VCS? or Different Distance measurement?

4D VCS was proposed too (LCR)

Anomalies in 4D VCS were found in LCR; solution requires each data packet records each node along its path during forwarding

Different distance measurement was prosposed (BVR), Manhattan style distance, indicated as a better solution

IEEE MASS 2006, October 12, 2006 12

Existing solutions do not reduce anomalies

4D Euclidean Distance to Node(21,20) Radio Range 1.5 Unit
70

Virtual Voids 4D Manhattan Distance to Node 0

120 100 80 60 40 20 0 50 40

Radio Range 1.5 Units

Virtual Voids

0 50 40 30 20 45 50

10 0 0 5 10

15

20

25

30

35

40

30 20

10 0 0 5 10

15

20

25

30

35

40

45

50

Eclidean Distance in 4D VCS

Manhattan Distance in 4D VCS

IEEE MASS 2006, October 12, 2006 13

Reducing Anomalies Greedy Forwarding Better

In terms of Average Path Stretch
Density Neighbors # 3.92 7.76 11.60 19.13 26.57 33.94 44.84 62.66 73.17 Optimal SP 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 GF GeoCS 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 CR Perimeter Routing 16.9200 13.1800 18.0365 23.7324 29.7087 31.7657 37.4030 41.9031 46.2990

4D VCS 1.0000 1.0000 1.0000 1.0494 1.0251 1.0409 1.0545 1.0850 1.0926

4D AVCS (d 1) 1.0000 1.0000 1.0000 1.0073 1.0010 1.0050 1.0035 1.0100 1.0074

BVR BT 2.1903 2.0996 2.1875 2.1411 2.1459 2.1609 2.1512 2.1342 2.1311

LCR BT 1.0000 1.0000 1.0000 1.0525 1.0277 1.0480 1.0564 1.1198 1.0982

Observation: if we can increase the ratio of greedy forwarding, we may improve the routing performance: either path stretch or overhead
IEEE MASS 2006, October 12, 2006 14

Why Anomalies in VCS?

Virtual Coordinate values are integral: quantization error or noise increases requiring more precise values for VCs

No discrimination among nodes in range: forwarding dilema requiring in range discrimination

Mapping from a continuous space to a discrete space: less forwarding candidates requiring continuous space

IEEE MASS 2006, October 12, 2006 15

Aligned VCS: Intuition

Anchor 1 2 3
A B

Node A and B are different as forwarding nodes, since with different regions of neighbors in their range.
IEEE MASS 2006, October 12, 2006 16

Aligned VCS (AVCS)

AVC of a given node is computed as a function of its VC and neighbors VC

Simplest value: average of the neighbors integral virtual coordinate values AVC coordinates with depth d are decided by its neighbors aligned virtual coordinates with depth d 1 Original integral virtual coordinates are AVC with depth 0

IEEE MASS 2006, October 12, 2006 17

AVCS (contd)
Euclidean Distance to Node (2, 8) VCS Forwarding Void
16 8 0 20 16 12 8 18 20 14 16 12 10

Euclidean Distance to Node (2, 8)

15 10 5 0 20 16 12 8

4 0 0 2 4 6 8

4 0 0 2 4 6 8

18 20 14 16 12 10

X (RR = 2.5 Units)

X (RR = 2.5 Units)

Forwarding Voids in 4D VCS

Aligned VCS without forwarding voids

IEEE MASS 2006, October 12, 2006 18

Simulation
Metrics: Greedy Ratio: how many pathes do not face any anomalies Path Stretch: the average length of all path (both GF and CR) comapred to optimal solution (SP) Simulator: NS-2: for network with less than 400 nodes Customer: for network with 1600 or 2500 nodes Based more than 30 networks used for each scenarios
IEEE MASS 2006, October 12, 2006 19

AVCS Performance: Greedy Forwarding Ratio

GF on GeoCS GF on GeoCS with 20% Loc Error GF on GeoCS with 40% Loc Error GF on 3D VCS (VCap) GF on 4D VCS (LCR) GF on 4D Aligned VCS depth 1 GF on 4D Aligned VCS depth 2

Greedy Ratio

100% 90% 80% 70%

5/400

13/400

29/400

49/400

81/400

Physical (cycle) Void Size (RR=1.5 units)

