Cost Changed Multicast
based on
General Coding Conditions
Wang Wenquan
5110309796
2014.5.10
contents
01 Background
02 Changed Cost
03 Free ride
04 Future Work
05 References
1、Background
Multicast
Multicast is communication between a single
sender and multiple receivers on a network.
Multicast saving network traffic and releasing the
sender from sending multiple copies of packets to
different receivers.
Shortest Path Tree(SPT) is often used in multicast
routing. For example, SBPT(shortest best path tree)、
DDSP(destination-driven shortest path)
1、Background
Network Coding
(a)shows the chain topology , and (b) shows the
topology. In both cases,R is the common relay node
for two flows. After receiving P1 and P2,R encodes
two packets and then broadcasts .
Shuo-Yen Robert Li in his famous article “Linear
Network Coding” indicates that network coding can
help to achieve the max capacity.
1、Background
GCC
The General Network Coding Conditions:
1) There exist upstream decode-capable nodes,
which can extract the intended native packet for the
node, on the considered flow.
2) There exist downstream acquisition nodes, which
can overhear enough packets (either native or
encoded) to decode, on other flows associated with
the encoding function at this node.
1、Background
GCC
The General Network Coding Conditions:
Extended Coding Graph
1、Background
GCC
Decoding-capable nodes
The nodes that are capable to decode the
encodes packet and retrieve the corresponding
native packet.
Acquisition nodes
The nodes that are not destinations but capable
to obtain those packets which are useful to
decode, either by overhearing or receiving.
2、 Changed Cost
In order to achieve high throughput and min cost, we
must try our best to Adding Coding Opportunities
We also notice that in most routing algorithm, flows
tend to pass through the nodes which has a lower
cost.
So we say:
A node who meets the coding conditions can decrease
the cost between its upstream and downstream.
Degree n: a node who meets the coding conditions
and can encode n packets.
2、 Changed Cost
Define
The flow L in node k is going to choose which node as
the next hop：if flow L pass through node i, and the
node i has a degree n, then
The new cost will less than the odd cost, which will
make the routing algorithm tend to choose node i.
2、 Changed Cost
2、 Changed Cost
What doer ∆ and err mean?
∆ : the cost caused by network coding, such as
computing complexity, transmission delay, IO
operations.
err: modification of new cost
3、Free Ride
Example:
1.Node k is going to choose the
next hop form his neighbors
2.Node a、b、j are direct
neighbors of node k
3.Node i、j、m、n are on the
same flow L
4.Node j meets the general
coding conditions and node n is
the decoding-capable code in
j’s downstream
3、Free Ride
If k chooses j as its next hop,
then the pack from k to n only
have a cost of Cost(k,j).
Because the packet can take a
free ride from j to n with flow L
already has paid the cost.
This suggests:
We can regard node n as the
direct neighbor of node k. In
such case, we call node n
“virtual neighbor” of k, and a
and b are “real neighbor” of k
3、Free Ride
What is the cost between k and
his virtual neighbor n ?
Where a b are the factors to
estimate the weight of the cost of k
and j and the cost of j and n,
We can not regard the cost of j and
n is zero, so we set a function f(x) to
estimate it.
3、Free Ride
Algorithm:
Initialize:
D(j) is the set of nodes, which are decoding-capable nodes and
in the downstream of node j.
R(k) is the set of nodes, which are real neighbors of node k.
V(k) is the set of nodes,which are virtual neighbors of node k.
Procedure
V(k)={}
for node j in N(k)
for node n in D(j)
do V(k)  n
end
end
4、Future Work
4、Future Work
1.In Changed Cost, is 1/n the best function? How
to get a suitable value of Δ and err to achieve
better result?
2. In Free Ride, how much will it improve the
multicast performance? And what is the best
function f(x) to estimate the cost of the node and
his virtual neighbor?
5、Reference
[1] R. Ahlswede, N. Cai, S.-Y. Li, and R. Yeung, “Network information
flow,” IEEE Transactions on Information Theory, vol. 46, no. 4, pp.
1204–1216, Jul. 2000.
[2] S.-Y. Li, R. Yeung, and N. Cai, “Linear network coding,” IEEE Transactions
on Information Theory, vol. 49, no. 2, pp. 371–381, Feb. 2003.
[3] P. Sanders, S. Egner, and L. Tolhuizen, “Polynomial time algorithms for
network information flow,” in Proceedings of the fifteenth annual ACM
symposium on Parallel algorithms and architectures, 2003.
[4] R. Koetter and M. M´edard, “An algebraic approach to network coding,”
IEEE Transactions on Networking, vol. 11, no. 5, pp. 782–795, Oct.
2003.
[5] X. Yin, Y. Wang, X. Wang, X. Xue, and Z. Li, “A graph minor perspective
to network coding: Connecting algebraic coding with network
topologies,” in Proceedings of IEEE INFOCOM, 2013.
[6] A. R. Lehman and E. Lehman, “Complexity classification of network
information flow problems,” in Proceedings of the fifteenth annual ACMSIAM
symposium on Discrete algorithms, ser. SODA ’04, 2004, pp.
142–150.
Thanks for the guidance of Mr. Tian and Mr. Wang!
Thank you for listening!
Q&A