Polylogarithmic Inapproximability of Radio Broadcast - PowerPoint PPT Presentation

1 / 18
About This Presentation
Title:

Polylogarithmic Inapproximability of Radio Broadcast

Description:

Radio Broadcast. Undirected graph, v V wants to broadcast ... The radio broadcast problem. Given: a graph G,v. Find: a minimum number of rounds schedule. ... – PowerPoint PPT presentation

Number of Views:28
Avg rating:3.0/5.0
Slides: 19
Provided by: yana9
Learn more at: https://crab.rutgers.edu
Category:

less

Transcript and Presenter's Notes

Title: Polylogarithmic Inapproximability of Radio Broadcast


1
PolylogarithmicInapproximability ofRadio
Broadcast
  • Guy Kortsarz
  • Joint work with Michael Elkin

2
Radio Broadcast
  • Undirected graph, v ? V wants to broadcast
  • A vertex receives the message if and only if
    exactly one its neighbors transmits

3
The radio broadcast problem
  • Given a graph G,v
  • Find a minimum number of rounds schedule.
  • Let opt denote the optimum number of rounds

4
Example
  • In the following graph the optimum is 3

R2
R1
R3
R3
Figure 1 opt 3
5
History
  • Chlamtac and Weinstein, 87
  • O(R? log2n) upper bound
  • Bar-Yehuda, Goldreich and Itai, Kowalski and
    Pelc, 04
  • O(R? log n log2n) upper bound
  • Alon, Bar-Noy, Lineal and Peleg, 89
  • R?(log2n) lower bound
  • Gaber and Mansour, 95
  • O(Rlog5n) upper bound
  • Elkin and Kortsarz, 04
  • RO(R1/2 log2n)O(Rlog4n) upper bound
  • Elkin and Kortsarz, 04
  • RO(R1/2 log n log3n)O(Rlog3n)
  • for planar graphs

6
Approximation status
Table1 The summary of previous and our results
Ref Type Bound
EK02 M.L.B. log n
KP04 M.U.B. log n, R ? log n
EK04 B O(R log4n)
KP04 A.U.B. R O(log2n), R?? log n
This paper A.L.B. R o(log2n)
7
Min-Rep
  • Input G(A,B,E)
  • Given A partition A ?Ai, B ?Bi

8
Goal
  • Choose overall few (representative) from X ? A ?
    B so that X is minimum, and
  • All superedges are covered

B3
Figure 3 An exact solution
9
The MIN-REP hardness result
  • In its full generality, due to Ran Raz
  • Yes instance is mapped to an exact cover
  • No instance every choice of complete cover needs
    average of
  • representatives per Ai, Bi

10
The star property
  • The hardness result holds even under the
    assumption of the star property

11
Set-Cover
  • Input B(V1, V2, H)
  • S ? V1 covers V2 if N(S) V2
  • Goal Minimum size V2 cover

12
The Lund and Yanakakis L.B
B
Figure 6 SET-COVER
13
The Lund and Yanakakis L.B
  • Yes instance An exact MIN-REP cover gives and
    exact cover
  • No instance In a no instance, every cover is
    large
  • log( M(A,B) ) gap

14
The reduction
B
15
A 3 rounds schedule for a YES instance
16
Witnesses for NO instance
Figure 9 Type 1 and Type 2 witnesses
17
  • Choose all deleted vertices as Type 1 witnesses
  • From every remaining round choose 2 witnesses.
    Type 2 witnesses
  • If of rounds is O (log n), then of witnesses
    is O (log n)
  • If v is not connected to all Type 1 witnesses,
    but is connected to all Type 2 witnesses, v
    doesnt get the message
  • Pr 1/pol for that
  • Use union bound over all schedules

18
Open problems
  • Prove O (R log2n) upper bound
  • If R O (log n) can we do better than log2n
    approximation?
  • Prove R O (log2n)(?)
  • or opt O (log2n)
Write a Comment
User Comments (0)
About PowerShow.com