Competitive Time and Traffic Analysis of Position-based Routing using a Cell-Structure
-
Upload
cailin-miles -
Category
Documents
-
view
37 -
download
0
description
Transcript of Competitive Time and Traffic Analysis of Position-based Routing using a Cell-Structure
Stefan Rührup 1
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and Complexity
Competitive Time and Traffic Analysisof Position-based Routing
using a Cell-Structure
Stefan Rührup and Christian Schindelhauer
Heinz Nixdorf Institute
University of Paderborn
Germany
IEEE WMAN‘05
Stefan Rührup 2
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityOutline
• Part I: Topology control for position-based routing
– Position-based routing: greedy forwarding and recovery
– Topology issues in position-based routing
– Abstracting from graph theory: the cell structure approach
• Part II: Performance measures and algorithms
– Competitive performance measures
– Single-path versus multi-path routing strategies
Stefan Rührup 3
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and Complexity
Part I
Topology Control for Position-based Routing
Part I
Topology Control for Position-based Routing
Stefan Rührup 4
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityPosition-based routing in a nutshell
Given: Source, location of the destination
Task: Deliver a message to the destination
Assumptions:
• A node can determine its own position
• Each node knows the positions of the neighbors
• The position of the target is known
transmission range
source
target (x,y)
Stefan Rührup 5
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityGreedy forwarding and recovery (1)
• With position informationone can forward a message in the "right" direction(greedy forwarding)
Example:
s
t
no routing tables, no flooding!
transmissionrange
progress boundary (circle around the
destination)
Stefan Rührup 6
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and Complexity
barrierbarrier??
Greedy forwarding and recovery (2)
• Greedy forwarding is stopped by barriers (local minima)• Recovery strategy: Traverse the border of a barrier
... until a forwarding progress is possible (right-hand rule)
transmissionrange
s
t
greedy
recoverygreedy
routing time depends on the size of barriers!
right-hand ruleneeds planartopology!
Stefan Rührup 7
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityThe Cell Structure
transmission radius(Unit Disk Graph)
v
Define a grid consisting of l l squaresDefine a grid consisting of l l squares
Stefan Rührup 8
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityThe Cell Structure
transmission radius(Unit Disk Graph)
v
nodes exchange beacon messages node v knows positions of ist neighbors
nodes exchange beacon messages node v knows positions of ist neighbors
Stefan Rührup 9
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityThe Cell Structure
v
node cell link cell barrier cell
each node classifies the cells in ist transmission range
each node classifies the cells in ist transmission range
Stefan Rührup 10
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityThe Cell Structure
v
node cell link cell barrier cell
each node includes the classification in its beacon messages (only constant overhead)
each node includes the classification in its beacon messages (only constant overhead)
Stefan Rührup 11
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityRouting based on the Cell Structure
• Routing based on the cell structure uses cell pathscell path = sequence of orthogonally neighboring cells
• Paths in the original network (here: unit disk graph) and cell paths are equivalent up to a constant factor
• no planarization strategy needed(required for recovery using the right-hand rule)
Stefan Rührup 12
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and Complexity
node cell link cell barrier cell
Routing based on the Cell Structure
v
virtual forwarding using cellsvirtual forwarding using cells
w
physical forwarding from v to w, if visibility range is exceeded
physical forwarding from v to w, if visibility range is exceeded
Stefan Rührup 13
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and Complexity
Part II
Performance Measures and Algorithms
Part II
Performance Measures and Algorithms
Stefan Rührup 14
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityPerformance Measures
• barriers make routing difficult• what is the worst case scenario?
it depends ...
• how difficult is a scenario?• what would the best algorithm do?
comparative ratios
Stefan Rührup 15
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityHow difficult is a scenario?
barrier
Stefan Rührup 16
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityHow difficult is a scenario?
perimeterperimeter
barrier
perimeter (p) = number of border cells
Stefan Rührup 17
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityWhat would the best algorithm do?
length of shortest barrier-free cell path (h)
Stefan Rührup 18
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and Complexity
• competitive ratio:
• competitive time ratio of a routing algorithm– h = length of shortest barrier-free path– algorithm needs T rounds to deliver a message
Competitive Ratio
solution of the algorithm
optimal offline solution cf. [Borodin, El-Yanif, 1998]
h
T
single-path
„“
Stefan Rührup 19
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and Complexity
• optimal (offline) solution for traffic:h messages (length of shortest path)
• this is unfair, because ...– offline algorithm knows the barriers– but every online algorithm has to pay
exploration costs• exploration costs:
sum of perimeters of all barriers (p)
• comparative traffic ratio cf. [Koutsoupias, Papadimitriou 2000]
Comparative Ratios
M = # messages usedh = length of shortest pathp = sum of perimeters
h+p
Stefan Rührup 20
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityComparative Ratios
• measure for time efficiency:
competitive time ratio
• measure for traffic efficiency:
comparative traffic ratio
• Combined comparative ratio
time efficiency and traffic efficiency
Stefan Rührup 21
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityAlgorithms under Comparative Measures
• Sinlge-path strategies:no parallelism, traffic-efficient (time = traffic)example: GuideLine/Recovery– follow a guide line connecting source and target– traverse all barriers intersecting the guide line
Time and Traffic:
• Multi-path strategies: speed-up by parallel exploration, increasing trafficexample: Expanding Ring Search– start flooding with restricted search depth– if target is not in reach then
repeat with double search depth
Time: Traffic:
Stefan Rührup 22
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityAlgorithms under Comparative Measures
GuideLine/Recovery (single-path)
Expanding Ring Search (multi-path)
traffictime
scenario
maze
open space
GuideLine/Recovery (single-path)
Expanding Ring Search (multi-path)
time ratio
trafficratio
combinedratio
Is that good?
It depends ... on the
Stefan Rührup 23
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityThe Alternating Algorithm
... uses a combination of both strategies:
1. i = 1
2. d = 2i
3. start GuideLine/Recovery with time-to-live = d3/2
4. if the target is not reached thenstart Flooding with time-to-live = d
5. if the target is not reached theni = 2 · igoto line 2
Combined comparative ratio:
Stefan Rührup 24
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and ComplexityConclusion
• cell structure abstracts from graph theoretical issues
• neighborhood information (= cell classification) causes only constant overhead in beacon messages
• implicit planarization,well-suited for position-based routing
• comparative performance measuresin relation to the difficulty of the scenario (optimal distance & perimeter of barriers)
• time and traffic efficiency
Stefan Rührup 25
HEINZ NIXDORF INSTITUTEUniversity of Paderborn, Germany
Algorithms and Complexity
Thank you for your attention!
Questions ...
Thank you for your attention!
Questions ...
Stefan Rü[email protected].: +49 5251 60-6722Fax: +49 5251 60-6482
Algorithms and ComplexityHeinz Nixdof InstituteUniversity of PaderbornFürstenallee 1133102 Paderborn, Germany