Architectures and Architectures and Algorithms for Resource Algorithms for Resource Allocation Allocation Mounire El Houmaidi Mounire El Houmaidi * , Mostafa A. Bassiouni , Mostafa A. Bassiouni * , and Guifang Li , and Guifang Li # * School of Electrical Engineering and Computer Science School of Electrical Engineering and Computer Science # School of Optics/CREOL School of Optics/CREOL University of Central Florida University of Central Florida
20
Embed
Architectures and Algorithms for Resource Allocation
Architectures and Algorithms for Resource Allocation. Mounire El Houmaidi * , Mostafa A. Bassiouni * , and Guifang Li # * School of Electrical Engineering and Computer Science # School of Optics/CREOL University of Central Florida. Outline. Motivation What is a Minimum Dominating Set (MDS) - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Architectures and Algorithms for Architectures and Algorithms for Resource AllocationResource Allocation
Mounire El HoumaidiMounire El Houmaidi**, Mostafa A. Bassiouni, Mostafa A. Bassiouni**, and Guifang Li, and Guifang Li##
**School of Electrical Engineering and Computer ScienceSchool of Electrical Engineering and Computer Science##School of Optics/CREOLSchool of Optics/CREOL
University of Central FloridaUniversity of Central Florida
OutlineOutline Motivation What is a Minimum Dominating Set (MDS) How to find k-MDS
– Algorithm– Example– What is Weighted MDS
Applications of k-MDS– Sparse placement of wavelength conversion
• k-LOSS(k-BLK) and F-SEARCH• Weighted k-MDS for non-uniform traffic• Limited wavelength conversion
– Placement of G-nodes for traffic grooming– Placement of FDLs
Optimize overall network performance by using dominating nodes [1-4]
1. M. El Houmaidi et. al., J. Opt. Net., 2:6, (OSA, 2003)2. M. El Houmaidi et. al., Proc. MASCOTS, (IEEE/ACM, 2003)3. M. El Houmaidi et. al., J. Opt. Eng., 43:1, (SPIE, 2004) 4. M. El Houmaidi et. al., Proc. OFC, (IEEE, 2004)
1
0
2
3
4
5
6
7
8
10
9
11
13
12
14
15
16
17
18
19
20
21
22
23
24
25
26
27
(U.S Long Haul Net.)
What is What is MDSMDS Given a graph G(V,E), determine a set with minimum
number of vertices D V such that every vertex in the graph is either in D or is at distance k or less from at least one member in D.
NP-Complete problem [1,2] .
Heuristic algorithms for sub-optimal solution.
Highly connected nodes dominate the entire topology.
1. Karp, Pl. Press, 19722. Lund, et. al., J. ACM, 1994
DefinitionsDefinitions
Neighbor (v): is the set of nodes sharing a link with v.
k-Neighbor (v): is the set of nodes that are at most within k hops away from a node v.
For k equals 0, 0-Neighbor(v) contains the node v only.
Definitions (Cont.)Definitions (Cont.) k-Connect(v): the connectivity index based on nodes within k hops of v is :
k-Master (v): represents the node p, member of k-Neighbor(v), with the highest k-Connect value over all nodes m that are at most k hops away from node v (i.e., all nodes mk-Neighbor(v))
We can achieve with 2-WMDS members as G-nodes (r=16) the same throughput as if all nodes in the network had the grooming capability (r is the grooming ratio)
OBS switch design with FDLs/OWCsOBS switch design with FDLs/OWCs
Input Link 2
DMX
MUX
MAIN CONTROL
DMX: De-multiplexor
MUX: Multiplexor
OWC: any-to- Converter
FDL: Fiber Delay Line
DMX
MUX
1
W
F.W
1
W
i O X C
Output Link 1
Output Link 2
Input Link 1
OWC
OWCW
.
.
.
Converter Bank
1
FDL
F.W + 2 F.W + 2
2
FDL1
F.W + 1F.W + 1
FDL Bank
OWC
OWCW
.
.
.
Converter Bank
1A 1
B 1
C 1 C 1
A 1
B 1
A 2
A 2
B 2
C 2
C 2
B 2
Fiber Delay Line designFiber Delay Line design
Variable delay: [0…Variable delay: [0…MAXDMAXD], where ], where MAXDMAXD = (2 = (200 + 2 + 211 +… +2 +… +2(max_d)(max_d)) x b) x b
λ1
. . .λW
λ1
. . .λW
…
20 21 22
…
2(max_d)OWC
OWC OWC
OWC
Benefits of FDLs and OWCsBenefits of FDLs and OWCsSwitch design benefits with NSFNET Topology
FDLs vs. OWCs in 2-WM DS nodes (W=16)
0
0.05
0.1
0.15
0.2
0.25
0.3
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Load intensity
Bu
rst
los
s p
rob
ab
ilit
y
No FDLs-No OWCs
OWCs only
FDLS only
FDLs and OWCs
FDLs vs. OWCs with JET signaling and W=16
In a fully connected network (all nodes are connected), OWC has no effect on the blocking performance but FDLs do.
FDLs and OWCs capabilities must be used judiciously and placed in nodes that maximize the performance.
k-LOSS heuristic [JIM99, MSS02]: Via simulation, Place OWC in nodes experiencing the highest blocking rates.