cost action 272 “packet-oriented service delivery via ...€¦ · on routing and traffic...
TRANSCRIPT
COST Action 272 “Packet-Oriented Service Delivery via Satellite”
On Routing and Traffic Engineering in Dynamic Satellite
Constellation Networks
TD-03-003-S
Anton Donner, Matteo Berioli, Markus Werner
German Aerospace Center (DLR) Institute of Communications and Navigation
{Anton.Donner,Matteo.Berioli,Markus.Werner}@dlr.de
1
Institute of Communications and Navigation 1
On Routing and Traffic Engineering in DynamicSatellite Constellation Networks
Anton Donner, Matteo Berioli, Markus Werner
German Aerospace Center (DLR)Institute of Communications and Navigation
Oberpfaffenhofen, Germany
Institute of Communications and Navigation 2
Outline
4 LEO/MEO Satellite Networks
4 MPLS Over Satellite
4 Route Computation Effort
4 Re-routing Algorithm and Performance
4 Summary
2
Institute of Communications and Navigation 3
Low/Medium Earth Orbit (LEO/MEO) Satellite Constellations
Owing to Van Allen belts 3 basic types of satellite constellations:
702 h700-1’500 kmLEO154-6 h10’000 km MEO324 h36’000 kmGEO
N° of satellite
PeriodAltitude
M-Star (LEO) inclined chosen as representative:
4 72 satellites
4 20° minimum elevation
4 2122 km footprint radius
Institute of Communications and Navigation 4
Considered LEO Satellite ConstellationM-Star allows the creation of a permanent, regularly meshed network topology of Intersatellite Links (ISLs).
Each satellite has:
•2 intra-orbit links•2 inter-orbit links
A certain number of ground stations has been considered, too :4 Continuously changing Ground Satellite Link (GSL) topology
3
Institute of Communications and Navigation 5
ConstellationMovement
Each ground-station chooses nearest satellite
⇒ worst case scenario
Institute of Communications and Navigation 6
Outline
4 LEO/MEO Satellite Networks
4 MPLS Over Satellite
4 Route Computation Effort
4 Re-routing algorithm
4 Summary
4
Institute of Communications and Navigation 7
Functional Blocks in an MPLS Based Satellite Network
Network state information• Link availability• Link capacity• Link delay
User behavior / QoS requirements• Admission Control
Route computation• Eventually able to guarantee QoS
LSP rerouting / creation / release master• Ingress LER or
Central Station
Institute of Communications and Navigation 8
Mapping MPLS on a Satellite Constellation
LERs on board
Continuously changing LERs
Each handover new negotiation
LERs on ground
Ground link inside MPLS domain
Possible to use MPLS rerouting features to manage handovers
(Processing power (LER) on ground)
MPLS Cloud
5
Institute of Communications and Navigation 9
Different possible scenarios
4Scenario 1: Distributed routing
4Scenario 2: Centralized routing, ingress LER maintains LSPs
4Scenario 3: Centralized routing, central station maintains LSPs
Institute of Communications and Navigation 10
(1) Distributed routing
Satellites and LERs maintain own LSDBs updated using OSPF-TE
4Direct adaptation of terrestrial approach
4Continuous flooding of network necessary (very high overhead)
4No sensible traffic engineering possible (e.g., re-routing of low priority traffic)
6
Institute of Communications and Navigation 11
(2) Centralized Routing
One LSDB in Central Control Station (CCS) LSPs are still maintained by ingress LERs
4CCS has global view of the network
4 Traffic engineering ⇒ better network utilization
4New protocol CCS-ground station necessary to bring CCS “orders” to LERs
Institute of Communications and Navigation 12
LSPs are maintained by central station
4CCS has global view of the network
4 Traffic engineering ⇒better network utilization
4 Fast LSP update
4New protocol CCS-satellites necessary to bring CCS new labels to LSRs
4Perfect synchronization between satellite necessary switching to the new labels
4Problems for systems with satellite diversity
(3) Completely Centralized
7
Institute of Communications and Navigation 13
Outline
4 LEO/MEO Satellite Networks
4 MPLS Over Satellite
4 Route Computation Effort
4 Re-routing algorithm
4 Summary
Institute of Communications and Navigation 14
Computational Effort
Route computation effort estimated in terms of raw path computation events due to:
4 LSP setup
4 LSP rerouting (only owing to handover)
It depends on:
4 Ground station latitude (footprints in general are not equally distributed)
4 Choosing serving satellite (shortest path, max. elevation, max. visibility)
4 Incoming LSP creation/release process considered
8
Institute of Communications and Navigation 15
Route Effort Estimation
4 Each ground station creates LSPs obeying a Poisson Process of rate λ
4 Each LSP is established exponentially distributed lifetime of average 1/µ
λΛ =∑LER
+=+=+=
τµλα
τµλαλα 2
12
rernewnet NN
NRRR
α (= 0.4) reflects the fact that satellites typically serve ground stations at all only during a certain share of time (appr. share of landmasses )
⇒ Mean duration τ of a ground station in the satellite footprint is responsible for constellation-caused re-routing
Estimation only valid for long LSP lifetimes!
