Today’s Menu

Dominant Pruning (DP)

DP and its variations Partial Dominant Pruning (PDP)

Total Dominant Pruning (TDP)

Multi-cover Dominant Pruning (MDP)

Prof. Ashikur Rahman CSE 6811: Wireless Ad 1 Prof. Ashikur Rahman CSE 6811: Wireless Ad 2
hoc networks hoc networks

Dominant pruning (DP) Some terminology

 Proactive
 A node selects some of its  Needs 2-hop information
neighbors for forwarding to  Periodic beacon messages
make sure that all of its 2-hop can be used to generate 2-hop
neighbors receive the packet. u neighbor info. u
 The selected set of nodes are  Let:
called forward list and  N(u) = nodes exactly 1-hop away
specified in the packet header  N(N(u))= nodes either 1-hop or 2-
 While selecting 1-hop hop away
neighbors, a node tries to  N(N(u)) – N(u) = nodes exactly
engage as minimum number 2-hops away
as possible

Prof. Ashikur Rahman CSE 6811: Wireless Ad 3 Prof. Ashikur Rahman CSE 6811: Wireless Ad 4
hoc networks hoc networks

Example
Dominant pruning explained

1 2
 The originator

3 4 5

Nodes  Two cases

1-hop or
2-hops
away  Intermediate
7 8 forwarding node
6
N(N(4)) = N({1,2,3,4,5,7})

N(4) = {1,2,3,4,5,7} = {exactly 1-hop} U {exactly 2-hop}

= {1,2,3,4,5,7} U {6,8} 5 Prof. Ashikur Rahman CSE 6811: Wireless Ad 6
hoc networks
Prof. Ashikur Rahman CSE 6811: Wireless Ad
hoc networks = {1,2,3,4,5,6,7,8}

1
 Originator case explained
Originator case explained
 Suppose node v is the
originator of a broadcast packet
B(φ, v) = N(v) Uv = N(N(v)) – N(v)
N(v)  Node v creates a list of neighbors who
should forward after receiving from v.

 Here is how v creates the forward list

v All its 1-hop neighbors in N(v) will
receive when v will broadcast. v
N(N(v))  Among its all 1-hop and 2-hop neighbors
Bv the following set of nodes Uv haven’t
received yet: Uv= N(N(v))- N(v)
Uv
= N(v)  The forward list is the minimum subset
of Bv such that each node in Uv is neighbor
= N(N(v))- N(v) of some node in Bv. Neighbors Neighbors of neighbors
that are not neighbors of v
7 8
Prof. Ashikur Rahman CSE 6811: Wireless Ad Prof. Ashikur Rahman CSE 6811: Wireless Ad
hoc networks hoc networks

DP Algorithm in action DP Algorithm in action

Neighbors

1 Neighbors
of neighbors
1 1 2
2 that are not
2 neighbors 4 5
3
6
3 4 5
3 6 7 8
4

Two solutions: {3,5}

5 8 and {7}
Engage as minimum as
7 8 possible
6
7
N(N(4)) = N(4) U {6,8}
N(4) = {1,2,3,4,5,7}
={1,2,3,4,5,6,7,8}
9 Prof. Ashikur Rahman CSE 6811: Wireless Ad 10
Prof. Ashikur Rahman CSE 6811: Wireless Ad hoc networks
hoc networks

Intermediate Forwarding node case Can be mapped to a set cover

problem
 Suppose node v received a
 Node v creates a list of neighbors who
broadcast from u.
should re-forward after receiving from v.
N(u) B(u, v) = N(v) –N(u) Uv= N(N(v))- N(v) –N(u)
N(v)
 Here is how v creates the forward list
 Among all its 2-hop neighbors, nodes in
N(u) already received the broadcast from u.
u v
 All its 1-hop neighbors in N(v) will v
receive when v will broadcast.
N(N(v))
 Among its all 1-hop and 2-hop neighbors
B(u, v) the following set of nodes Uv haven’t
received yet: Uv= N(N(v))- N(v) –N(u)
Uv
= N(v) –N(u)  The forward list is the minimum subset of
B(u,v) such that each node in Uv is Neighbors Neighbors of neighbors
= N(N(v))- N(v) –N(u) neighbor of some node in B(u,v). that are not neighbors of v
11 12
Prof. Ashikur Rahman CSE 6811: Wireless Ad
Prof. Ashikur Rahman CSE 6811: Wireless Ad and u
hoc networks
hoc networks

2
DP algorithm DP Algorithm in action

1 2

3 4 5

7 8
6

N(u) = {1,2,3,4,5,7} N(N(u)) = N(u) U {6,8}

Prof. Ashikur Rahman CSE 6811: Wireless Ad 13

={1,2,3,4,5,6,7,8} 14
Prof. Ashikur Rahman CSE 6811: Wireless Ad
hoc networks
hoc networks

DP Algorithm in action (for

Node 7) DP Algorithm in action
N(N(v))- N(v) –N(u)
1
1 2 N(N(7))- N(7) –N(4)

2
{1, 2, ,3 ,4 ,5, 6, 7, 8}
– {4, 6, 7, 8}
3 4 5 – {1, 2, 3, 4, 5, 7}

4 3
Empty set

5 1 2
7 6
7 8 4 5
6 3

8
6 7 8

Prof. Ashikur Rahman CSE 6811: Wireless Ad 15 Prof. Ashikur Rahman CSE 6811: Wireless Ad 16
hoc networks hoc networks

DP Algorithm in action A slight modification

N(N(v))- N(v) –N(u)

N(N(7))- N(7) –N(4) 1 2

{1, 2, ,3 ,4 ,5, 6, 7, 8}
– {4, 6, 7, 8}
– {1, 2, 3, 4, 5, 7}
3 4 5

Empty set

1 2
7 6
3 4 5 8
6 7
8
6 7 8
9
Prof. Ashikur Rahman CSE 6811: Wireless Ad 17 Prof. Ashikur Rahman CSE 6811: Wireless Ad 18
hoc networks hoc networks

3
 DP Algorithm in action Further optimization in DP!

B(u, v) Uv
Optimization is
only possible if
7 6 we can reduce
size of U set !

