light paths
Post on 06-Apr-2018
270 Views
Preview:
TRANSCRIPT
-
8/3/2019 Light Paths
1/27
Optical NetworksBM-UC Davis 122
Part III
Wide-Area (Wavelength-Routed)
Optical Networks
1. Virtual Topology Design
2. Wavelength Conversion
3. Control and Management
-
8/3/2019 Light Paths
2/27
Optical NetworksBM-UC Davis 123
Lightpaths and Wavelength Routing
Lightpath
Virtual topology
Wavelength-continuityconstraint
Wavelength conversion
Packet routing
-
8/3/2019 Light Paths
3/27
Optical NetworksBM-UC Davis 124
Illustrative example
WA
CA1
CA2
UT
CO
TX
NEIL
MI
NY
NJPA
MD
GA
-
8/3/2019 Light Paths
4/27
Optical NetworksBM-UC Davis 125
Solution 1a: Infocom94 and ToN-Oct96
More than one laser filter pair at any node can tune to the samewavelength
-
8/3/2019 Light Paths
5/27
Optical NetworksBM-UC Davis 126
Solution 1b: Infocom94 and ToN-Oct96
All laser filter pairs at any node must be tuned to differentwavelengths
-
8/3/2019 Light Paths
6/27
Optical NetworksBM-UC Davis 127
Virtual Topology
-
8/3/2019 Light Paths
7/27
Optical NetworksBM-UC Davis 128
Wavelength Routing Switch (WRS)Details of the UT Node
-
8/3/2019 Light Paths
8/27
Optical NetworksBM-UC Davis 129
Optimization Problem Formulation
On virtual topology connection matrix Vij
iTV ij
ij jRV j
i
ij
On physical route variablespijmn
mn
ij
mn Pp
ij
ij
mn Vp
jikifppn
ijkn
m
ijmk ,
ij
n
ij
in Vp
ij
m
ij
mj Vp
On virtual topology traffic variables sdij
0sd
ij
sd
j
sd
sj
sd
j
sd
sj
dskifj
sd
kj
i
sd
ik,
CVij
ds
sd
ij
,
On coloring of lightpaths cijk
ij
k
ij
k Vc
knmcp ijk
ij
ij
mn,,1
Objective: Optimality criterion(a) Delay minimization:
(b) Maximizing offered load (equivalent to minimizing maximum flow in a link):
ij sd mn sd
sd
ij
mn
ij
mn
sd
ijC
dpMinimize
1
jisd
sd
ij ,maxmin
New optimality criterion
(c) Minimize average hop distance
ji ds
sd
ij
ds sd
Minimize, ,,
1
-
8/3/2019 Light Paths
9/27
Optical NetworksBM-UC Davis 130
Solution Approach to Virtual Topology WDM WAN Design
1. Choice of optimal virtual topology Simulated annealing; optimization based on maximizing throughput,
minimizing delay, maximizing single-hop traffic, etc.
2.Routing of lightpaths over the physical topology Alternate-path routing, multicommodity flow formulation, randomized
routing
3. Wavelength assignment: Coloring of lightpaths to avoidwavelength clashes
Graph-coloring algorithms, layered graph models
4. (Optimal) routing of packets over the virtual topology
Shortest-path routing, flow-deviation algorithm, etc.
5. Iterate
Check for convergence and go back to Step 1, if necessary.
-
8/3/2019 Light Paths
10/27
Optical NetworksBM-UC Davis 131
Details of Virtual Topology Design
Simulated Annealing
Start with random virtual topology Perform node exchange operations on two random nodes
Route packet traffic (optimally) using flow deviation
Calculate maximum trafficscaleup for current configuration
If maximum scaleup is higher then previous maximum,then accept current configuration;else accept current configuration with certain decreasing probability
Repeat until problem solution stabilizes (frozen).
