Post on 26-Jan-2016
description
CCSDS Fall 2009 Meeting ESTEC 28/10/2009
Progress Report on
ESA/ESOC DTN Testbed
V. Tsaoussidis, DUTH – Greece
ESA DTN Testbed
CCSDS Fall 2009 Meeting 2/13
Fundamental Issues: Does it work? Can it work for Space?(Where) Is it functionally richer than CFDP, for example?Is it efficient enough, performance-wise?Can we quantify the gain from the use of DTN?
Departing Question:What are the driving needs for DTN?
No, not just interoperability
DTN should not just work, but work where others fail. – By Definition.Evaluate it in stressed environments – search every aspect of it.
When we say search, we mean search.
3
Design a testbed for
Dynamic control of network parameters• Emulation of fundamental network parameters (bandwidth, PER, connectivity
availability)• Realistic and dynamic adaptation to parameter changes in real-time (packet
loss rate emulating solar storms, etc.)
Scalability• Efficient scaling over a large number of communication nodes
Transparency• Network emulation should be transparent to upper layer protocols and
applications
Flexibility• Emulation of any desirable communication topology• Incorporation of new protocols, architectures, mechanisms• Interoperability with other DTN testbeds• Reusable infrastructure
CCSDS Fall 2009 Meeting 3/13
Architecture
CCSDS Fall 2009 Meeting 4/13
Architecture: Administrative Part
Graphical User Interface• Input of experiment parameters regarding:
Nodes (number, data production – consumption, storage size)Links (bandwidth, error rate, propagation delay, availability)Available protocols
• Modification of parameters while the experiment is on progress
• Real-time presentation of testbed statistics and status information
Kinematics Modeling System• Creation of the communication scenarios• Creation of the corresponding control data for the nodes
Central Management System• Handling of the communication between the various testbed components• Exchange of control data and status reports with the Emulation Nodes
CCSDS Fall 2009 Meeting 5/13
Architecture: Emulation Part
Emulation Nodes• Control Daemon
sets node parametersgenerates status reportscommunicates with Central Management System through the Control Plane
• Node Protocol StackProtocol stack under evaluation
CCSDS Fall 2009 Meeting
Individual PCs communicate through the Data plane, exchanging files such as images, measurements, etc.
6/13
Testbed Topology
CCSDS Fall 2009 Meeting
Ten Emulation Nodes• Suitable for complex space
communication scenarios
Intercontinental Link with MIT - Boston
• Suitable for terrestrial scenarios
Geostationary Link (HellasSat Geo Satellite)
• Suitable for low-orbit scenarios
7/13
Protocol Stack
CCSDS Fall 2009 Meeting
DTN Implementation: Interplanetary Overlay Network (ION)
• Bundle Protocol• Asynchronous Message Service (AMS)• Licklider Transmission Protocol (LTP)• Contact Graph Routing (CGR)• Interoperates with DTN2
An advanced application layer protocol is
required (e.g. CFDP)
CCSDS File Delivery Protocol (CFDP)
• Automatic, reliable file transfer• File segmentation• Remote file management and directory
listing Lacks dynamic routing support
8/13
Complicated Scenario
CCSDS Fall 2009 Meeting 9/13
Progress so far
Integration of CFDP into ION protocol stack• Performance evaluation of CFDP over ION versus CFDP as a stand alone
application
Integration of CCSDS Space Packet protocol• Implementation of the protocol’s basic functionality
Evaluation of several DTN routing protocols • Comparison of Contact Graph Routing (CGR) with Probabilistic Routing
Protocol using a History of Encounters and Transitivity (PRoPHET) and Flood routing
Design of efficient space oriented DTN transport protocols
• DS-TP• DTTP
CCSDS Fall 2009 Meeting 10/13
Progress so far
Integration of CFDP into ION protocol stack
Objective: Exploit CFDP file management features• CFDP on endpoints in unacknowledged mode• All intermediate nodes run ION• Reliability is achieved by the underlying network (BP, LTP)
Integration:
• Middleware Application
• Receives CFDP PDUs• Each CFPD PDU is encapsulated into a bundle• Requires mapping between CFDP ID and DTN EID
Validation – Evaluation: CFDP over BP/TCP and CFDP over BP/LTP
CCSDS Fall 2009 Meeting 10/13
Progress so far
Integration of CCSDS Space Packet Protocol into ION
Objective: Deployment of an architecture possibly used by ESA in future DTN space communications • Basic implementation of the Space Packet Protocol below LTP
• Each LTP segment is encapsulated into a Space Packet
• LTP Engine Number to APID mapping
• Independent Packet Sequence Count for each LTP span
• Intermediate nodes supporting only the Space Packet Protocol
• Static routing of space packets based on APID between non-DTN nodes
CCSDS Fall 2009 Meeting 10/13
Progress so far
Evaluation of several DTN routing protocols
Objective: Comparison of Contact Graph Routing (CGR) with Probabilistic Routing Protocol using a History of Encounters and Transitivity (PRoPHET) and Flood routing
• Observations
• CGR outperforms both PRoPHET and Flood Routing when delay is in the order of seconds
• PRoPHET does not perform well even for short delay values
• CGR needs to have predetermined contact plans in order to operate and cannot cope with opportunistic contacts
CCSDS Fall 2009 Meeting 10/13
Work-in-Progress
Evaluation of Fragmentation and Bundle-size performance Design and implementation of an efficient DTN routing scheme, using parameters such as
• Resource availability• Custody requirements• Foreign agency assets exploitation
Implementation of an advanced Kinematics module• Dynamic adjustment of link characteristics, based on real planet and satellite
trajectories, random solar storms etc.
CCSDS Fall 2009 Meeting 11/13
Suggestion
16
Test functionality – prove it is operational
Test efficiency – custody, fragmentation, routing
Test efficiency – when others fail.