The Interplanetary Internet Challenges for the Internet's Evolution TERENA Networking Conference May 2000 Scott Burleigh, Vint Cerf Bob Durst, Adrian Hooke Keith Scott, Eric Travis, Howard Weiss
A Candidate Sharing of Issues and Technologies Generally high bandwidth-delay products Possible data loss due to bit-errors and/or transient link outages Potentially asymmetric data rates Satellite, Wireless Surface to Orbit Relay Power constrained end systems Episodic/disjoint connectivity Networks may need to be self-organizing Surface Operations Ultra high bandwidth-delay product typically >> transaction size ‘Ping-pong of bursts’ replaces streaming Internet backbone Long Haul to Earth Channels often periodically unidirectional Need for progressive/selectable reliability Need for store-and-forward delivery Mobile/ Roving applications Mars Moon Earth
The Basic IPN Concept: Construct a “Network of internets” Deploy standard internets in low latency remote environments (e.g., on or around remote rocks) Connect distributed internets via an interplanetary backbone that handles the high latency deep space environment. Create gateways and relays to interface between low and high latency environments
i I Wired Tetherless Deployed Internets Stable Backbone Security Interplanetary Gateways Inter-Internet Dialog and Nodes IPN Technology Thrust Areas earth.sol IG I i ii i i i i i i i i i I i i i i I i i i I mars.sol luna.sol
The Good News Is... Much of the ongoing work in terrestrial mobile ad hoc networking is interesting, relevant, and timely We can probably apply these emerging terrestrial technologies to our in-situ space communications needs, rather than developing them from scratch But… Anything deployed on remote rocks is going to be slow to evolve
What’s a Backbone? A set of high-capacity, high-availability links between network traffic hubs Terrestrial backbone links are between hubs like Lisbon and Paris. Interplanetary backbone links are between hubs like Europa and Mars. Availability is a matter of perspective... Deployed Internets Stable Backbone Security Interplanetary Gateways Inter-Internet Dialog and Nodes
Differences Between Terrestrial and Interplanetary Backbones Terrestrial Backbone Interplanetary Backbone <.1 10 to 10,000 WiredUntethered, moving Copper, glass Space “low” Very high “low” High (power is costly) Delay (light sec) Connectivity Medium Deployment $ Operations $ Repair, upgrade $ “low” Very high
On the Interplanetary Backbone: Communications capacity is expensive Bits count Round Trips hurt Interactive protocols don’t work Internet protocol suite doesn’t scale well with increasing latency Negotiation is impractical Reliable in-order delivery takes too long Protocols need to be connectionless Congestion control and flow control are difficult Reliance on forward coding versus retransmission for error recovery Custody, store-and-forward data transfer is fundamental “Chatty Telephony” gives way to “Bundled Mail” as the model of operations
Communication in a Micro-Power Environment: It's a Full-Stack Job Link Layer: Effects of very low SNRs in conjunction with QOS Multiple simultaneous coding schemes needed? Resource reservation/interference avoidance Signaling of link status to upper layers: availability; capacity and congestion status; error status Network Layer: MANETs with both fast- and slow-moving elements Changing resource availability Synthesis and propagation of link status Resource allocation: integrated services?, adaptive Self organization
Communicating in a Micro-Power Environment: It's a Full-Stack Job Transport Layer: Power-efficiency requires careful balance between link layer coding and transport layer retransmission Adaptive to changing network conditions QOS-responsive Tolerant of link outages and mixed losses Application Layer: Service location in infrastructure-free environment Efficient network management and control Health and status of mobile nodes (not just networking aspects)
While communications with distant rocks can be exciting... The IPN is really about exploring the evolutionary paths the Internet may take as the dynamic range of interconnected environments continues to increase... Autonomy - functional and/or administrative Mobility - not just nodes, but entire networks Impedance Matching - keeping the backbone fair, but allowing subnets to be aggressive While preserve the end-to-end semantics of communications
Bundle Space Bundle space supports end-to-end transfer across IPN domains and/or heterogeneous network protocol stacks Bundle Transport Application Transport Network Transport Application Network Transport Network Bundle Network of internets spanning dissimilar environments
Inter-Internet Dialog: Design Principles Names are the means of reference Routing between domains based upon names Late-Binding Separate addressing domains for each internet; names converted to local addresses in destination domain Indirection Inherent dependence on intermediate relay agents Custodial transfer “Bundles” as a common end-to-end mailing mechanism Deployed Internets Stable Backbone Security Interplanetary Gateways Inter-Internet Dialog and Nodes
Bundled Data Transfers Traffic is inherently non-interactive A single self-contained request generates a single self-contained response Distinctly taciturn, “non-chatty” dialog Bundling Protocol lives above the local transport protocols, but below end-to-end applications Minimally, carries names end-to-end in “metadata” Potential applicability to new Internet applications Deployed Internets Stable Backbone Security Interplanetary Gateways Inter-Internet Dialog and Nodes
Single Name Space, Late Name-to-Address Binding(s) Internet Interplanetary Backbone Name Space - Common Across All Internets Name-to-Address Binding Space A Name-to-Address Binding Space B Name-to-Address Binding Space C mars.sol.earth.sol.ipn.sol
From Telephony Back to the Pony Express Custody Transfers and Bundles Custody Transfers Interplanetary Backbone Internet “Return Receipt” “Shipping Receipt and Tracking #” R R AA IPN Relay: mitigation for R 2 effects; no custody transfer A R Bundle Agent: builder and consumer of bundles IPN Gateway: custody transfer and routing between IPN domains G G G G
IPN Security Security of user data flowing through the IPN Security of the IPN backbone Deployed Internets Stable Backbone Security Interplanetary Gateways Inter-Internet Dialog and Nodes IG I i ii i i i i i i i i i I i i i i I i i i I mars.sol luna.sol earth.sol
IPN Security Requirements access control to the IPN will be required because space-based assets will have limited available resources. authentication will be required to perform access controls. data integrity will be required to assure that what was sent is received. data privacy will be required to assure that unauthorized users cannot obtain information.
We can only have end-to-end security where there are end-to-end protocols, so: - security cannot be performed in the IPN at transport or below IPSEC In the IPN, end-to-end protocols (e.g. TCP) are terminated at the IPN gateways. End-to-end like security can only be applied to the data (e.g.,TLS/SSL, S/ ) - is TLS/SSL the right answer (many round- trip negotiations)? NO! - is secure model is a better fit? YES - key management - “PKI to the Planets?” IPN Security Constraints
Mars may need women, but... The IPN Needs Participants Help expand the gene pool: Robert Rumeau (CNES) Centre National d'Etudes Spatiales Toulouse CEDEX Eric Travis (NASA) Global Science and Technology, Inc. Greenbelt, Maryland USA (Dinosaurs Welcome; Spacecraft optional)