Advances in Optical Network Design with p-Cycles:

1
Advances in Optical Network Design with
p-Cycles Joint optimization and pre-selection of
candidate p-cycles (work in progress) Wayne D.
Grover, John Doucette grover_at_trlabs.ca,
doucette_at_trlabs.ca TRLabs and University of
Alberta Edmonton, AB, Canada Related papers
available at www.ee.ualberta.ca/grover IEEE
LEOS Summer Topicals 2002 Mont Tremblant,
Quebec, Canada July 2002
2
Outline

• What are p- Cycles ?
• Why do we say they offer mesh-efficiency with
  ring-speed ?
• Optimal design with p-Cycles
• non-joint or spare capacity only design
• jointly optimized design
• What makes a good p-Cycle ?
• The idea of Preselection
• Preselection by Topological Score, by A Priori
  Efficiency (AE)
• Application of Preselection to Joint and
  non-joint p_cycle Design Problems

3
Important Features of p-Cycles

• Working paths go via shortest routes over the
  graph
• p-Cycles are formed only in the spare capacity
• Can be either OXC-based or on ADM-like nodal
  devices
• a unit-capacity p-cycle protects
• one unit of working capacity for on cycle
  failures
• two units of working capacity for straddling
  span failures
• Straddling spans
• there may be up to N(N-1)/2 -N straddling span
  relationships
• straddling spans each bear two working channels
  and zero spare
• Only two nodes do any real-time switching for
  restoration
• protection capacity is fully preconnected
• switching actions are known prior to failure

4
The Unique Position p-Cycles Occupy

• p -cycles
• BLSR speed
• mesh efficiency

Path rest, SBPP
Speed
Span (link)rest.
200 ms
BLSR
50 ms
100
50
200
Redundancy
5
Backgrounder p-Cycles
Ring network
p-Cycle
Spare Capacity
x2 protection coverage on eachstraddling span
Protection Coverage
Able to restore 9 working wavelength channels
Able to restore 29 working wavelength
channels (on 19 spans)
6
Motivation for Joint Optimization

• In joint optimization the working route
  assignments are chosen in conjunction with
  survivability considerations
• example of the effect this can have

2 working channel-hops 12 spares in total TOTAL
Capacity 14
2e working 6 spares in total TOTAL Capacity 8
e
7
Approaches to p-Cycle Network Design
(non-joint)
(joint)
Route all lightpath requirementsvia
shortest-paths
enumerateeligible working routes
enumerategraph cycles
enumerategraph cycles
I.L.P. solution forp-cycle formation
Heuristic algorithm(s) forp-cycle formation
all in one I.L.P. solution
working routes working capacity
working routes working capacity
p-cycles spare capacity
p-cycles spare capacity
8
Integer Linear Programming (I.L.P) Formulation
(for the joint problem)

• Objective Function
• Minimize total cost of working and spare
  capacity
• Subject To
• A. All lightpath requirements are routed.
• B. Enough WDM channels are provisioned to
  accommodate the routing of lighpaths in A.
• C. The selected set of p-cycles give 100 span
  protection.
• D. Enough spare channels are provisioned to
  create the p-cycles needed in C.
• E. Integer p-cycles decision variables, integer
  capacity

9
Comments Approaches to p-Cycle Network Design

• Non-joint problem
• several heuristic algorithms under development
• however, optimal solution is quite fast too
• no real difficulties here
• Joint design problem
• I.L.P more complex to solve (coupled integer
  decision variables and constraint systems)
• Idea use I.L.P. but with reduced number of
  preselected candidate cycles
• need some a priori view as to what makes a
  candidate cycle a promising as p-cycle

10
Preselection Criteria (1) Topological Score (TS)
TS
Credit rules 1 for an on-cycleprotection
relationship 2 for a straddling
spanprotection relationship
Examples
6 spans, all on-cycle(equiv. To a ring) TS 6
7 spans on-cycle 2 straddlers TS 7 22 11
on-cycle
straddlers
By itself TS tends to like large cycles
(Hamiltonian maximizes TS)no regard to
corresponding cost of the cycle
11
Preselection Criteria (2) a Priori Efficiency
(AE)
Examples
AE
AE is defined as TS j-------------- Cost of
cycle j
TS 6 Cost 6 hops --gt AE 1
Note all rings have AE 1
TS 11 Cost 7 hops --gt AE 1.57

• Preselection hypothesis
• choose a small number of elite cycle
  candidates based on AE
• Let I.L.P. formulation assemble final design

12
COST239 European Study Network
Copenhagen

• Pan European optical core network
• planning model defined by COST 239 study group
  for optical networks
• 11 nodes, 26 spans
• Average nodal degree 4.7
• Demand matrix
• Distributed pattern
• 1 to 11 lightpaths per node (average 3.2)

London
Berlin
Amsterdam
Brussels
Luxembourg
Prague
Zurich
Paris
Vienna
Milan
13
Results(1) Benefits of Preselection by AE Metric
(non-joint design)
COST239 non-joint designs Solution quality vs.
No. candidate p-cycles in designc
500 cycles
2000 cycles
14
Results(2) Benefits of AE Metric Pre-Selection
(Joint Design)
COST239 joint designs Solution quality vs. No.
candidate p-cycles in designc
200 cycles
2000 cycles
15
Benefits of AE Metric Pre-Selection (Joint Design)
Additional Test Network 20 nodes, 40 spans, 190
demand pairs
2000 cycles
18,000 cycles
mipgap
16
Where the Preselection Heuristic Can Really
Help... Exponential Nature of Cycle Enumeration
Illustrated for 40 nodes and a varying of spans
(hence connectivity)
The preselection strategy will help us keep the
problem sizes manageable, i.e., in this range,
avoiding the combinatorial explosion that
happens over here.
17
How Much Does Joint Design Improve Efficiency?
COST-239 Joint design uses 5 more working
capacity, and 43 less spare capacity for total
network capacity reduction of 13. Network
redundancy 39
working
spare
(4 p-cycles)
(7 p-cycles)
joint
non-joint
18
Summary Main Findings

• Jointly optimized p-cycle protected OTNs can be
  extremely efficient
• as little as 39 redundancy observed (for 100
  span protection)
• Joint design is a more complex problem, however
• Solution time reduced by preselection of a small
  number of elite cycle candidates based on AE
• Further Work and Applications
• Other test networks
• Incremental application to dynamic demands
• Strategies for wavelength conversion
• Design heuristics