Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Photonics in Switching
Lena Wosinska, Assoc. Professor Next Generation Optical NETworks NEGONET Royal Institute of Technology KTH Electrum 229, 164 40 Kista, Sweden
wosinska@kth.se
9/24/2009
Outline
Introduction Photonic Circuit Switching
WDM network design WDM network elements
Summary
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school 3
9/24/2009
Switched networks
1 Switching nodes
not concerned with contents of data purpose: provide switching facility in general not fully connected provides data to transfer connected via switching nodes physical connections between nodes
A End nodes
Links
Switching
Switching
Circuit Switching
Packet Switching
Datagram
9/24/2009
SDH/SONET Networking
9/24/2009
Photonics in switching
Optical circuit switching (OCS) p g( )
Wavelength-routed networks Relatively mature technology today Providing lightpaths WDM network elements: OLT, OADM, OXC
9/24/2009
Wavelength-Routed Networks
W-R switches Provide lightpaths Problems: Low bandwidth efficiency Large granularity
1 A
Wavelength-routing switch Access (client) node (e.g., IP router): contains (tunable) transmitters and receivers
L. Wosinska Tutorial at BONE Summer school 11
Network
Network elements
Fiber Optical line amplifier (OLA) Optical line terminal (OLT) Optical add-drop multiplexer (OADM) Optical cross-connect (OXC)
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school 12
9/24/2009
Definitions
Lightpath topology:
Logical (virtual) topology
Physical topology:
Fiber topology
Grooming:
Kind of time multiplexing: packing a low speed channels into higher speed channels.
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school
13
Objective
Seattle New York
Given a traffic matrix (a forecast) and a fiber (physical) topology: design the network that fits the traffic forecast or/and optimize the (existing) network
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school
14
9/24/2009
Network design
Given
the fiber (physical) topology and traffic matrix (obtained by forecasting): in packets/second
Determine
the lightpath (virtual or logical) topology (Lightpath topology design: LTD): grooming physical routes through the network and wavelength assignment (RWA): map the LTD into the physical topology
15
Heuristic solution
Hard to determine th li ht th t H dt d t i the lightpath topology l jointly with the routing and wavelength assignment Split into separate LTD and RWA problems
Solve the LTD problem and then realize the obtained LTD within the optical layer (i.e. for the obtained LTD solve RWA problem).
16
9/24/2009
LTD:
Given: Gi
Determine the topology and routing of packets p gy g p Objective (an optimization problem):
Minimize the maximum load that any lightpath must carry
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school 17
RWA:
9/24/2009
RWA approaches
ILP formulation Heuristic
Routing sub-problem Wavelength assignment sub-problem
19
Routing sub-problem
Fixed routing
Fixed shortest-path routing
Fixed-alternate routing
Routing table at each node that contains a list of fixed routes to each destination
Adaptive routing
The route is chosen dynamically, depending of the network state
Fault-tolerant routing
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school 20
10
9/24/2009
Routing algorithms
Shortest Path selects the shortest source-destination path (# of links/nodes) Least Loaded Routing avoids the busiest links Least Loaded Node avoids the busiest nodes
21
WA sub-problem
Static WA sub-problem sub problem
Graph coloring
Dynamic WA sub-problem
Random (R) Wavelength Assignment First-Fit (FF) Least-Used (LU)/SPREAD d ( )/S Max-Used (MU)/PACK Least Loaded (LL)
etc.
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school 22
11
9/24/2009
Construct a graph G so that each lightpath in the system is represented by a node. Undirected edge between nodes in the G if the corresponding lightpaths share a physical link Color the nods of G such that no two adjacent nodes have the same color.