Flow Deviation
Perform shortest-path routing of the traffic
Select path with large traffic congestion Route a fraction of this traffic to less-congested links
Repeat above two steps iteratively, until solution is acceptable
-
8/3/2019 Light Paths
11/27
Optical NetworksBM-UC Davis 132
NSFNET Traffic Matrix (11:45 PM to midnight, ET, Jan. 12, 1992)
-
8/3/2019 Light Paths
12/27
Optical NetworksBM-UC Davis 133
The WDM Advantage
Transceivers/node
Scaleup
4 106
5 1356 163
-
8/3/2019 Light Paths
13/27
Optical NetworksBM-UC Davis 134
Delay Components in a WDM Solution
-
8/3/2019 Light Paths
14/27
Optical NetworksBM-UC Davis 135
Scaling of Bandwidth The WDM Advantage
No WDM (Physical Topology)
Mbpsp
p
pH
CL
WDM (with P transmitters/receivers per node)
Mbpsvv
vv
HCNP
HCL
WDM Advantage
vp
v
v
p
pv
p
p
v
p
v
H
P
H
H
L
NP
H
H
L
L
IncreasingPdecreasingHv
C= link speed (Mbps)
Hp= avg. hop distance (physical)
N= number of nodes
-
8/3/2019 Light Paths
15/27
Optical NetworksBM-UC Davis 136
Problems/Limitations of Solution 1
Nonlinear objective functions.
Nonlinear constraints on wavelength continuity.
Resorted to heuristicsOptimal virtual topology design (Simulated Annealing)
Optimal packet routing on V.T. (Flow Deviation Algorithm)
No routing and wavelength assignment(Shortest-path lightpath routing; no constraints onwavelengths).
-
8/3/2019 Light Paths
16/27
Optical NetworksBM-UC Davis 137
Highlights/Contributions of Solution 2
Complete Virtual Topology Design
Linear formulationOptimal solution
Objective: Minimize average hop distance
Assume: Wavelength conversion(Sparse conversion provides almost full conversion benefits).
Resource Budgeting Tradeoffs Important/Expensive Resources: Transceivers and
wavelengths
Dont under-utilize either of them!
Hardware cost model.
Optimal Reconfiguration Algorithm
Minimize reconfiguration time.
-
8/3/2019 Light Paths
17/27
Optical NetworksBM-UC Davis 138
Optional Constraints / Simplifying Assumptions
Need scalability.
Physical topology is a subset of the virtual topology.
Bounded lightpath length
Prevent long convoluted lightpaths from occuring.
Prune the search spaceConsider Kshortest paths (bounded K).
-
8/3/2019 Light Paths
18/27
Optical NetworksBM-UC Davis 139
Two Solutions from the LP
(a) Two-wavelengthsolution
(b) Five-wavelengthsolution
-
8/3/2019 Light Paths
19/27
Optical NetworksBM-UC Davis 140
Hop Distance, Transceiver + Wavelength Utilization
-
8/3/2019 Light Paths
20/27
Optical NetworksBM-UC Davis 141
Average Hop Distance
-
8/3/2019 Light Paths
21/27
Optical NetworksBM-UC Davis 142
Transceiver Utilization
-
8/3/2019 Light Paths
22/27
Optical NetworksBM-UC Davis 143
Wavelength Utilization
-
8/3/2019 Light Paths
23/27
Optical NetworksBM-UC Davis 144
Heuristic Solutions
-
8/3/2019 Light Paths
24/27
Optical NetworksBM-UC Davis 145
WDM Network Cost Model
2/log2111
m
N
m
mx
N
j
jN
i
im
i
i
i
it WC
W
R
W
TMCRTCC
-
8/3/2019 Light Paths
25/27
Optical NetworksBM-UC Davis 146
Reconfiguration Algorithm
Generate linear formulations F(1)and F(2)corresponding to traffic
matrices sd1 and sd2.
Derive solutions and S(1)and S(2), corresponding to F(1)and F(2)
Modify F(2)to F(2)by adding the new constraint:
New objective function for F(2):
or
Although modis nonlinear, above reconfiguration formulation is linearsince the variables ps and Vs are binary.
2, ,
1
OPTji ds
sd
ij
sd sd
ij mn
ij
mn
ij
mn ppMinimize )1()2(:
ijijij VVMinimize )1()2(:
-
8/3/2019 Light Paths
26/27
Optical NetworksBM-UC Davis 147
Reconfiguration Statistics
-
8/3/2019 Light Paths
27/27
Optical NetworksBM-UC Davis148
Summary of Virtual Topology Design Principles
Use WDM to scale up an existing fiber-based WAN(Networks information carrying capacity increased
manifold)
Employ packet-switched virtual topology
imbedded on a physical topology as if we have a virtual Internet
(which is reconfigurable under user control) need optimum graph-imbedding algorithms
Reuse electronic switch of existing WAN as part of the WRS in the scaled-up WAN
top related