Upload
morty
View
33
Download
0
Tags:
Embed Size (px)
DESCRIPTION
Mathematical Challenges in Telecommunicati on a Tutorial. Martin Grötschel IMA workshop on Network Management and Design Minneapolis, MN, April 6, 2003. Comment. This is an abbreviated version of my presentation. The three films I showed are not included. - PowerPoint PPT Presentation
Citation preview
http://www.zib.de/[email protected]
Institute of Mathematics, Technische Universität Berlin (TUB) DFG-Research Center “Mathematics for key technologies” (FZT 86) Konrad-Zuse-Zentrum für Informationstechnik Berlin (ZIB)
Martin Grötschel
Mathematical Challenges in Telecommunication
a Tutorial
Martin Grötschel
IMA workshop on Network Management and Design
Minneapolis, MN, April 6, 2003
MartinGrötsche
l
2
Comment
This is an abbreviated version of my presentation.
The three films I showed are not included.
I also deleted many of the pictures and some of the tables: in some cases I do not have permission to publish
them since they contain some confidential material or information not for public use,
in other cases I am not sure of the copy right status (and do not have the time and energy to check it).
MartinGrötsche
l
3
The ZIB Telecom Group
Andreas Bley
Andreas Eisenblätter
Martin Grötschel
Thorsten Koch
Arie Koster
Roland Wessäly
Adrian Zymolka
ZIB Associates
Manfred Brandt
Sven Krumke
Frank Lutz
Diana Poensgen
Jörg Rambau
and more:
Clyde Monma (BellCore, ...)
Mechthild Opperud (ZIB, Telenor)
Dimitris Alevras (ZIB, IBM)
Christoph Helmberg (Chemnitz)
ZIB Telecom Team
MartinGrötsche
l
4
ZIB Partners from Industry
Bell Communications Research
Telenor (Norwegian Telecom)
E-Plus (acquired by KPN in 01/2002)
DFN-Verein
Bosch Telekom (bought by Marconi)
Siemens
Austria Telekom (controlled by Italia Telecom)
T-Systems Nova (T-Systems, Deutsche Telekom)
KPN
Telecel-Vodafone
Atesio (ZIB spin-off company)
MartinGrötsche
l
5
Contents
1. Telecommunication: The General Problem2. The Problem Hierarchy: Cell Phones and
Mathematics3. The Problem Hierarchy: Network Components
and Math4. Network Design: Tasks to be solved
Addressing Special Issues:5. Frequency Assignment6. Locating the Nodes of a Network 7. Balancing the Load of Signaling Transfer Points8. Integrated Topology, Capacity, and Routing
Optimization as well as Survivability Planning9. Planning IP Networks10.Optical Networks11.Summary and Future
MartinGrötsche
l
6
Contents
1. Telecommunication: The General Problem2. The Problem Hierarchy: Cell Phones and
Mathematics3. The Problem Hierarchy: Network Components
and Math4. Network Design: Tasks to be solved
Addressing Special Issues:5. Frequency Assignment6. Locating the Nodes of a Network 7. Balancing the Load of Signaling Transfer Points8. Integrated Topology, Capacity, and Routing
Optimization as well as Survivability Planning9. Planning IP Networks10.Optical Networks11.Summary and Future
MartinGrötsche
l
7
What is the Telecom Problem?
Design excellent technical devices and a robust network that survivesall kinds of failures and organize the traffic such that high quality telecommunication betweenvery many individual units at many locations is feasibleat low cost!
Speech DataVideo Etc.
MartinGrötsche
l
8
What is the Telecom Problem?
Design excellent technical devices and a robust network that survivesall kinds of failures and organize the traffic such that high quality telecommunication betweenvery many individual units at many locations is feasibleat low cost!
This problem This problem is is
too general too general to be solved to be solved
in in one step.one step.
