1

Network Coding: Theory and

Practice
Apirath Limmanee
Jacobs University
2
Overview
Theory
Max-Flow Min-Cut Theorem
Multicast Problem
Network Coding
Practice
3
Max-Flow Min-Cut Theorem
Definition
Graph
Min-Cut and Max-Flow
4
Definition
(From Wiki) The max-flow min-cut
theoremis a statement in optimization
theory about maximal flows in flow networks
The maximal amount of flow is equal to the
capacity of a minimal cut.
In layman terms, the maximum flow in a
network is dictated by its bottleneck.
5
Graph
Graph G(V,E): consists of a set
and a set
V consists of sources, sinks, and other nodes
A member e(u,v) of E has a to
send information from u to v
A
D
S
B
C
T
V of vertices
E of edges.
S
A B
D C
T
3 3
3 3
2
2
2
4
capacity c(u,v)
6
Min-Cuts and Max-Flows
Cuts: Partition of vertices into two sets
Size of a Cut = Total Capacity Crossing the Cut
Min-Cut: Minimum size of Cuts = 5
Max-Flows from S to T
Min-Cut = Max-Flow
S
A B
D C
T
3 3
3 3
2
2
2
4 S
A
D
3 3
3
2
2
B
C
T
3
3
2
2
4
3
2
3
2
3
2
1
7
Multicast Problem
Butterfly Networks: Each
edges capacity is 1.
Max-Flow from A to D = 2
Max-Flow from A to E = 2
Multicast Max-Flow from
A to D and E = 1.5
Max-Flow for each
individual connection is
not achieved.
A
B C
F
G
D E
8
Network Coding
Introduction
Linear Network Coding
Transfer Matrix
Network Coding Solution
Connection between an Algebraic Quantity
and A Graph Theoretic Tool
Finding Network Coding Solution
9
Introduction
Ahlswede et al. (2000)
With network coding,
every sink obtains the
maximum flow.
Li et al. (2003)
Linear network coding
is enough to achieve the
maximum flow
A
B C
F
G
D E
b
1
b
2
b
1
b
1
b
2
b
2
b
1
+b
2
b
1
+b
2
b
1
+b
2
10
Linear Network Coding
Random Processes in a Linear Network
Source Input:
Info. Along Edges:
Sink Output:
Relationship among them
{ ; ),... , ( ), , ( ) , (
1 0
l v x l v x l v X =
The index is
a time index
{ ; ),... ( ), ( ) (
1 0
e y e y e Y =
{ ; ),... , ( ), , ( ) , (
1 0
l v z l v z l v Z =
_ _
= =
+ =
) ( ) ( :
,
) (
1
,
) ( ) , ( ) (
e tail e head e
e e
v
l
e l
e Y l v X e Y

_
=
=
v e head e
j e
e Y j v Z
) ( :
,
) ( ) , (
e comes
out of v
Weighted
Combination of
processes
generated at v
Weighted
Combination of
processes from
adjacent edges of e
Weighted
Combination
from all
incoming edges
11
Transfer Matrix
e
4
v
1
v
2
v
3
v
4
e
1
e
2
e
3
e
5
e
6
e
7
) 1 , (
1
v X
) 2 , (
1
v X
) 3 , (
1
v X
) 1 , (
4
v Z
) 2 , (
4
v Z
) 3 , (
4
v Z
) 3 , ( ) 2 , ( ) 1 , ( ) (
1 1 1
, 3 , 2 , 1 1
v X v X v X e Y
e e e
+ + =
) 3 , ( ) 2 , ( ) 1 , ( ) (
2 2 2
, 3 , 2 , 1 2
v X v X v X e Y
e e e
+ + =
) 3 , ( ) 2 , ( ) 1 , ( ) (
3 3 3
, 3 , 2 , 1 3
v X v X v X e Y
e e e
+ + =
) ( ) ( ) (
2 , 1 , 4
4 2 4 1
e Y e Y e Y
e e e e
+ =
) ( ) ( ) (
2 , 1 , 5
5 2 5 1
e Y e Y e Y
e e e e
+ =
) ( ) ( ) (
4 , 3 , 6
6 4 6 3
e Y e Y e Y
e e e e
+ =
) ( ) ( ) (
4 , 3 , 7
7 4 7 3
e Y e Y e Y
e e e e
+ =
) ( ) ( ) ( ) 1 , (
7 1 , 6 1 , 5 1 , 4
7 6 5
e Y e Y e Y v Z
e e e
+ + =
) ( ) ( ) ( ) 2 , (
7 2 , 6 2 , 5 2 , 4
7 6 5
e Y e Y e Y v Z
e e e
+ + =
) ( ) ( ) ( ) 3 , (
7 3 , 6 3 , 5 3 , 4
7 6 5
e Y e Y e Y v Z
e e e
+ + =
)) 3 , ( ), 2 , ( ), 1 , ( (
1 1 1
v X v X v X x Let =
)) 3 , ( ), 2 , ( ), 1 , ( (
4 4 4
v Z v Z v Z z=
M x z =
B A M
e e e e
e e e e e e e e e e
e e e e e e e e e e

