algorithms for radio networks winter term 2005/2006 21 dec 2005 10th lecture

16
1 HEINZ NIXDORF INSTITUTE University of Paderborn Algorithms and Complexity Christian Schindelhauer Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture Christian Schindelhauer [email protected]

Upload: phuc

Post on 13-Jan-2016

28 views

Category:

Documents


2 download

DESCRIPTION

Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture. Christian Schindelhauer [email protected]. Radio Broadcasting. Broadcasting A sender distributes a message to n radio stations Radio Broadcasting Undirected Graph G=(V,E) describes possible connections - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

1

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

Algorithms for Radio NetworksWinter Term 2005/2006

21 Dec 200510th Lecture

Christian Schindelhauer

[email protected]

Page 2: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

Algorithms for Radio Networks 2

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

Radio Broadcasting

Broadcasting– A sender distributes a message to n radio stations

Radio Broadcasting– Undirected Graph G=(V,E) describes possible connections

• If edge {u,v} exists, u can transmit to v and vice versa

• If no edge exists, then there is no reception and no interference

– One frequency, stations communicate in a round model

– If more than one neighbored station send at the same time, no signal is received (not even an interference signal)

Main problem: – Graph G=(V,E) is unknown to the participants– Distributed algorithm avoiding conflicts

Page 3: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

Algorithms for Radio Networks 3

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

Radio Broadcasting without ID

Theorem

There is no deterministic broadcasting algorithm for the radio broadcasting problem (without id)

Proof:

Consider the following graph:1. Blue node sends (at any time)

a message to the neighbors

2. As soon they are informed, they behave completely synchronously – because they use the same algorithm

– so, they send (or do not send) always at the same time

3. Red node does not receive any message.

Page 4: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

Algorithms for Radio Networks 4

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

A simple random algorithm (I)

Every station uses the following algorithmSimple-Random(t) begin if message m is available then for i ← 1 to t do r ← result of a fair coin toss (0/1 with prob. 1/2)

if r = 1 then send m to all neighbors fi od fi endTheorem

For appropriate c>1 we have: Simple-Random informs the complete network with probability of at least 1-O(nk) within time c 2Δ/Δ (D+ log n).

Page 5: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

Algorithms for Radio Networks 5

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

Extending the Deterministic Model

Model too restrictiv

New deterministic model:

– Every of the n players knows his unique id number from the set {1,..,n}

Probabilistic model:

– Die number n of players is known

– The maximal degree Δ is known

– But no ID is available

Page 6: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

Algorithms for Radio Networks 6

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

Decay (I)

Idee: randomized thinning out of the players

Decay(k,m)

begin

j ← 1

repeat

j ← j + 1

Send message to all neighborsr ← result of fair coin toss (0/1 with prob. 1/2)

until r=0 oder j > k

end

Page 7: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

Algorithms for Radio Networks 7

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

Decay (II)

d neighbors are informed All d neighbors start simultanously (k,m) P(k,d): Prob. that message is received by d neighbors within

at most k rounds:

Lemma

For d≥2 :

Page 8: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

Algorithms for Radio Networks 8

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

BGI-Broadcast[Bar-Yehuda, Goldreich, Itai 1987]

All informed players have synchronized round counters, i.e.

– Time is attached to each message

– and incremented in each round

BGI-Broadcast(Δ,) begin

k ← 2 log Δt ← 2 log (N/)wait until message arrives

for i ← 1 to t do

wait until (Time mod k) = 0

Decay(k,m)

od

end

Theorem

BGI-Broadcast informs all nodes with probability 1- in time O((D+log(n/)) log Δ)

Page 9: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

Algorithms for Radio Networks 9

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

Changing the Game: New Models

Probabilistic mode:

– Number n of players is known

– The maximal degree Δ is known

– But no ID Restriction: What if the maximal degree is not known?

– Corollary

• BGI-Broadcast informs all nodes with probability1- in time O((D+log(n/)) log n)

Determinististic model:

– Each of the n players knows a unique identifier (id) of the set {1,..,n} and knows n

Page 10: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

Algorithms for Radio Networks 10

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian SchindelhauerDeterminism versus Probabilism

TheoremFor every distributed deterministic Radio-Broadcasting algorithm using IDs there is a graph with D=2 that cannot be completely informed within time n-2.

TheoremBGI-Broadcast informs all nodes with probability 1- in time O((D+log(n/)) log Δ) for any e>0.

TheoremFor any constant >0 BGI-Broadcast informs all nodes of a graph with D=2 with probability 1- in time O((log n)2).

Page 11: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

Algorithms for Radio Networks 11

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

Decay

d neighbors are informed All d neighbors start simultanously (k,m) P(k,d): Prob. that message is received from d neighbors

within at most k rounds:

Lemma

For d≥2 :

Page 12: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

Algorithms for Radio Networks 12

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

Proof of Lemma (Part I)

P(k,d): Prob. that the message is received from d neighbors within at most k rounds

0 neighbored players are informed:

– P(1,0)= 0 Chance of being informed in the first round by nobody

– P(2,0)= 0

– P(3,0)= 0

– ... 1 neighbored player is informed:

– P(1,1)= 1 One player cannon cause any conflict

– P(2,1)= 1 stays informed in the next roundd

– P(3,1)= 1 etc.

– ...

Page 13: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

Algorithms for Radio Networks 13

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

Proof of Lemma (Part I)

P(k,d):–Prob. that the message is received from d neighbors within at most k rounds

2 neighbored players are informed:–P(2,1)= 0

•Two nodes send in the first round.•No chance

–P(2,2)= P(no player continues) P(1,0) +

P(one player continues) P(1,1) +

P(two players continue) P(1,1)= 1/4 P(1,0) + 1/2 P(1,1) + 1/4 P(2,1)= 0 + 1/2 + 0 = 1/2

Page 14: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

Algorithms for Radio Networks 14

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

Survey of Randomized Broadcasting Algorithms

Lower bounds for random algorithms concerning expected round time:– Alon, Bar-Noy, Linial, Peleg, 1991

(log2n) for diameter D=1– Kushiletz, Mansour, 1998

(D log (n/D)) Expected round time of random algorithms

– Gaber, Mansour, 2003 O(D+ log5 n) if the network is known

– Bar-Yehuda, Goldreich, Itai, 1992O((D+log n) log n) (presented

here)– Czumaj, Rytter, 2003: O(D log (n/D) + log2 n)– Bar-Yehuda, Goldreich, Itai, 1992

O(n log n) if D is unknown

Page 15: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

Algorithms for Radio Networks 15

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

Survey of Deterministic Algorithms

Lower bounds for deterministic algorithms concerning expected round time:

– Bar-Yehuda, Goldreich, Itai, 1992(n) (presented here)

Worst case time of deterministic algorithms– Chlebus, Gasieniec, Gibbons, Pelc, Rytter, 1999

O(n11/6)– Chlebus, Gasieniec, Östlin, Robson, 2000

O(n3/2)– Chrobak, Gasieniec, Rytter, 2001,

O(n log2 n)– Kowalski, Pelc, 2002 O(n log n log D)

Page 16: Algorithms for Radio Networks Winter Term 2005/2006 21 Dec 2005 10th Lecture

16

HEINZ NIXDORF INSTITUTEUniversity of Paderborn

Algorithms and ComplexityChristian Schindelhauer

Thanks for your attention!End of 11th lectureNext lecture: We 18 Jan 2006, 4pm, F1.110Next exercise class: Th 19 Jan 2006, 1.15 pm, F2.211 or Tu 24 Jan 2006, 1.15 pm, F1.110Next mini exam Mo 13 Feb 2006, 2pm, FU.511