Approach in Practice:• Decompose whenever possible• Look at a hierarchy of problems• Address the individual problems one by one• Recompose to find a good global solution
MartinGrötsche
l
9
Contents
1. Telecommunication: The General Problem2. The Problem Hierarchy: Cell Phones and
Mathematics3. The Problem Hierarchy: Network Components and
Math4. Network Design: Tasks to be solved
Addressing Special Issues:5. Frequency Assignment6. Locating the Nodes of a Network 7. Balancing the Load of Signaling Transfer Points8. Integrated Topology, Capacity, and Routing
Optimization as well as Survivability Planning
9. Planning IP Networks10.Optical Networks11.Summary and Future
MartinGrötsche
l
10
Cell Phones and Mathematics
Designing mobile phones•Task partitioning•Chip design (VLSI)•Component design
•Computational logic•Combinatorial optimization•Differential algebraic equations
Producing Mobile Phones•Production facility layout•Control of CNC machines•Control of robots •Lot sizing•Scheduling•Logistics
•Operations research•Linear and integer programming•Combinatorial optimization•Ordinary differential equations
Marketing and Distributing Mobiles•Financial mathematics •Transportation optimization
cellphonepicture
Chip Design
CMOS layout for four-transistor static-memory cell
CMOS layout for two four-transistor static-memory cells.
Compacted CMOS layout for two four-transistor static-memory cells.
PlacementRouting
Compactification
Schematic for four-transistor static-memory cell
ZIB-Film 9:01 – 9:41
MartinGrötsche
l
12
Design and Production of ICs and PCBs
Integrated Circuit (IC)Printed Circuit Board (PCB)
Problems: Logic Design, Physical DesignCorrectness, Simulation, Placement of
Components, Routing, Drilling,...
Production and Mathematics: Examples
CNC Machine for 2D and 3D cutting and welding
(IXION ULM 804)Sequencing of Tasks
and Optimization of Moves
Mounting DevicesMinimizing Production Time
via TSP or IP
Printed Circuit Boards
Optimization of Manufacturing
SMD
MartinGrötsche
l
14
Drilling 2103 holes into a PCB
Significant Improvementsvia TSP
(Padberg & Rinaldi)
MartinGrötsche
l
15
Siemens Problem
before after
printed circuit board da1
MartinGrötsche
l
16
Siemens Problem
before after
printed circuit board da4
MartinGrötsche
l
17
Mobile Phone Production Line
Fujitsu Nasu plant
MartinGrötsche
l
18
Contents
1. Telecommunication: The General Problem2. The Problem Hierarchy: Cell Phones and
Mathematics3. The Problem Hierarchy: Network Components
and Math4. Network Design: Tasks to be solved
Addressing Special Issues:5. Frequency Assignment6. Locating the Nodes of a Network 7. Balancing the Load of Signaling Transfer Points8. Integrated Topology, Capacity, and Routing
Optimization as well as Survivability Planning9. Planning IP Networks10.Optical Networks11.Summary and Future
MartinGrötsche
l
19
Network Components
Design, Production, Marketing, Distribution:Similar math problems as for mobile phones
•Fiber (and other) cables•Antennas and Transceivers •Base stations (BTSs)•Base Station Controllers (BSCs)•Mobile Switching Centers (MSCs)•and more...
