Practical and Efficient Broadcast in Mobile Ad hoc Networks - PowerPoint PPT Presentation

1 / 13
About This Presentation
Title:

Practical and Efficient Broadcast in Mobile Ad hoc Networks

Description:

The Ohio State University. Practical and Efficient Broadcast in Mobile Ad hoc ... (Using PDP algo): Node v is chosen as. the forwarder by all the source nodes. ... – PowerPoint PPT presentation

Number of Views:103
Avg rating:3.0/5.0
Slides: 14
Provided by: Tan156
Category:

less

Transcript and Presenter's Notes

Title: Practical and Efficient Broadcast in Mobile Ad hoc Networks


1
Practical and Efficient Broadcast in Mobile Ad
hoc Networks
Infocom 2007 Paper
  • L. (Erran) Li, R. Ramjee, M. Buddhikot, S. Miller
  • Bell Labs, Lucent

Slides by Tan Apaydin
2
  • Problem Broadcasting packets to be received by
    all nodes.
  • Wired vs. Wireless
  • Observation Due to the broadcast nature of
    wireless media, not all nodes need to transmit.

3
PREVIOUS WORK On BROADCAST SUPPORT
  • Probabilistic Each node re-broadcast packet to
    its neighbors with a given forwarding probability
    (fp).
  • Challenge tune the fp.
  • Deterministic Predetermine neighbor nodes to
    forward the broadcast packet.
  • In this paper deterministic approach.

4
DETERMINISTIC APPROACH
  • Two popular schemes
  • Complete topology
  • 1 or 2 hop topology
  • First one above is not realistic for ad hoc
    mobile networks.
  • In this paper 2-hop topology info is kept.

5
What is presented in the paper?
  • 1-) Select forwarding nodes to prevent each node
    broadcasting each packet PDP algorithm which is
    already published by other authors.
  • 2-) NETWORK CODING Two algorithms for encoding
    packets.
  • - XOR Based
  • - Reed-Solomon Based (FEC forward error
    correction)

6
Partial Dominant Pruning (PDP)
  • N(u) Neighbors, N(N(u)) 2-hop neighbors
  • Node u sends packet to node v
  • v selects a forward list (min. nodes to
    rebroadcast to cover all nodes in N(N(v))
  • Nodes in N(u) already received
  • Nodes in N(v) will have when v broadcasts
  • Nodes in N(N(u) n N(v)) will also receive
  • Select forward node list from
  • N(N(v)) N(u) N(v) N(N(u) n N(v))
  • (A greedy set cover algorithm applied)

7
Neighbor Reception Table at node v
Source nodes r, s, t, u
- (Using PDP algo) Node v is chosen as the
forwarder by all the source nodes. - v will
cover their respective 2-hop neighborhood.
  • At least 1 node missing a pkt v makes 5
    transmissions.
  • Send (p1 XOR p2) 4 trans.
  • Reed Solomon 3 trans.

By sending nk bits, rcv can recover k bit
erasures. Send 3 pkts, rcvs recover up to 3
missing pkts.
8
CODEB Overview
  • Coding-based broadcast protocol
  • Like COPE, new layer btw IP MAC
  • Opportunistic listening exchange neighbor
    information, construct reception report tables.
  • Forwarder selection and pruning Forwarder set is
    stamped in the pkt header.
  • Opportunistic coding

9
- CODEB Algorithm - Line 5?
10
  • Problem is NP-hard
  • They apply a greedy algorithm.

Coding algorithm with XOR
11
  • The largest of pkts received by a neighbor is X
  • k P - X
  • At least k pkts need to be encoded

Coding algorithm with Reed-Solomon
As long as a node receives at least n-k distinct
native pkts (doesnt matter which set), node can
decode the rest of the missing pkts.
Vandermonde Matrix
12
  • How is packet decoding done?
  • - Similar to COPE
  • - Each node maintains a Packet Pool.
  • Theorem For a reliable network, all nodes
    receives the set of pkts injected into the
    network.

13
EXPERIMENTS
  • XOR up to 45 gain.
  • Reed-Solomon up to 60 gain.
  • Coding-based deterministic approach vs.
    Coding-based probabilistic approach
  • Deterministic performs better.
Write a Comment
User Comments (0)
About PowerShow.com