Multi-Domained, Multi-Homed
Mobile Networks
Mobile Platform Internet (MPI)
http://www.multicasttech.com/mpi/
mailing list:
MPI-subscribe@multicasttech.com
Terry Davis – Boeing (terry.l.davis@boeing.com)
Will Ivancic – NASA Glenn (william.d.ivancic@nasa.gov)
1
Common Sectors
• Aviation
• Maritime
• Trains
• Trucking (Shipping)
• Automotive
• Others ???
2
How Do You Select and
Implement the Routing Path?
High Speed SatCom Network Entertainment Low Speed SatCom Network
• Globally Available • Globally Available
• Affected by Weather • Low Bandwidth
• Higher Bandwidth Mobile Network • High Latency
• High Latency • Very High Cost
Command and
• High Cost Control Operations • Redundant
Low Rate VHF
• Reliable High Speed Terrestrial
• Low Latency • Not Available when Mobile
High Speed LOS Network • High Bandwidth
• Globally Available • Low latency
• High Bandwidth • Lower Cost
• Low Latency Internet
• Lower Security
• Moderate Cost
Destination Network
Destination Network
(for Operations)
(for Entertainment)
Destination Network
How do you decide which path the data should take? (for Command & Control)
How do you cause the network(s) to route the data via this path? 3
Aviation Specific Issues
• Safety of Life / Safety of Flight
– Time-Critical command and control for Air Traffic Control
• Fast convergence time is essential!
• New radio link technologies are “uncertified” for Air Traffic
Control / Air Operations Communications (ATC/AOC)
• Regulatory requirements force network design
• Three independent network domains
– (required for regulatory, QoS, & security)
– Passenger & In-Flight-Entertainment
– Airline Operations
– Air Traffic Control
• Service providers may be authorized to carry one, two, or all
services.
• ATC will be a “closed network”
4
• Multiple security and authentication architectures
Airplane Communications
5
In-Air Communication
• Multiple networks with varying criteria for utilizing
different links
– Aircraft Control Domain
– Airline Information Services Domain
– Passenger Information and Entertainment Services
Domain
• Often multiple links will be active to the same
domain simultaneously.
• May need to have connectivity to 10 or more
ISPs depending on what airports one flies into
– Need to autonomously connect to service providers
– Each airport controls the ISP contracts
6
ISO Aeronautics Telecommunication Network
(ATN) Island Routing Domain Confederation
Mobile RD
Mobile RD
Another
ATN Island ATN Backbone RDC
ATN TRD ATN TRD
Mobile RD
ATN Island RDC
ATN TRD
ATN ERD
ATN ERD OSI IDRP Routing
• Controlled topology to ensure
rapid route prorogation
Routing Protocols requires
ERD – End Routing Domain
ownership/coordination of
RD – Routing Domain
infrastructure
RDC – Routing Domain Confederation
• Injecting routes 7
TRD – Transit Routing Domain
• Not deployed due to cost
E-BGP Control
BGP/OSPF for Air Traffic requiresthe
If you don’t own
configuration with each
infrastructure, you
Autonomous Systems.
probably will not be
This is a scalability
Air Ground allowed to inject routes!
problem.
BGP/OSPF
BGP
OSPF
Mobile-1
AREA 1
BGP/OSPF
BGP
Mobile-2 OSPF
AREA 0
BGP/OSPF
OSPF
AREA N
BGP
BGP/OSPF
Mobile-N
This may work for ATC, but not for AOC or Passenger Services 8
(AOC is a closed network and will not propagate routes to the world)
Multiplexing at the Radio
ATC
ATC
NEMO-1
NEMO-1 MUX Satellite
HA
AOC
AOC
NEMO-2
NEMO-2 MUX WiFi
HA
Passenger
Passenger
NEMO-3
NEMO-3 MUX VHF
HA
If one requires a separate nemo
router for each domain, then, failure
to Multiplex requires a radio for each
interface! 9
Antenna Systems
Note, this picture does not show:
• Satellite links
• Passenger service links
• Gate links (WiFi)
• Gate links (umbilical cord)
10
SATCOM
Communication Multiplexing AERO-1
and Display
at the Router SATCOM
AERO-HH
Mobile
Air Traffic Network 1 VHF
Management Voice/DATA
LAN
Mobile
Router HF
Mobile Voice/DATA
Operations Network 2 NEM0-1
LAN NEMO-2 INMARSAT
(Avionics) Swift 64
NEMO-3
High-Rate
Satellite
Sensor Controller
(Optional Display)
WiFi Max
Mobile
Network 3 GateLink
Passenger Cellular
Services
Future Links
11
Global HAHA for NEMO
Primary
VPN
X VPN
Endpoint
Secondary
Home Agent
Endpoint
Home Agent
Secondary Home Agent
provides pseudo route optimization
which is useful over long distances
Beware of VPN tunnels.