MartinGrötsche
l
20
Component „Cables“
MartinGrötsche
l
21
Component „Antennas“
MartinGrötsche
l
22
Component „Base Station“
Nokia MetroSite
Nokia UltraSite
Component„Mobile Switching Center“:
Example of an MSC Plan
MartinGrötsche
l
24
Contents
1. Telecommunication: The General Problem2. The Problem Hierarchy: Cell Phones and
Mathematics3. The Problem Hierarchy: Network Components
and Math4. Network Design: Tasks to be solved
Addressing Special Issues:5. Frequency Assignment6. Locating the Nodes of a Network 7. Balancing the Load of Signaling Transfer Points8. Integrated Topology, Capacity, and Routing
Optimization as well as Survivability Planning9. Planning IP Networks10.Optical Networks11.Summary and Future
MartinGrötsche
l
25
Network Design: Tasks to be solved Some Examples
Locating the sites for antennas (TRXs) and
base transceiver stations (BTSs)
Assignment of frequencies to antennas
Cryptography and error correcting encoding for wireless communication
Clustering BTSs
Locating base station controllers (BSCs)
Connecting BTSs to BSCs
MartinGrötsche
l
26
Network Design: Tasks to be solved Some Examples (continued)
Locating Mobile Switching Centers (MSCs)
Clustering BSCs and Connecting BSCs to MSCs
Designing the BSC network (BSS) and the MSC network (NSS or core network) Topology of the network Capacity of the links and components Routing of the demand Survivability in failure situations
Most of these problems turn out to be Combinatorial Optimization or
Mixed Integer Programming Problems
MartinGrötsche
l
27
Connecting Mobiles: What´s up?
BSC
MSC
BSC
BSC
BSC
BSC
BSC
BSC
MSC
MSCMSC
MSC
BTS
MartinGrötsche
l
28
Connecting Computers or other Devices
The mathematical problems are similar
but not identical
MartinGrötsche
l
29
Contents
1. Telecommunication: The General Problem2. The Problem Hierarchy: Cell Phones and
Mathematics3. The Problem Hierarchy: Network Components and
Math4. Network Design: Tasks to be solved
Addressing Special Issues:5. Frequency Assignment6. Locating the Nodes of a Network 7. Balancing the Load of Signaling Transfer Points8. Integrated Topology, Capacity, and Routing
Optimization as well as Survivability Planning
9. Planning IP Networks10.Optical Networks11.Summary and Future
MartinGrötsche
l
30
FAP
Film
MartinGrötsche
l
31
Antennas & Interference
xx
antenna
backbone network
xx
xxsite
xx
cell
co- & adjacentchannel
interferencecell
MartinGrötsche
l
32
Interference
Level of interference depends on distance between transmitters geographical position power of the signals direction in which signals are transmitted weather conditions assigned frequencies
co-channel interference adjacent-channel interference
MartinGrötsche
l
33
Separation
Parts of the spectrum forbidden
at some locations: government regulations, agreements with operators in
neighboring regions, requirements military forces, etc.
Site
Blocked channels
Frequencies assigned to the same
location (site) have to be separated
MartinGrötsche
l
34
Frequency Planning Problem
Find an assignment of frequencies to transmitters that satisfies all separation constraints all blocked channels requirements
and either avoids interference at all
or minimizes the (total/maximum) interference
level
MartinGrötsche
l
35
Minimum InterferenceFrequency Assignment Problem
Integer Linear Program:
min
. . 1
1 , ( )
1 ,
1 , 1
, , 0,1
co ad
v
co co ad advw vw vw vw
vw E vw E
vff F
dvf wg
co covf wf vw v w
ad advf wg vw
co advf vw vw
c z c z
s t x v V
x x vw E f g d vw
x x z vw E f F F
x x z vw E f g
x z z
MartinGrötsche
l
36
A Glance at some Instances
k 267 56,8 2 151,0 238 3 69B-0-E-20 1876 13,7 40 257,7 779 5 81
f 2786 4,5 3 135,0 453 12 69h 4240 5,9 11 249,0 561 10 130
E-Plus Project
MartinGrötsche
l
37
Region Berlin - Dresden
2877 carriers
50 channels
Interference
reduction: 83.6%
MartinGrötsche
l
38
Region Karlsruhe
2877 Carriers
75 channels
Interference
Reduction:83.9 %
MartinGrötsche
l
39
The UMTS Radio Interface
Completely new story
see talk by Andreas Eisenblätter on Monday
MartinGrötsche
l
40
Contents
1. Telecommunication: The General Problem2. The Problem Hierarchy: Cell Phones and
Mathematics3. The Problem Hierarchy: Network Components
and Math4. Network Design: Tasks to be solved
Addressing Special Issues:5. Frequency Assignment6. Locating the Nodes of a Network 7. Balancing the Load of Signaling Transfer Points8. Integrated Topology, Capacity, and Routing
Optimization as well as Survivability Planning9. Planning IP Networks10.Optical Networks11.Summary and Future
MartinGrötsche
l
41
G-WiN Data
G-WiN = Gigabit-Wissenschafts-Netz of the DFN-Verein
Internet access of all German universities and research institutions
• Locations to be connected: 750• Data volume in summer 2000: 220 Terabytes/month • Expected data volume in 2004: 10.000 Terabytes/month
Clustering (to design a hierarchical network):• 10 nodes in Level 1a 261 nodes eligible for• 20 nodes in Level 1b Level 1• All other nodes in Level 2
MartinGrötsche
l
42
G-WiN Problem
• Select the 10 nodes of Level 1a.• Select the 20 nodes of Level 1b.