23
First-Fit (FF)
All wavelengths are numbered Assign the first available wavelength
Least-Used (LU)/SPREAD
Select the wavelength that is the least used in the network g
Max-Used (MU)/PACKED
Select the most used wavelength in the network
24
12
9/24/2009
Multiplexer
To merge the incoming channels
13
9/24/2009
OADM
Functions
Provide lightpaths Protection switching (rerouting) Performance monitoring Wavelength conversion
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school 28
14
9/24/2009
Types of OXCs
Electrical or optical core
Transparency Cost Size
Optical signal
Transparency
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school 29
All-optical OXCs
Transparency at the cost of:
Grooming
Higher demand for lightpaths
No aggregation of low bitrate demands
Wavelength conversion
Higher bl ki Hi h blocking of lightpath demands f li ht th d d
Signal regeneration
More constrained routing of lightpaths
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school 30
15
9/24/2009
32
16
9/24/2009
J. Chen, A. Jirattigalachote, Novel Node Krakow Sept. 29, 2009 L. Wosinska and L Thyln,L. WosinskaArchitectures for Wavelength-Routed WDM Networks with Wavelength Conversion Capability, in Proc. of ECOC08, Brussels, Belgium, September 2008 Tutorial at BONE Summer school
33
Performance evaluation
J. Chen, A. Jirattigalachote, Novel Node Krakow Sept. 29, 2009 L. Wosinska and L Thyln,L. WosinskaArchitectures for Wavelength-Routed WDM Networks with Wavelength Conversion Capability, in Proc. of ECOC08, Brussels, Belgium, September 2008 Tutorial at BONE Summer school
34
17
9/24/2009
35
18
9/24/2009
OPS Networks
Large capacity High bandwidth efficiency Rich routing functionalities Great flexibility and reliability
37
Increase flexibility
Problems
Technological problems
Optical control functions Synchronization Optical buffering
19
9/24/2009
Multiplexing of lower-speed streams (and reverse operation, i.e. demultiplexing) Contention resolution (e.g. buffering of packets if output busy)
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school 39
Electronic packet switching typically rely on the time domain by means of queuing What about optical packet switching ?
Queuing in optical domain is difficult
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school 40
20
9/24/2009
Increases delay and network load Creates variable delays and potentially reordering
switch fabric
OC OC OC
switch fabric
OC OC OC
21
9/24/2009
L. Wosinska and G. 2009 high capacity optical networks, in Proc. NFOEC02, Dallas, Krakow Sept. 29,Karlsson, A photonic packet switch for L. Wosinska Texas, September 2002 Tutorial at BONE Summer school
43
COMPARISON
Architecture
Buffer I Dedicated II Shared Low High Complex
L. Wosinska and G. 2009 high capacity optical networks, in Proc. NFOEC02, Dallas, Krakow Sept. 29,Karlsson, A photonic packet switch for L. Wosinska Texas, September 2002 Tutorial at BONE Summer school
44
22
9/24/2009
Comparison. N=20
A1
A2
L. Wosinska and G. 2009 high capacity optical networks, in Proc. NFOEC02, Dallas, Krakow Sept. 29,Karlsson, A photonic packet switch for L. Wosinska Texas, September 2002 Tutorial at BONE Summer school
45
Optical buffering
Fiber delay lines (FDLs)
Not random access Require synchronization Supported packet format
Constant packet size Some configurations support variable packet size
A certain granularity
46
23
9/24/2009
Cheap and easy to manufacture Several kilometers long Slow down factor - zero No flexibility in terms of storage time Requires synchronization Many architectures proposed to introduce variable delay
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school 47
EIT
48
24
9/24/2009
EIT, cont.
coupling beam
Arriving IP packet
Memory cell
Optical fiber
49
EIT, cont.
Phase 1 : writing
25
9/24/2009
EIT, cont.
Phase 1 : writing
Cell length The memory cell needs to be long enough to fit the entire packet IP packet of 1500bytes at 2,5Gb/s is 1,4 km long in free space and about 1km long in an optical fiber
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school 51
EIT, cont.
Phase 2 : storage The coupling beam is turned off
52
26
9/24/2009
EIT, cont.
Phase 3 : reading The coupling beam is turned back on
53
EIT, cont.