If security is not considered up front, it may negate route optimization! 12
Global HaHa
• Draft allows for
– Distribute HAs at IP layer
• HA to HA communications at layer 3 to get rid of home link in
configurations where it is not needed
– MIP proxy for local mobility management
• to allow a mobile router to bind locally as it is roaming far away
from any of its own home agents
– Pseudo route optimization in the infrastructure
• No Security
• Concerns
– How is routing accomplished to get to appropriate HA?
– How many home agents will aviation need?
– How fast can one switch to a new home agent? 13
Monami6 for Aeronautics
14
Multiple Links to Aircraft
• Failover for safety-of-life critical control
traffic
• Separation of governmental control, airline
corporate, and passenger network traffic
– Differing priorities on same link
– Differing link forwarding selection according to
policy
15
Policy-Based Routing,
All Links Active
P-DATA
Mobile High speed link P-DATA
Router P-DATA P-DATA
AOC
Home
int1
Agent P-DATA
ATC Low latency link
AOC AOC
P-DATA int2 ATC
Reliable link
int3
ATC ATC AOC
Routing Routing
Policy Policy
16
Policy-Based Routing,
Critical Link Active
P-DATA Mobile High speed link
Router P-DATA
AOC Home
int1 Agent
ATC Low latency link
AOC
P-DATA int2 ATC
Reliable link
int3
ATC ATC
Routing Routing
Policy Policy
17
Policy-Based Routing,
Passengers Link Active
P-DATA High speed link
Mobile P-DATA
Router P-DATA P-DATA AOC ATC
AOC Home
int1 Agent P-DATA
ATC Low latency link
P-DATA int2 AOC
Reliable link
int3
ATC
Routing Routing
Policy Policy
18
Multiple Care-of Addresses
• Some governments provide VHF/UHF air-
to-ground network infrastructure
• Multiple commercial service providers for
satellite and other links
Multiple HAs
• Removes potential single-point-of-failure
for safety-of-life traffic
19
Within Monami6 ...
• Assume an aircraft has a NEMO MR with X links
and care-of addresses, and Y geographically
distributed home agents
– X*Y bidirectional tunnels
– But only a few active at one time
• At MR, downlink tunnel selection (CoA and HA
pair) based on policy for traffic class of each
packet
• At HA, uplink tunnel selection (CoA) based on
policy
• Different security mechanisms for each tunnel
based on policy 20
Desires
• Multiple CoA option works for NEMO
– GOOD based on draft-ietf-monami6-
multiplecoa-00
• Support use of flow binding option for
NEMO HA-MR tunnel selection
– draft-soliman-monami6-flow-binding-02 is
currently worded very mobile-node (not
NEMO) centric, although seems to work well
for NEMO too
21
Recommendations / Desires
22
Recommendations / Desires
• Need Route Optimization for Large Mobile
Platforms that move over large geographical
areas.
– Global HAHA is a potential
– Other proposals also should be considered
• Including work being done in IRTF MOBOPTS
• It is highly desirable to have multiple NEMOs in
a single router (multi-domained nemos)
– What are the security implications of this?
• Need to examine Security Architecture relative
to mobility mechanisms
– If we are not careful, security may undo route
optimization.
– Ensure the IKEv2 and revised IPsec architecture are
considered 23
Recommendations / Desires
• Aspects of MANET Autoconfiguration may be
applicable to traditional routing for large mobile
systems regarding the air/ground interface
addressing.
• MONAMI6 currently appears to be handling the
needs of large mobile platforms.
• It would be advantageous if the filtering
mechanisms and policy-base routing for
monami6 worked with a hybrid
nemo/mip/traditional routing configurations.
24
Non-IETF (?) Issues or Not?
• Automated access to networks –
particularly access to radio systems
(Internet Docking for Mobile Networks)
• Implementation solutions are a
combination of architecture, protocol and
policy. Is it appropriate to have an
“Informational document” for large mobile
platforms?
25
Mobile Platform Internet (MPI)
• Web Page:
– http://www.multicasttech.com/mpi/
• To Join to the mailing list, E-mail to:
– MPI-subscribe@multicasttech.com
26