8 9 v

Prof. Ashikur Rahman CSE 6811: Wireless Ad 19 Prof. Ashikur Rahman CSE 6811: Wireless Ad 20
hoc networks hoc networks

Basic Idea N(u) N(v) PDP N(u) N(v)

Case 1:
u v u v
Neigbor
of a
common
N(N(v)) N(N(v))
neighbor
B(u, v) B(u, v)
X = N(u) ∩ N(v)
Uv P = N(X) Uv
= N(v) –N(u) = N(v) –N(u)

= N(N(v))- N(v) –N(u) = N(N(v))- N(v) –N(u) - P

Prof. Ashikur Rahman CSE 6811: Wireless Ad 21 Prof. Ashikur Rahman CSE 6811: Wireless Ad 22
hoc networks hoc networks

TDP N(u) N(v) Example in DP

N(N(u))
1
3
Case 2: 2
u v
Common 5 4
in both
N(N(u)) 6
N(N(v)) 7
and
N(N(v)) B(u, v) 8
9 11
Uv 10 12
= N(v) –N(u)
U = N(N(v) ) – N(N(u))

= N(N(v) ) – N(N(u))
Prof. Ashikur Rahman CSE 6811: Wireless Ad 23 Prof. Ashikur Rahman CSE 6811: Wireless Ad 24
hoc networks hoc networks

4
Example in DP Example in DP
1
1
2
3
1 3 2
3 5 4
2 5
5 4
4 6
6 6 7
7 7
8 8
9 11 8
9 11
10 12
10 12
9
11

Prof. Ashikur Rahman CSE 6811: Wireless Ad 25 Prof. Ashikur Rahman CSE 6811: Wireless Ad 26
hoc networks 10 hoc networks

Example in DP In PDP
3
1
3
4
2
1 12
3 5 4
2
5 4 6
7 8 7
6
7
10 8
8
9 11 9 11
10 12
10 12
11

Prof. Ashikur Rahman CSE 6811: Wireless Ad 27 Prof. Ashikur Rahman CSE 6811: Wireless Ad 28
hoc networks hoc networks

In PDP In TDP
3
1
3
4
2
12 5 4

6
7 8 7
10 8
9 11

10 12
11

Prof. Ashikur Rahman CSE 6811: Wireless Ad 29 Prof. Ashikur Rahman CSE 6811: Wireless Ad 30
hoc networks hoc networks

5
 Is redundancy always a bad
In TDP thing?
3
Of course not! (think about “un-trusted” environment)

4  Network with selfish nodes

Untrusted
Network  Noisy environment
12
 Network with intentionally un-coopeartive nodes

Mica2 Mote
7 8
Example: Noisy
10 environment:
Level of Arsenic
monitoring in
11 Bangladesh.
A joint project by
1 UCLA, MIT and
BUET.
Prof. Ashikur Rahman CSE 6811: Wireless Ad 31 Prof. Ashikur Rahman CSE 6811: Wireless Ad 32
hoc networks hoc networks

Multi-cover Dominant Pruning

Trade off game (MDP)
B(u, v) Uv
 Un-trusted environment demands fault tolerant design of protocols! N(u) N(v)

 To make broadcasting algorithms fault tolerant we need increased redundancy!

u v
Minimize Increase Fault- v
Redundancy Tolerance
N(N(v))
B(u, v)
 Wouldn’t be nice if we could control redundancy with the help of some tunable Uv
parameter(s)?  Reformulate the problem as a set-multicover problem to make it fault-tolerant.
 Now the forward list is the minimum subset of B(u, v) such that each node in Uv
is neighbor of at least m nodes in B(u,v).
 m becomes the tunable parameter. With m = 1 the algorithm becomes dominant
pruning and m =∞ the algorithm becomes simple flooding.
Prof. Ashikur Rahman CSE 6811: Wireless Ad 33 Prof. Ashikur Rahman CSE 6811: Wireless Ad 34
hoc networks hoc networks

MDP Example MDP Example

12
2
12

2 3

3
4
4

6 11 11
10 5 1 5

6
7 8 10
7
9 1
8

13
13
Prof. Ashikur Rahman CSE 6811: Wireless Ad 35 9
Prof. Ashikur Rahman CSE 6811: Wireless Ad 36
hoc networks hoc networks

6
MDP Example MDP Example
2 2
12 12

3 3

4 4

11 11
5 5

6 6
10 10
7 7
1 1
8 8
13 13
9
Prof. Ashikur Rahman CSE 6811: Wireless Ad 37 9
Prof. Ashikur Rahman CSE 6811: Wireless Ad 38
hoc networks hoc networks

MDP Example MDP Example

2 2
12 12

3 3

4 4

11 11
5 5

6 6
10 10
7 7
1 1
8 8
13 13
9
Prof. Ashikur Rahman CSE 6811: Wireless Ad 39 9
Prof. Ashikur Rahman CSE 6811: Wireless Ad 40
hoc networks hoc networks

MDP Example MDP Example

12
2
12

3 2
3
4
4

11 6 11
5 10 5 1

6
10 7 8
7
1 9 Forwarding list =
8 { 3, 4, 7, 8}

13
9
Prof. Ashikur Rahman CSE 6811: Wireless Ad 41 13
Prof. Ashikur Rahman CSE 6811: Wireless Ad 42
hoc networks hoc networks