Variable coupling power
The packet is slowed down in the cell No storage of light We regulate the slowdown factor by varying the coupling power
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school 54
27
9/24/2009
EIT, cont.
material Quantum dots slow down factor 40 in room temperature 107 in very low temperature 105 storage time 8.7ns
Atomic vapor
Slow down factor and storage time depend on the material, temperature, coupling power, bandwidth and wavelength
Krakow Sept. 29, 2009 L. Wosinska Tutorial at BONE Summer school 55
Optical cavities
Optical cavities use optical resonance in photonic structures
Slow down factor of 104 (depending on the number of side cavities) Storage time: 50 ns Chip scale implementation of the system foreseeable
56
28
9/24/2009
Tuning of the intensity of the control field Temperature and mechanical stress Cost
57
Comparison
storage time EIT
Up to 0.5 ms Order of cm Close to 0K or 80C Depends on the material
cell size
temperature
bandwidthwavelength
Optical cavities
Order of ns Size of a chip Room temp. No limitations
58
29
9/24/2009
Example 1:
OPS with hybrid buffer.
Optical l demultiplexer Parallel electrical p and optical buffer positions Switching matrix Optical inputs Optical outputs
KrakowL. Wosinska, J. Haralson, L. Thyln, Benefit of Implementing Novel Optical Buffers Sept. 29, 2009 L. Wosinska in an Asynchronous Photonic Packet Switch, in Porc. ECOC04, Stockholm, Sept. 2004 Tutorial at BONE Summer school
59
ASSUMPTIONS
The traffic load is uniformly distributed between the outputs. The traffic load at all inputs is identical.
B u ffe r
1
. . . . . .
N N + 1 N + N 1 N
1 N
. . .
. . .
The transparency class (i.e. packets that can not be converted to the electrical signal) represents 20% of the total traffic
KrakowL. Wosinska, J. Haralson, L. Thyln, Benefit of Implementing Novel Optical Buffers Sept. 29, 2009 L. Wosinska in an Asynchronous Photonic Packet Switch, in Porc. ECOC04, Stockholm, Sept. 2004 Tutorial at BONE Summer school
60
30
9/24/2009
load
1 calculated
KrakowL. Wosinska, J. Haralson, L. Thyln, Benefit of Implementing Novel Optical Buffers Sept. 29, 2009 L. Wosinska in an Asynchronous Photonic Packet Switch, in Porc. ECOC04, Stockholm, Sept. 2004 Tutorial at BONE Summer school
61
1,E-01
1,E-02
1,E-03 0 2 4 6 8 10 12 14 16
The loss probability goes down to a certain point and than stays constant as the buffer increases.
KrakowL. Wosinska, J. Haralson, L. Thyln, Benefit of Implementing Novel Optical Buffers Sept. 29, 2009 L. Wosinska in an Asynchronous Photonic Packet Switch, in Porc. ECOC04, Stockholm, Sept. 2004 Tutorial at BONE Summer school 62
31
9/24/2009
The lowest achievable packet loss probability for a given number of buffer positions reaches a limit that cannot be overcome by increasing the maximum storage time.
KrakowL. Wosinska, J. Haralson, L. Thyln, Benefit of Implementing Novel Optical Buffers Sept. 29, 2009 L. Wosinska in an Asynchronous Photonic Packet Switch, in Porc. ECOC04, Stockholm, Sept. 2004 Tutorial at BONE Summer school 63
Storage time of 0.5ms is enough to obtain any value of loss probability for any traffic load.
KrakowL. Wosinska, J. Haralson, L. Thyln, Benefit of Implementing Novel Optical Buffers Sept. 29, 2009 L. Wosinska in an Asynchronous Photonic Packet Switch, in Porc. ECOC04, Stockholm, Sept. 2004 Tutorial at BONE Summer school 64
32
9/24/2009
Example 2:
A1
JiaJia Chen and L. Wosinska, Krakow Sept. 29, 2009 Novel Architectures of Asynchronous Optical Packet Switch, in Proc. of European L. Wosinska Conference on Optical Communication ECOC07, Berlin, Germany, September 2007 Tutorial at BONE Summer school
65
Evaluation
JiaJia Chen and L. Wosinska, Krakow Sept. 29, 2009 Novel Architectures of Asynchronous Optical Packet Switch, in Proc. of European L. Wosinska Conference on Optical Communication ECOC07, Berlin, Germany, September 2007 Tutorial at BONE Summer school
66
33
9/24/2009
Summary
Switched networks Photonic circuit switching
WDM network design: solving offline LTD and RWA
Objective: minimize number of wavelengths
Dynamic scenario
Oblective: minimize blocking probability
68
34