Future Internet Resilience Summary of Networking Research at The University of Kansas ITTC

Transcription

1 Future Internet Resilience Summary of Networking Research at The University of Kansas ITTC James P.G. Sterbenz Department of Electrical Engineering & Computer Science Information Assurance, Communication & Network System Labs Information Technology & Telecommunications Research Center The University of Kansas June Sterbenz

2 ITTC Networking Research Major Themes Major related research themes future Internet architecture and infrastructure resilient and survivable networks information assurance and security disruptive and novel communication paradigms 10 June 2010 ResiliNets Overview 2

3 Collaborators ITTC Networking Research Collaborators and Funding regional: K-State, UMKC, UNL, national: Rutgers, Penn State, CMU, ORNL, international: U. Lancaster UK, ETH Zürich, TU-Munich, Funding NSF FIND, GENI, DoD DARPA, CTEIP, EU FP6 SAC, FP7 FIRE Industry: Sprint, 10 June 2010 ResiliNets Overview 3

4 Resilient Networks Motivation Increasing reliance on network infrastructure Increasingly severe consequences of disruption Increasing attractiveness as target from bad guys Internet is critical infrastructure interdependent with other CI, e.g. power grid 10 June 2010 ResiliNets Overview 4

5 Resilience Resilient Networks Resilience Definition provide and maintain acceptable service in the face of faults and challenges to normal operation Challenges faults unintentional misconfiguration or operational mistakes large scale disasters (natural and human-made) malicious attacks from intelligent adversaries environmental challenges (wireless, mobility, delay) unusual but legitimate traffic service failure at a lower level 10 June 2010 ResiliNets Overview 5

6 Resilience Scope Relationship to Other Disciplines Survivability many targetted failures Fault Tolerance (few random) Challenge Tolerance Traffic Tolerance Disruption Tolerance environmental delay energy mobility connectivity Robustness Complexity Trustworthiness Dependability reliability maintainability safety availability integrity confidentiality Security nonrepudiability AAA auditability authorisability authenticity legitimate flash crowd attack DDoS Performability QoS measures 10 June 2010 ResiliNets Overview 6

7 Resilience Architecture ResiliNets Strategy: D 2 R 2 + DR Real time control loop: D 2 R 2 defend passive active detect remediate recover Background loop: DR diagnose refine Defend Recover Diagnose Defend Refine Detect Remediate [ComNet 2010] 10 June 2010 ResiliNets Overview 7

8 Resilience Architecture ResiliNets Principles service requirements normal behaviour threat and challenge models metrics heterogeneity resource tradeoffs complexity state management self-protection connectivity redundancy diversity multilevel context awareness translucency self-organising and autonomic prerequisites tradeoffs enablers behaviour adaptable evolvable Resilience Prerequisites: to understand and define resilience Tradeoffs: recognise and organise complexity Enablers: architecture and mechanisms for resilience Behaviour: require significant complexity to operate 10 June 2010 ResiliNets Overview 8

9 Resilience Architecture Multilevel Resilience and Cross-Layering ResiliNets Cube multilevel protocol layers planes mechanisms D 2 R 2 +DR strategy D 2 R 2 control plane DR mgt. plane Cross-layering knobs and dials are metrics K,D N P adaptive applications & overlays disruption tolerant transport survivable topology fault tolerant network elements architecture & engineering application session E2E network HBH link MAC physical data plane ARQ EC ARQ management plane control plane FEC APR FEC CDMA TDMA UWB QAM cross-layer composable programmable & autonomic mechanisms redundant & diverse context-aware & adaptive Defend Detect Remediate Recover D iagnose R e fi ne D 2 R 2 +DR resilience strategy 10 June 2010 ResiliNets Overview 9

10 Resilience Quantification State Space: Operational Resilience Operational resilience minimal degradation in the face of challenges Resilience state remains in normal operation Normal Operation Operational State N Partially Degraded Severely Degraded S 0 10 June 2010 ResiliNets Overview 10