• Each Level 1a node has to be linked to two Level 1b nodes.• Link every Level 2 node to one Level 1 node.
• Design a Level 1a Network such that•Topology is survivable (2-node connected)•Edge capacities are sufficient (also in failure situations)•Shortest path routing (OSPF) leads to balanced capacity use (objective in network update)
• The whole network should be „stable for the future“.• The overall cost should be as low as possible.
MartinGrötsche
l
43
Potential node locations for the 3-Level Network of the G-WIN
Red nodes are potentiallevel 1 nodes
Cost: Connection between nodes Capacity of the nodes
Blue nodes are all remaining nodes
MartinGrötsche
l
44
Demand distribution
The demand scales with the height of each red line
AimSelect backbone nodes and
connect all non-backbone nodes to a backbone node
such that theoverall network cost is minimal
(access+backbone cost)
MartinGrötsche
l
45
G-WiN Location Problem: Data
set of locations
set of potential Level 1a locations (subset of )
set of possible configurations at
location in Level 1a
For , and :
connection costs from to
tr
p
p
ip
i
V
Z V
K
p
i V p Z k K
w i p
d
affic demand at location
capacity of location in configuration
costs at location in configuration
1 if location is connected to (else 0)
1 if configuration is used at loc
kp
kp
ip
kp
i
c p k
w p k
x i p
z k
ation (else 0)p
MartinGrötsche
l
46
G-WiN Location/Clustering Problem
min
1
1
p
k kip ip p p
p Z i V p Z k K
ipp
k ki ip p p
i k
kp
k
kp
p
w x w z
x
d x c z
z
z const
Each location i must be connected to a Level 1 node
Capacity at p must be large enough
Only one configuration at each Location 1 node
All variables are 0/1.
# of Level 1a nodes
MartinGrötsche
l
47 Solution: Hierarchy & Backbone
MartinGrötsche
l
48
G-WiN Location Problem: Solution Statistics
The DFN problem leads to ~100.000 0/1-variables.Typical computational experience:
Optimal solution via CPLEX in a few seconds!
A very related problem at Telekom Austria has~300.000 0/1-variables plus some continuous variables
and capacity constraints.Computational experience (before problem specific fine tuning):
10% gap after 6 h of CPLEX computation,60% gap after „simplification“ (dropping certain capacities).
MartinGrötsche
l
49
Contents
1. Telecommunication: The General Problem2. The Problem Hierarchy: Cell Phones and
Mathematics3. The Problem Hierarchy: Network Components
and Math4. Network Design: Tasks to be solved
Addressing Special Issues:5. Frequency Assignment6. Locating the Nodes of a Network 7. Balancing the Load of Signaling Transfer Points8. Integrated Topology, Capacity, and Routing
Optimization as well as Survivability Planning9. Planning IP Networks10.Optical Networks11.Summary and Future
MartinGrötsche
l
50
Re-Optimization of Signaling Transfer Points
Telecommunication companies maintain a signaling network (in adition to their communication transport network). This is used for management tasks such as:
Basic call setup or tear down
Wireless roaming
Mobile subscriber authentication
Call forwarding
Number display
SMS messages
Etc.
