On Optimal Diversity in Network-Coding-Based Routing in Wireless Networks Qiao Xiang 1 , Hongwei Zhang 2 , Jianping Wang 3 , Guoliang Xing 4 , Shan Lin 5 and Xue Liu 1 1 School of Computer Science, McGill University, Canada 2 Department of Computer Science, Wayne State University, US 3 Department of Computer Science, City University of Hong Kong, China 4 Department of Computer Science and Engineering, Michigan State University, US 5 Department of Electrical and Computer Engineering, Stony Brook University, US April 29th, 2015 Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 1/ 41
44
Embed
On Optimal Diversity in Network-Coding-Based …cpslab.cs.mcgill.ca/people/qiaoxiang/files/slides/slides-ONCR... · On Optimal Diversity in Network-Coding-Based Routing in Wireless
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
1School of Computer Science, McGill University, Canada2Department of Computer Science, Wayne State University, US
3Department of Computer Science, City University of Hong Kong, China4Department of Computer Science and Engineering, Michigan State University, US5Department of Electrical and Computer Engineering, Stony Brook University, US
April 29th, 2015
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 1/ 41
Introduction
Outline
1 Introduction
2 System Model and Problem Definition
3 An Analytical Framework for NC-based Routing
4 Optimizing Diversity of NC-based Routing
5 ONCR: an Optimal NC-Based Routing Protocol
6 Performance Evaluation
7 Conclusion and Future Work
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 2/ 41
Introduction
Introduction
Network Coding (NC)
First proposed in wired networks
Provide benefits on throughput and robustness
Naturally extended into wireless environment
Network-Coding-based routing, e.g., MORE,
CodeOR, CCACK and MIXIT
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 3/ 41
Introduction
Network-Coding-Based Routing: An Example
Differences from Opportunistic Routing
Packets are divided into batches and encoded
No communication needed between forwardersEvery node broadcasts
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 4/ 41
Introduction
Network-Coding-Based Routing Protocols
Three key challenges:
How to select the set of forwarders?
Routing Diversity Control
When to stop broadcasting? ACK Scheme
When to start broadcasting? Rate Control
Existing protocols:
Focus on improving network throughput
Various ACK and rate control schemes are
proposed
Routing diversity control is overlooked
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 5/ 41
Introduction
Routing Diversity
Utilizing all routing diversity
High contention and collision
Compromising network throughput and data delivery cost
Not suitable for resource-constrained wireless networks, e.g.,sensor networks
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 6/ 41
Introduction
Our Contribution
An analytical framework for estimating the cost
of NC-based routing
A greedy optimal algorithm to minimize the costof NC-based routing
ONCR, a fully distributed minimal costNC-based routing protocol
Performance improvement of ONCR over
state-of-the-art protocols on sensor testbed
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 7/ 41
System Model and Problem Definition
Outline
1 Introduction
2 System Model and Problem Definition
3 An Analytical Framework for NC-based Routing
4 Optimizing Diversity of NC-based Routing
5 ONCR: an Optimal NC-Based Routing Protocol
6 Performance Evaluation
7 Conclusion and Future Work
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 8/ 41
System Model and Problem Definition
System Model and Problem Definition
System model
A directed graph G = (V ,E ) with one source S
and one destination T
Edge (i , j) ∈ E with link reliability Pij =1
ETXij
Node i has a forwarder candidate set FCSi , i.e.,
one-hop neighbors of i
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 9/ 41
System Model and Problem Definition
System Model and Problem Definition
EST-NC Problem
Let forwarder set FSi = FCSi for each node i
Estimate the total transmission cost to deliverK linear independent packets from S to T .
MIN-NC Problem
Determine the forwarder set FSi for each node i
Minimize the total transmission cost to deliver
K linear independent packets from S to T
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 10/ 41
An Analytical Framework for NC-based Routing
Outline
1 Introduction
2 System Model and Problem Definition
3 An Analytical Framework for NC-based Routing
4 Optimizing Diversity of NC-based Routing
5 ONCR: an Optimal NC-Based Routing Protocol
6 Performance Evaluation
7 Conclusion and Future Work
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 11/ 41
An Analytical Framework for NC-based Routing
An Analytical Framework for NC-based Routing
How does it work?
1 Define the whole forwarder set as a virtual
node VS
2 Compute the transmission cost from the S to VS
3 Sort forwarders in non-descending order of theirtransmission cost
4 Each forwarder only forwards its effective load
with corresponding cost
5 Sum up all transmission cost
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 12/ 41
An Analytical Framework for NC-based Routing
An example
CSVS(K ) =
K
PSVS
=K
1− (1− P1)(1 − P3)(1− P5)
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 13/ 41
An Analytical Framework for NC-based Routing
An Analytical Framework for NC-based Routing
How does it work?
1 Define the whole forwarder set as a virtual
node VS
2 Compute the transmission cost from the S to VS
3 Sort forwarders in non-descending order of theirtransmission cost
4 Each forwarder only forwards its effective load
with corresponding cost
5 Sum up all transmission cost
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 14/ 41
An Analytical Framework for NC-based Routing
An example
P2 ≥ P4 ≥ P6
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 15/ 41
An Analytical Framework for NC-based Routing
An Analytical Framework for NC-based Routing
How does it work?
1 Define the whole forwarder set as a virtual
node VS
2 Compute the transmission cost from the S to VS
3 Sort forwarders in non-descending order of theirtransmission cost
4 Each forwarder only forwards its effective load
with corresponding cost
5 Sum up all transmission cost
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 16/ 41
An Analytical Framework for NC-based Routing
An Analytical Framework for NC-based Routing
Definition
For a node j in the forwarder candidate set FCSi ,
the effective load Lj is defined as the number oflinear independent packets received by j but none of
the nodes in FCSi that has lower transmission costto the destination.
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 17/ 41
An Analytical Framework for NC-based Routing
An example
P2 ≥ P4 ≥ P6
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 18/ 41
An Analytical Framework for NC-based Routing
Packets Recevied
KSA =
KP1
1− (1− P1)(1 − P3)(1 − P5)
KSB =
KP3
1− (1− P1)(1 − P3)(1 − P5)
KSC =
KP5
1− (1− P1)(1 − P3)(1 − P5)
Effective Load
LA = KSA
LB = KS ′
B = KKSB
K(1− P1) = KS
B (1− P1)
LC = KS ′
C = KSC (1− P1)(1− P3)
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 19/ 41
An Analytical Framework for NC-based Routing
An example
Qiao Xiang et al. (McGill) IEEE INFOCOM’15 04/29/2015 20/ 41