11 Resilience Quantification State Space: Service Resilience Service resilience acceptable service given degraded operation Resilience state remains in acceptable service Resilience R = area under trajectory for particular scenario resilience R over all scenarios Service Parameters P Acceptable Impaired Unacceptable Normal Operation S 0 Operational State N Partially Degraded Severely Degraded S c S c 10 June 2010 ResiliNets Overview 11

12 Resilience Quantification D 2 R 2 + DR Relationship to State Space Real time control loop: D 2 R 2 defend keeps toward origin passive active detect when leaves remediate pushes back recover back to origin Background loop: DR diagnose refine tightens trajectory Service Parameters P Acceptable Impaired Unacceptable Normal Operation Defend S 0 Operational State N Partially Degraded S r Recover S r Severely Degraded S c S c Remediate 10 June 2010 ResiliNets Overview 12

13 Resilience Evaluation Topology Generation: KU-LoCGen Generation of realistic topologies Multilevel hierarchy level 1: represents (tier 1) backbone level 2: represents access networks around a backbone PoP level 3: represents subscriber nodes Constrained generation geographic node location (infrastructure or population) constrained link location (based on exiting fiber runs) constrained cost (fixed + variable cost) graph-theoretic constraints for resilient diversity 10 June 2010 ResiliNets Overview 13

14 Resilience Evaluation Evaluating Challenges in State Space Topology generation use KU-LoCGen Challenge simulation random failures intelligent attacks degree, betweeness, etc. large scale disasters Example hurricanes, blackouts resilience of alternatives based on Sprint PoPs unacceptable impaired acceptable normal partially degraded degraded severely # of link cuts 10 June 2010 ResiliNets Overview 14 topology

15 challenge specification.txt node coordinates.txt ITTC Resilience Evaluation KU-CSM Challenge Simulation adjacency matrix.txt simulation description.cc ns-3 simulator (C++) plotted trace KU-CSM Challenge Simulation Module challenge specification describes challenge scenario network coordinates provide node geo-locations } adjacency matrix specifies link connectivity KU-LoCGen input to conventional ns-3 simulation run generates trace to plot results 10 June 2010 ResiliNets Overview 15

16 Resilience Evaluation KU-CSM Challenge Simulation Example: evolving area-based challenge example circle moving from Orlando to NY Performability analysis: packet delivery ratio PDR varies with # links nodes down 10 June 2010 ResiliNets Overview 16

17 Enabling Future Internet Research GpENI Overview Great Plains Environment for Network Innovation part of NSF GENI program affiliated with EU FP7 FIRE programme / ResumeNet project Programmable network infrastructure (L1 7) Midwest US optical backbone International testbed Conduct experiments in: future Internet architectures resilience and survivability experiment application end-to-end router topology VLAN lightpath Programmability Gush, Raven PlanetLab Quagga, XORP, Click VINI cross-evaluation with analyticaland simulation-based eval. [TridentCom 1 RF,photonics site-specific 2010] 10 June 2010 ResiliNets Overview GpENI Layer DCN

18 Enabling Future Internet Research GpENI Midwest Optical Node Cluster GpENI cluster 5 10 PCs GpENI mgt. L4: PlanetLab L3: prog. routers GbE switch prog. routers VINI, XORP, click, arbitrary site interconnection L2: GpENI/GENI VLAN SNMP cluster monitoring Ciena optical switch L1 GpENI interconnection PlanetLab GENIwrap prog. nodes GbEnet Ciena optical ctl. frwk. aggr. mgr. PLC VINI DCN GENI VLANs GpENI management & control site specific KUAR, sensor, GpENI optical backbone to Internet2 and KC SPP then to MAX 10 June 2010 ResiliNets Overview 18