MartinGrötsche
l
51
Signaling Transfer Point (STP)
CCD
CCD
CCD
CCD CCD
CCD
CCD
CCD
CCD
Link-Sets STP
Cluster
Cluster
Cluster
CCLK
CCD=Common Channel Distributors, CCLK=Common Channel Link Controllers
CCD=routing unit, CCLK=interface card
MartinGrötsche
l
52
STP – Problem description
Target
Assign each link to a CCD/CCLK
Constraints
At most 50% of the links in a linkset can be assigned to a single cluster
Number of CCLKs in a cluster is restricted
Objective
Balance load of CCDs
MartinGrötsche
l
53
STP – Mathematical model
DataC set of CCDs j
L set of links i
Di demand of link i
P set of link-sets
Q set of clusters
Lp subset of links in link-set p
Cq subset of CCDs in cluster q
cq #CCLKs in cluster q
Variables
0,1 , , ijx i L j C
1ijx if and only iflink i is assigned to CCD j
MartinGrötsche
l
54
STP – Mathematical model
2
min
1
,
0,1
p
p q
q
ijj C
i iji L
i iji L
iji L j C
ij qi L j C
ij
L
y z
x i L
D x y j C
D x z j C
x p P q Q
x c q Q
x
Assign each link
Upper bound of CCD-load
Diversification
Lower bound of CCD-load
CCLK-bound
Integrality
Min load difference
MartinGrötsche
l
55
STP – current (former) solution
Minimum: 186 Maximum: 404 Load difference: 218
0
50
100
150
200
250
300
350
400
450
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18CCD
Tra
ffic
Lo
ad
MartinGrötsche
l
56
STP – „Optimal solution“
Minimum: 280 Maximum: 283 Load difference: 3
0
50
100
150
200
250
300
350
400
450
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18CCD
Tra
ffic
Loa
d
MartinGrötsche
l
57
STP – Practical difficulty
Problem: 311 rearrangements are necessary to migrate to the optimal
solution
Reformulation with new objective
Find a best solution with a restricted number of
changes
MartinGrötsche
l
58
STP – Reformulated Model
*, ( )
2
min
1
,
0,1
p
p q
q
ijj C
i iji L
i iji L
iji L j C
ij qi L j C
ij
iji L j C j j i
L
y z
x i L
D x y j C
D x z j C
x p P q Q
x c q Q
x
x B
Assign each link
Upper bound of CCD-load
Diversification
Lower bound of CCD-load
CCLK-bound
Integrality
Restricted number of changes!
Min load difference
MartinGrötsche
l
59
STP – Alternative Model
*, ( )
2
min
1
,
0,1
p
p q
q
iji L j C j j i
ijj C
i iji L
i iji L
iji L j C
ij qi L j C
ij
L
x
x i L
D x y j C
D x z j C
x p P q Q
x c q Q
x
y z D
Assign each link
Upper bound of CCD-load
Diversification
Lower bound of CCD-load
CCLK-bound
Integrality
Restricted load difference
Min # changes
MartinGrötsche
l
60
STP – New Solutions
0
50
100
150
200
250
300
350
400
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
CCD
Tra
ffic
Lo
ad White: D=50, (alternative) Minimum: 257 Maximum: 307 D=Load difference: 50 Number of changes: 12
Orange: B=8, (reformulated)
Minimum: 249 Maximum: 339 Load difference: 90 Number of changes=B: 8
MartinGrötsche
l
61
STP – Experimental results
Max changes 0 5 10 15 20
Load differences
218 129 71 33 14
1 hour application of CPLEX MIP-Solver for each case
MartinGrötsche
l
62
STP - Conclusions
It is possible to achieve 85%
of the optimal improvement with less than
5% of the changes necessary to obtain a load balance optimal solution !