Institute of Communications and Navigation 16
ISL-delay.datGSL-delay.dat
Matlab
ISL & GSLdelay
Simulator Overview
Matlab
Graphical analysis
NamNetwork visualization
Packet levelanimation
out.txtdetailed.txt
out.nam
NS 2.1b9a & MNS_v2.0 Event-driven Simulator
MPLS signalingDynamic topology
Mns_routingRouting Algorithm
Process of Creation-Release requests
Sat-sim.tcl
Pre-calculation NS2 Simulation Visualization
Each simulation is divided into three steps:
1. Matlab precomputation of deterministic delay variation
2. Event-driven simulation based on Network Simulator (Ns-2) with MNS_V2.0 (MPLS) extension
Mns_routing is a C++ object which has been developed to manage LSP creation & rerouting
3. Graphical result visualization
9
Institute of Communications and Navigation 17
Outline
4 LEO/MEO Satellite Networks
4 MPLS Over Satellite
4 Route Computation Effort
4 Re-routing algorithm
4 Summary
Institute of Communications and Navigation 18
Minimum-Switched-Links (MSL) Re-routing algorithm
[ ]1,0
)(),(0
)(),(1
)()1()(
∈
∉∈
=
∆+⋅−=∆+
α
α
tpji
tpjiX
ttdXttc ijij
HANDOVER in the interval [t , t+∆t]
⇒ Re-routing
4 Link cost at time (t+∆t) …
4 … depends on link presence in the LSP at time (t)
4 α : tunable parameter
10
Institute of Communications and Navigation 19
MSL Re-routing Algorithm Performance (1/5)
0 200 400 600 800 1000 1200 1400 1600−5
0
5
10
15
20
25
30
35
40
45
Simulated time [s]
Num
ber
of a
void
ed li
nk s
witc
hing
s (A
α)
α = 0.3α = 0.5α = 0.7
0 200 400 600 800 1000 1200 1400 16000
50
100
150
200
250
300
350
400
450
Simulated time [s]
Tot
al n
umbe
r of
sw
itche
d lin
ks (
L α)
α = 0.3α = 0.5α = 0.7non−opt
Short-lasting LSPs (short simulation, high number of LSPs: 200)
Institute of Communications and Navigation 20
MSL Re-routing Algorithm Performance (2/5)
0 200 400 600 800 1000 1200 1400 1600−0.5
0
0.5
1
1.5
2
2.5
3
3.5
Simulated time [s]
Avo
ided
link
sw
itchi
ngs
per
LSP
(a α)
α = 0.3α = 0.5α = 0.7
Short-lasting LSPs (short simulation, high number of LSPs: 200)
11
Institute of Communications and Navigation 21
MSL Re-routing Algorithm Performance (3/5)
0 1000 2000 3000 4000 5000 60000
10
20
30
40
50
60
70
80
Simulated time [s]
Tot
al n
umbe
r of
sw
itche
d lin
ks (
L α)
α = 0.3α = 0.5α = 0.7non−opt
0 1000 2000 3000 4000 5000 6000−10
−5
0
5
10
15
20
Simulated time [s]
Num
ber
of a
void
ed li
nk s
witc
hing
s (A
α)
α = 0.3α = 0.5α = 0.7
Long-lasting LSPs (long simulation, small number of LSPs: 30)
Institute of Communications and Navigation 22
MSL Re-routing Algorithm Performance (4/5)
Long-lasting LSPs (long simulation, small number of LSPs: 30)
0 1000 2000 3000 4000 5000 6000−4
−2
0
2
4
6
8
10
Simulated time [s]
Avo
ided
link
sw
itchi
ngs
per
LSP
(a α)
α = 0.3α = 0.5α = 0.7
12
Institute of Communications and Navigation 23
MSL Re-routing Algorithm Performance (5/5)
Overall results: Saving of Signalling Effort
0.3 0.5 0.70
2
4
6
8
10
12
14
16
8.71 %
5.53 %
10.10 %
6.63 %
12.02 %
7.35 %
α
%
Short simulation (200 LSPs)Long simulation (30 LSPs)
Institute of Communications and Navigation 24
Summary
4 Suitability of MPLS for satellite constellations
4 Mapping of logical MPLS components on physical network
4 Re-routing effort estimation by analysis of re-routing events
4 Minimum-Switched-Links (MSL) re-routing algorithm and performance