19 Enabling Future Internet Research GpENI Midwest Optical Backbone Physical topology to Smith Ctr. KS (eventual link to CO) CCD C42 Gp ENI multiwavelength optical backbone current or imminent deployment 4 universities in 3 states 1 switch/year with current funding new nodes GpENI Ciena CoreDirector GpENI Ciena CN4200 GpENI node cluster patch WTC fiber C C42 Gp ENI Power Plant C-band n λs KSU KS KU/Qwest fiber Rathbone Hall SFBB fiber C Ellsworth Hall C-band n λs C42 Nichols Hall splice Gp ENI KU KS KU/Qwest fiber Qwest POP KC MO UNL NE Gp ENI Avery Hall Scott Center KU/Qwest fiber dark fiber CCD Internet2 POP KC MO 2 λs UNL (L3) fiber MOREnet fiber 4 λs C42 Gp ENI Newcomb Hall Ethernet Flarsheim Hall UMKC MO 10 June 2010 ResiliNets Overview 19

20 Enabling Future Internet Research GpENI European Expansion European GpENI partners Asia SDSMT 13 nations 24 research institutions ~120 nodes more under discussion DSU USD UNL UMC KSU KU UMKC IIT Internet2 IU GMOC ISCTE Lisboa HEAnet UC Dublin JANET Lancaster Cambridge Bern DANTE ETH SWITCH UPC Barcelona G-Lab Kaiserslautern Karlsruhe Konstanz U-Zürich Passau Wien München Tampere UT Uppsala SICS TKK Helsinki KTH Stockholm Karlstads Simula NORDUnet Warszawa GÉANT2 Skt. Peterburg IIRAS Bucharest Moscow to Beijing Bilkent 10 June 2010 ResiliNets Overview 20

21 Enabling Future Internet Research GpENI Asian Expansion IIT Guwahati Europe CUC Beijing POSTECH Asian GpENI partners 3 nations 5 research institutions 25 nodes more under discussion ERNET IIT Mumbai IISc Bangalore APAN Internet2 Asia SDSMT DSU USD UNL UMC KSU KU UMKC IIT IU GMOC 10 June 2010 ResiliNets Overview 21

22 ITTC Networking Research Selected Project Examples Weather disruption-tolerant networking (Sterbenz) Highly-dynamic airborne networking Information security and privacy SDRs and cognitive networking Sensor networking (Sterbenz) (Luo) (Minden, Evans) (Frost) 10 June 2010 ResiliNets Overview 22

23 Mesh architecture WDTN Project Overview high degree of connectivity alternate diverse paths severely attenuated mm wave alternate mm, lower-freq. RF fiber bypass (competitor) Solution [INFOCOM 2009] reroute before link failures they occur P-WARP predictive routing image radar to predict weather XL-OSPF instantaneously reactive routing cross-layered with BER estimation CO/POP G 10 June 2010 ResiliNets Overview 23

24 Airborne Networking Project Scenario Very high relative velocity Mach 7 10 s contact dynamic topology Communication channel limited spectrum asymmetric links data down omni C&C up directional Multihop among TAs through relay nodes 10 June 2010 ResiliNets Overview 24 GS GW TA RN TAs TA test article relay node Internet TA RN GS GW GS ground station GW gateway

25 Airborne Network Project Protocol Stack and Interoperability GW RN or TA TA TCP AeroTP AeroTP TCP IP AeroNP AeroN RP AeroNP IP link/mac inet MAC inet MAC inet MAC link/mac PHY inet PHY inet PHY inet PHY PHY HMI apps gnet TmNS AeroTP: TCP-friendly transport AeroNP: IP-compatible forwarding TA peripherals AeroRP: routing [MILCOM 2008] 10 June 2010 ResiliNets Overview 25

26 InfoSec and Privacy Projects [Bo Luo PI] CAT: A node-failure-resilient anonymous communication protocol through commutative path hopping [INFOCOM 2010] protect the identity/privacy of communication participants group-based path probing & commutative path hopping: resilient to relay node failures Secure in-network operations for smart grids in-network operations: distribute operations (e.g. aggregation) into smart meters Secure operations: perform operations without revealing the data, using applied crypto methods 10 June 2010 ResiliNets Overview 26