MartinGrötsche
l
63
Contents
1. Telecommunication: The General Problem2. The Problem Hierarchy: Cell Phones and
Mathematics3. The Problem Hierarchy: Network Components
and Math4. Network Design: Tasks to be solved
Addressing Special Issues:5. Frequency Assignment6. Locating the Nodes of a Network 7. Balancing the Load of Signaling Transfer Points8. Integrated Topology, Capacity, and Routing
Optimization as well as Survivability Planning9. Planning IP Networks10.Optical Networks11.Summary and Future
MartinGrötsche
l
64
Network Optimization Film
ZIB Film 1994, 9:40
MartinGrötsche
l
65
Network Optimization
Networks
Capacities Requirements
Cost
MartinGrötsche
l
66
What needs to be planned? Topology Capacities Routing Failure Handling (Survivability)
IP Routing Node Equipment Planning Optimizing Optical Links and Switches
DISCNET: A Network Planning Tool(Dimensioning Survivable Capacitated
NETworks)
Atesio ZIB Spin Off
MartinGrötsche
l
67
The Network Design Problem
200
65
258
134
30
42Düsseldorf
Frankfurt
Berlin
Hamburg
München
Communication Demands
Düsseldorf
Frankfurt
Berlin
Hamburg
München
Potential topology &
Capacities
MartinGrötsche
l
68
Capacities
PDH
2 Mbit/s
34 Mbit/s
140 Mbit/s
SDH
155 Mbit/s
622 Mbit/s
2,4 Gbit/s
... WDM (n x STM-N)
Two capacity models : Discrete Finite Capacities Divisible Capacities
(P)SDH=(poly)synchronous digital hierarchy
WDM=Wavelength Division MultiplexerSTM-N=Synchronous Transport Modul with N STM-1 Frames
MartinGrötsche
l
69
Survivability
H B
D
F
M
120
H B
D
F
M
3030
60Diversification„route node-disjoint“
Path restoration„reroute affected demands“(or p% of all affected demands)
6060
H B
D
F
M
H B
D
F
M
H B
D
F
M
60
60
Reservation„reroute all demands“(or p% of all demands)
120
H B
D
F
M
H B
D
F
M
H B
D
F
M
60
60
MartinGrötsche
l
70
Model: Data & Variables
SsOperating
states:
suvs PDuvSs ,,
0)( Pf s
uvPath variables:
Capacity variables: Ee}1,0{),( tex
Supply Graph: G=(V,E)
Z0
eCeT
ee CC 1
s
uvP PValid Paths:
Demand Graph: H=(V,D) Duvduv ,
Duvuv ,Duvluv ,
Duvuv ,
MartinGrötsche
l
71
Model: Capacities
Capacity variables : e E, t = 1, ..., Te {0,1}t
ex
Cost function :
1
mineT
t te e
e E t
k x
Capacity constraints : e E 0 11 0eTe e ex x x
0
eTt t
e e et
y c x
MartinGrötsche
l
72
Model: Routings
Path variables :
Capacity constraints :
Demand constraints :
, , ss uvs S uv D P
( ) 0suvf P
0
0
:
( )uv
e uvuv D P e P
y f P
e E
0
0 ( )uv
uv uvP
d f P
uv D
Path length restriction
MartinGrötsche
l
73
Model: Survivability (one example)
Path restoration „reroute affected demands“
H B
D
F
M
120
60
60
H B
D
F
M
for all sS, uvDs
for all sS, eEs
60
60
H B
D
F
M
MartinGrötsche
l
74
Mathematical Model
topology decisison
capacity decisions
normal operation routing
component failure routing
suvs PDuvSs ,,0)( Pf s
uv
)(:
0
0
PfyDuv PeP
uve
uv
Ee
0
)(0
uvPuvuv Pfd Duv
1
mineT
t te e
e E t
k x
eTtEe ,,1,
, 1, , ee E t T {0,1}tex
te
te xx 1
eT
t
te
tee xcy
0Ee
LP-based Methods
Feasibleintegersolutions
LP-based relaxation
Convex hull
Objectivefunction
Cuttingplanes
Feasibleintegersolutions
LP-based relaxation
Convex Convex hullhull
Objectivefunction
Cuttingplanes
Flow chart
Optimalsolution
InitializeLP-relaxation
Runheuristics
SolveLP-relaxation
Separation algorithms
AugmentLP-relaxation
Solve feasibilityproblem
Separation algorithms
Inequalities?
