Title: Optical Multicasting for Interactive Realtime Application in Sparse Splitting Optical Networks
1Optical Multicasting for Interactive Real-time
Application in Sparse Splitting Optical Networks
APAN Network Research Workshop 2007
- Ju-Won Park, Hyunyong Lee, and JongWon Kim
- 2007/ 08/ 27
2Contents
- Introduction
- Related Work
- Constrained Optical Multicast Routing
- Problem statement
- The proposed light-tree construction algorithm
- Experiment Results
- Conclusion
3Introduction
4Multicast over WDM networks
- Multicast in IP over WDM Networks
- IP layer multicast
- Multicast via WDM unicast
- WDM layer multicast
- Multicast tree constructed by the IP layer can
make copies of a data packet and transmit a copy
to each of its child - Require O/E/O conversion
- Undesirable
- Inefficient
- Long latency
5Multicast over WDM networks
- Construct a virtual topology consisting of a set
of lightpaths from the multicast source to each
destination (b) - Using multiple unicasts
- Inefficient bandwidth large multicast session
- WDM switches make copies of data packets in the
optical domain via light splitting (c) - More desirable transmission to different
destinations can now share bandwidth on common
link - Useful to support high-bandwidth multicast
application such as HDTV. - WDM layer multicast potential advantages
- Knowledge of the physical topology more
efficient multicast routing is possible - Light splitting is more efficient than copying
packets - Avoid the electronic processing bottleneck
- Support of coding format and bit-rate
transparency across both unicast and multicast
6Related Work
7Related Work
- The main mechanism of transport over optical
network is light-path, a point to point all
optical channel connecting from source to
destination. - To incorporate optical multicasting capability, a
light-tree, light-forest concept is introduced. - The problem of constructing a light-tree that
spans a given source and a set of destinations is
similar to the Steiner tree problem which is
known to be NP-complete - Consider several new issues and complexities for
QoS provisioning of optical multicasting - Sparse splitting (X. Zhang, J. Wei and C. Qiao,
Constrained Multicast Routing in WDM Networks
with Sparse Light Splitting, in J. of Lightwave
Technology, vol. 18, no. 12, December 2002.) - Power constraint (Y. Xin and G. Rouskas,
Multicast routing under optical layer
constraints, In Proc. of INFOCOM 2004) - Delay boundary (M. Chen, S.Tseng, B. Lin,
Dynamic multicast routing under delay
constraints in WDM networks with heterogeneous
light splitting capabilities, in Computer
Communications 29 (2006) 1492-1503)
8Constrained Optical Multicast Routing
- Problem statement
- The proposed light-tree construction algorithm
9Problem Statement
- Sparse splitting optical network
- MC (multicast capability) node
- MI (multicast in-capability) node
- We define a delay function which assigns a
nonnegative weight to each link the network - To deliver interactive real-time application via
light-tree, we consider three parameters - Adequate signal quality power constraint
- End-to-end delay boundary
- inter-destination delay variation boundary
10Constrained Optical Multicast Routing
- Goal
- Every member of session is connected
- Satisfy the delay and inter-destination delay
variation tolerance - Balanced tree to guarantee a certain level of
optical signal power - The way
- Adopt hierarchical approach
11Constrained Optical Multicast Routing
- Make multicast backbone network
- Build the auxiliary MC network as referred as
multicast backbone network, - Every MC node is included.
- Adjacent MC node is connected using logical link
if there is available wavelength on the path. If
there are multiple path between MC nodes, the
shortest path is selected. - The delay of logical link is equal to the delay
summation of path
12Constrained Optical Multicast Routing
13Constrained Optical Multicast Routing
- Build the light-tree based on application
requirement - Source searches the MC node which is nearest from
source as referred to primary MC node. - The primary MC node is unique of each session
- Build the light-tree which has primary MC node as
root in multicast backbone network based on
constraints.
14Constrained Optical Multicast Routing
15Constrained Optical Multicast Routing
- Each destination selects a adequate MC node
- The MC selection by receiver is a key to
construct feasible light-tree - Each MI node finds the subset of on-tree MC nodes
which satisfy the delay boundary - MI node chooses the MC node which has minimum
fanout in subset and then, join the light-tree by
connection with selected MC node
16Constrained Optical Multicast Routing
17Constrained Optical Multicast Routing
- Completed light-tree meets the delay boundary
with balanced aspect. - It does not satisfy the inter-destination delay
variation boundary. - Reduce the inter-destination delay variation by
swapping MI nodes
18Constrained Optical Multicast Routing
19Constrained Optical Multicast Routing
- Advantages
- Source need not know about the location of
destinations. - Every destination need not find the minimum cost
path from itself to source. It just must find
the location of MC node which satisfies
application requirement. - Simple construction of member-only light-tree
- The procedure of joining the light-tree is only
performed at member. - The procedure of dynamic addition or deletion of
members in a group is simple. - Join The node which wants to join in the
multicast session can be connected to its nearest
MC node. - Leave The node which wants to leave can be
disconnected send the prune message to connected
MC node.
20Experiment Results
21Experiment Results
22Experiment Results
23Conclusion
24Conclusion Future Work
- To support multicast in optical network
- a balanced light-tree to guarantee signal quality
- Delay and inter-destination delay variation along
all source-destination paths in the tree should
be bounded in sparse splitting optical network. - The proposed algorithm is heuristic approach to
obtain the feasible light-tree - Wavelength assignment algorithm should be
explored in future research. - Minimize wavelength cost
25QA