=
7 3 6 3
7 4 4 2 6 4 4 2 5 2
7 4 4 1 6 4 4 1 5 1
, ,
, , , , ,
, , , , ,
0

=
3 2 1
3 2 1
3 2 1
, 3 , 3 , 3
, 2 , 2 , 2
, 1 , 1 , 1
e e e
e e e
e e e
A

=
3 , 2 , 1 ,
3 , 2 , 1 ,
3 , 2 , 1 ,
7 7 7
6 6 6
5 5 5
e e e
e e e
e e e
B



12
Network Coding Solution
We want
Choose to be an
identity matrix.
Choose B to be the
inverse of

=
3 , 2 , 1 ,
3 , 2 , 1 ,
3 , 2 , 1 ,
7 7 7
6 6 6
5 5 5
e e e
e e e
e e e
B



,
3 2 1
3 2 1
3 2 1
, 3 , 3 , 3
, 2 , 2 , 2
, 1 , 1 , 1

=
e e e
e e e
e e e
A



B A M
e e e e
e e e e e e e e e e
e e e e e e e e e e

=
7 3 6 3
7 4 4 2 6 4 4 2 5 2
7 4 4 1 6 4 4 1 5 1
, ,
, , , , ,
, , , , ,
0


M x z =
x z=
A

7 3 6 3
7 4 4 2 6 4 4 2 5 2
7 4 4 1 6 4 4 1 5 1
, ,
, , , , ,
, , , , ,
0
e e e e
e e e e e e e e e e
e e e e e e e e e e



NETWORK CODING
SOLUTION EXISTS IF
DETERMINANT OF M
IS NON-ZERO
13
Connection between an Algebraic
Quantity and A Graph Theoretic Tool
Koetter and Medard (2003): Let a linear
network be given with source node , sink
node , and a desired connection
of rate . The following three statements
are equivalent.
1. The connection is possible.
2. The Min-Cut Max-Flow bound is satisfied
3. The determinant of the transfer
matrix is non-zero over the Ring
, ) , )
' '
= , , , c

'
) (c R
, ) , ) ' ' = , , , c
) ( ) ( c R c R
M
[ ,... ,..., ,..., ..., F
, , , 2 j e e e e l

14
Finding Network Coding Solution
Koetter and Medard (2003): Greedy Algorithm
Let a delay-free Communication Network G and a
Solvable multicast problem be given with one
source and N receivers. Let R be the rate at which
the source generates information. There exists a
solution to the network coding problem in a finite
field with m
F
2

) 1 ( log
2
+ < NR m
15
Random Network Coding
Jaggi, Sanders, et al. (2003): If the field size
is at least , the encoding will be
invertible at any given receiver with prob. at
least , while if the field size is at least
then the encoding will be invertible
simultaneously at all receivers with prob. at
least .
/ E
1 / T E
1
16
Practical Issues
Network Delay
Centralized Knowledge of Graph Topology
Packet Loss
Link Failures
Change in Topology or Capacity
17
Thank You
18
You Are Welcome.