IEEE MASS 2006, October 12, 2006 20

AVCS Performance: Greedy Forwarding Ratio over BVR

Original BVR on 4D VCS BVR on 4D Alinged VCS depth 1 BVR on 4D Alinged VCS depth 2 BVR on 4D Alinged VCS depth 3 BVR on 4D Alinged VCS depth 4

100% 90%

Greedy Ratio

80% 70% 60% 50% 40% 30% 5/400 13/400 29/400 49/400 81/400

Hole (cycle) Size (RR = 1.5 units)

IEEE MASS 2006, October 12, 2006 21

AVCS Performance: Path Stretch

1.35 Shortest Path GPSR on GeoCS GPSR with 20% Loc Error GPSR with 40% Loc Error GR on 4D VCS (LCR) GR on 4D Aligned VCS (depth=1)

Path Stretch to SP (=1.0)

1.30 1.25 1.20 1.15 1.10 1.05 1.00 5/400

13/400

29/400

49/400

81/400

Physical (cycle) Void Size (RR=1.5 units)

IEEE MASS 2006, October 12, 2006 22

AVCS Performance: Path Stretch over BVR

Original BVR on 4D VCS BVR on 4D Alinged VCS (depth 1) BVR on 4D Alinged VCS (depth 2) BVR on 4D Alinged VCS (depth 3) BVR on 4D Alinged VCS (depth 5) BVR on 4D Alinged VCS (depth 5)

Path Stretch to SP (=1.0)

2.2 2.0 1.8 1.6 1.4

5/400

13/400

29/400

49/400

81/400

Physical (cycle) Void Size (RR=1.5 units)

IEEE MASS 2006, October 12, 2006 23

AVCS Performance: Depth

100%

Greedy Ratio

97.5% 95% 92.5% 90% Depth 1 Depth 2 Depth 3 Depth 4

5/400

13/400

29/400

49/400

81/400

Physical (cycle) Void Size (RR=1.5 units)

IEEE MASS 2006, October 12, 2006 24

AVCS Performance: GF ratio with random deployment

100% 90% 80%

Greedy Ratio

70% 60% 50% 40% GF on GeoCS GF on GeoCS with 20% Loc Error GF on GeoCS with 40% Loc Error GF on 3D VCS (VCap) GF on 4D VCS (LCR) GF on 4D Aligned VCS depth 1 12 14 16 18 20

Normalized Density
IEEE MASS 2006, October 12, 2006 25

AVCS Performance: Path stretch with random deployment

1.50 1.45 GPSR on GeoCS GPSR on GeoCS with 20% Loc Error GPSR on GeoCS with 40% Loc Error GR on 4D VCS (LCR) GR on 4D Alinged VCS (depth 1)

Path Stretch to SP (=1.0)

1.40 1.35 1.30 1.25 1.20 1.15 1.10 1.05 1.00 11 12 13 14 15

16

17

18

19

20

21

Normalized Density
IEEE MASS 2006, October 12, 2006 26

Conclusions
Greedy Forwarding performs much better than complementary routing phase; Virtual Coordinates System with simple integral values create more anomalies than Geometric Routing; Aligned VCS help reduce anomalies, enhancing performance; Geometric Routing in VCS (AVCS) can provide equivalent, or even better performance, than geographic routing; Further, stateless routing can approach that of stateful routing protocols, such as shortest path routing.
IEEE MASS 2006, October 12, 2006 27

Thank You !
Code is available on my website http://www.cs.binghamton.edu/kliu

Questions?

IEEE MASS 2006, October 12, 2006 28

Back up : Multiple Physical voids

28

100% 90%

80% 70% 60% 50% 40% GF on GeoCS GF on GeoCS with 20% Loc Error GF on GeoCS with 40% Loc Error GF on 3D VCS (VCap) GF on 4D VCS (LCR) GF on 4D Aligned VCS depth 1 GF on 4D Aligned VCS depth 2 1 2 3 4 5 16

Average Path Length

26

Greedy Ratio

Shortest Path GPSR on GeoCS GPSR on GeoCS with 40% Loc Error GR on 4D VCS GR on 4D Aligned VCS depth 1

24

22

20

18 1 2 3 4 5 16

Number of Holes

Number of Holes (Physical Voids)

IEEE MASS 2006, October 12, 2006 29