Transcript i,j
A Multicast Mechanism
in WiMax Mesh Network
Jianfeng Chen, Wenhua Jiao, Pin Jiang, Qian Guo
Asia-Pacific Conference on Communications, 2006.
(APCC '06)
Mei-zhen chen
Outline
Introduction
Overview of WiMax Mesh Network
Architecture Framework for Multicast
Performance Evaluation
Conclusion
Introduction
The standard of IEEE 802.16 only defines the
multicast connection in PMP mode in one hop range.
The Connection ID (CID) used for the multicast
service is the same for all Subscribe Stations (SSs)
that participate in the multicast.
How to construct a multicast tree in mesh mode is
undefined in the standard.
Introduction (cont.)
A multicast tree construction mechanism is proposed
based on the unicast routing tree in WiMax mesh
network.
The broadcasting feature of omni-directional antenna
is used to improve the transmission efficiency.
Overview of WiMax Mesh Network
Two
scheduling methods:
Centralized scheduling
• Mesh BS is responsible for collecting bandwidth request
from SSs and managing resources.
Distributed scheduling
• A processing of three-way handshake.
Overview of WiMax Mesh Network
-assumption of the physical layer
Only time division mode (TDD) is supported in Mesh
mode.
The MAC layer is assumed to schedule data to time
division multiple access (TDMA).
This paper only considers the single carrier channel
for both uplink and downlink traffic.
Overview of WiMax Mesh Network
–control messages
In
centralized scheduling,
MSH-CSCF (Mesh centralized scheduling
Configuration) message carries the information of
channel configuration and routing tree information.
MSH-CSCH (Mesh centralized scheduling)
message carries the information of bandwidth
request and grant.
Overview of WiMax Mesh Network
–control messages (cont.)
BS
SSs
MSH-CSCF
(broadcast)
Forward to SSs
Routing tree
MSH-CSCH
Determine the amount of
granted resources for
each link
transmit resource request
(Request message)
MSH-CSCH
(Grant message)
propagate
along the route tree
The SSs determine its
actual uplink and downlink
transmission time for data
transmission
Architecture Framework for Multicast
-Multicast Architecture
Architecture Framework for Multicast
-Multicast Architecture (cont.)
Application/IP layer
This layer utilizes the services of the routing layer
to satisfy the multicast requirements of
applications.
It consists of two modules:
• Data packet transmit/receive controller
• Multicast session initiator/terminator
Architecture Framework for Multicast
-Multicast Architecture (cont.)
MAC
layer
The major module in this layer is tree construction
module.
The input information to make the decision come
from three modules:
• Multicast Information Handler
• Neighbor List Handler
MSH-NCFG (Mesh network Configuration) message
• Centralized Scheduling
Architecture Framework for Multicast
-Multicast Architecture (cont.)
Frame
Structure layer
The main function of this layer is to encapsulate
frame packets according to the format.
The message will be placed in control sub-frame
and the uplink and downlink data will be arranged
in data sub-frame.
Architecture Framework for Multicast
-Tree Construction
Definitions of node:
Source node:start point of the multicast session.
Receiver node:one of end points of the multicast session.
Tree node:located in the multicast tree but not receiver
node.
Agent node:selected for broadcasting to its one-hop
neighbor in some specific frame.
Architecture Framework for Multicast
-Tree Construction (cont.)
Architecture Framework for Multicast
-Mathematical Model and algorithm
Let T = (V, E) denote the routing tree defined in the
MSH-CSCF message,
where
The nodes V are access points.
• All nodes in V are labeled with an integer.
• the root node is labeled with 0. (Mesh BS)
The links E are bi-directional wireless links between
neighboring nodes.
• All links are also labeled with integer.
The routing tree T has L layers.
Architecture Framework for Multicast
-Mathematical Model and algorithm (cont.)
Let G= (M,P) denote the multicast routing tree,
where
M are the group member of multicast tree,
include BS, source node S, Tree node T
={ i, i V (i is a Tree Node)}, Agent Nodes A
={ i, i V (i is an Agent Node)} and Receiver
nodes R ={ i, i V (i is a Receiver Node)}.
The set P are links between neighboring multicast
tree nodes.
Architecture Framework for Multicast
-Mathematical Model and algorithm (cont.)
H(i) :hop count from node i to BS.
D(i,j) :the hop count between node i and j.
Fi={(j)| (j,i)E} :the neighboring father of node i.
C(i) :all the children of node i.
Ni= {(j)|D(i,j) =1 H(j)≥H(i)} :one-hop neighbor set of node i
with same or larger hop count than i.
Multicast Exponent MEi = | O( j ) || j ( Ni R) ( Ni A) :for
each node to select candidates for agent nodes.
Architecture Framework for Multicast
-Mathematical Model and algorithm (cont.)
Architecture Framework for Multicast
-Mathematical Model and algorithm (cont.)
Architecture Framework for Multicast
-Mathematical Model and algorithm (cont.)
Performance Evaluation
It used the network to compare proposed multicast
scheme with traditional scheme.
There are total 19 nodes in, the source node is node
14 and the nodes in blue are receiver nodes.
Performance Evaluation (cont.)
Following the algorithm of shortest path and
multicast tree construction
Layer 0
0
2
0
Layer 1
Layer 2
Layer 3
Layer 4
2
2
3
5
2
1
1
0
0
1
1
0
4
2
1
Performance Evaluation -frame-slot
Bandwidth efficiency:
9.4% v.s. 5.5%
Performance Evaluation
–the variation of latency
20-rounds’ simulation
Traffics: Poisson distribution
Proposed mechanism:
average Delay/maximal delay:
21.6 / 28 frame slots
traditional mechanism:
52%
average Delay/maximal delay:
44.5 / 92.4 frame slots
Performance Evaluation
–delay for random choice of source/receiver node
200-rounds’ simulation
Source node: randomly
Receiver node: randomly (10)
Traffics: Poisson distribution
30%
80%
Performance Evaluation
–the variation for multicast duration
Source node: randomly
Receiver node: randomly
Num of Receiver node:
4,7,10,13,16
65%
6%
Conclusion
The paper have presented a multicast mechanism in
IEEE802.16 based mesh network.
The proposed model can save more than 50% of the
frame-slots in peak value.
Simulation results can improve efficiency remarkably.