No
Yes
x variablesinteger?
Yes
No
Feasibleroutings?
Yes
No
Polyhedral combinatorics
Valid inequalities (facets)
Separation algorithms
Heuristics
Feasibility of a capacity vector
LP-based approach:
MartinGrötsche
l
77
Finding a Feasible Solution?
Heuristics Local search Simulated Annealing Genetic algorithms ...
Nodes Edges Demands Routing-Paths
15 46 78 > 150 x 10e6
36 107 79 > 500 x 10e9
36 123 123 > 2 x 10e12
Manipulation of– Routings– Topology– Capacities
Problem Sizes
78
MartinGrötsche
l
How much to save?
Real scenario• 163 nodes• 227 edges• 561 demands
34% potential savings!==
> hundred million dollars
PhD Thesis: http://www.zib.de/wessaely
MartinGrötsche
l
79
Contents
1. Telecommunication: The General Problem2. The Problem Hierarchy: Cell Phones and
Mathematics3. The Problem Hierarchy: Network Components
and Math4. Network Design: Tasks to be solved
Addressing Special Issues:5. Frequency Assignment6. Locating the Nodes of a Network 7. Balancing the Load of Signaling Transfer Points8. Integrated Topology, Capacity, and Routing
Optimization as well as Survivability Planning9. Planning IP Networks10.Optical Networks11.Summary and Future
MartinGrötsche
l
80
Comment
9. Planning IP Networks10.Optical Networks11.Summary and Future
The lecture ended after about 100 minutes. The last three topics above were not covered.
MartinGrötsche
l
81
Summary
Telecommunication Problems such as
• Frequency Assignment• Locating the Nodes of a Network Optimally• Balancing the Load of Signaling Transfer Points• Integrated Topology, Capacity, and Routing
Optimization as well as Survivability Planning
• Planning IP Networks• Optical Network Design• and many others
can be succesfully attacked with optimization techniques.
MartinGrötsche
l
82
SummaryThe mathematical programming approach
• Helps understanding the problems arising• Makes much faster and more reliable planning
possible• Allows considering variations and scenario
analysis• Allows the comparision of different technologies• Yields feasible solutions• Produces much cheaper solutions than traditional
planning techniques• Helps evaluating the quality of a network.
There is still a lot to be done, e.g., for the really important problems, optimal solutions are way out of reach!
MartinGrötsche
l
83
The Mathematical Challenges
Finding the right ballance between flexibility and controlability of future networks
Controlling such a flexible network
Handling the huge complexity
Integrating new services easily
Guaranteeing quality
Finding appropriate Mathematical Models
Finding appropriate solution techniques (exact, approximate , interactive, quality guaranteed)
http://www.zib.de/[email protected]
Institute of Mathematics, Technische Universität Berlin (TUB) DFG-Research Center “Mathematics for key technologies” (FZT 86) Konrad-Zuse-Zentrum für Informationstechnik Berlin (ZIB)
Martin Grötschel
Mathematical Challenges in Telecommunication
The End
Martin GrötschelIMA workshop on
Network Management and Design
Minneapolis, MN, April 6, 2003