27 SDR and Cognitive Radio Projects [Gary Minden and Joseph Evans PIs] KUAR: KU agile radio experimental system: wireless networking & radio research 5.8 GHz UNII band; independent 30MHZ tx/rx signal paths, signal processing is entirely in an FPGA and GPU Application sharing radio frequency spectrum with multiple users configure radio software for specific missions adaptation to dynamic RF environment and other users radio network control and resource management Cognitive networking new dynamic routing algorithms exploiting SDR technologies 10 June 2010 ResiliNets Overview 27

28 Rx Antenna SDR and Cognitive Radio Projects KUAR Diagrams Rx LNA Rx IF Filter I Rx Mixer Q ADC Digital DSP/FPGA/GP Phy packets Software Organisation Cos Tx Antenna Band Freq. Rx Freq Tx IFFilter I DSP/FPGA/GP DAC Phy packets Digital Tx Amp Tx Mixer Q Cos Tx Frequency LNA MGA Gain = GHz P1dB = +4.3dBm@6.0 GHz NF =1.8dB@6.0 GHz GHz (-77dBm) AD8347 AD ADF ADF SMV3300A 20 FOX801BE BGA MGA MGA MGA-545P MGA HMC488MSG8 100 ERA-1S M mA 440mA Lu mped (-100dBm) +8 db i -4dB -4 db +1 9dB Active RX Antenna Module I²C DPB Input Microstrip Active TX Antenna Module MGA-545P8 Gain = dB@5.8 GHz P1dB = +21dBm@5.8 GHz P SAT = +22dBm@5.8 GHz -5dB +17d B -3dB MC68HC08 Microcontroller -5dB MGA Gain = +17dB@6.0 GHz P1dB = +15dB m@6.0 GHz P SAT = +18dBm@6.0 GHz AD8347 I-Q DEMODULATOR 800 MHz G Hz HMC488MS8G CLo ss = 10 db P1dB = +8 dbm GHz LO Drive 0dBm (-66dBm) (-81dB m) LNA R I 0 L +19dB d B -5dB 0-30dB RX A ttenuation GAIN Control CONTROL 3.4 GHz SPI Bus (+3. 5dBm) (+3. 5dBm) OPTIONAL For use with passive anten na 3dB POWER DI VDER -2dB -1 0d B MGA Gain = GHz P1dB =+17dBm@4.0 GHz LO 3 +9dB (+25dBm) ERA-1SM +8dB i GHz Gain = +6dB@6.0 GHz 3.4 GHz P1dB = +12dBm (+17dB m) (+15dB m) (-2dBm) (-3d Bm ) (+21dBm) (+9dBm) (-8dBm) (+7dBm) L PA R I +9dB -5 db SMV3300A (+5d Bm) A D5601 6BITDAC TX IF GA IN CO NTROL -5dB (+3d Bm) BGA2031 Gain = +23dB@1.9 GHz (2.7V ) CT RL ΔG = 56 GHz P1dB =+13dBm@1.9 GHz IDET Jumper Select GHz ADF411 3 QDET AD BITDAC AD8349 I-Q MODULATOR 700 MHz GHz Σ RX IF G AIN CONTROL 0 90 BASEBAND I BW=30 MHz BASEBAND Q ADF GHz ADF GHz RX LO 1 FOX801BE-160 TCXO 16.0 MHz REF CLK OUT TX LO 2 ADF GHz ADF GHz BASEBAND I BW=30 MHz BASEBAND Q 5 GHz RF PCB Block Diagram D. DePardo 27 JULY June 2010 ResiliNets Overview 28

29 Transportation Security SensorNet [Victor Frost PI] Objective and problem KC SmartPort is encouraging development transport systems require visibility, accountability, efficiency, security Transportation security approach sensing, communications, and information integration integrate sensor information and real-time tracking with trade data documents to correlate expand the ORNL SensorNet technologies to mobile rail network environment 10 June 2010 ResiliNets Overview 29

30 End Questions? 10 June 2010 ResiliNets Overview 30