Intelligent Content- Aware Data Priori2za2on and Synchroniza2on across Disconnected, Intermi<ent, Limited (DIL) Networks

Transcription

1 Intelligent Content- Aware Data Priori2za2on and Synchroniza2on across Disconnected, Intermi<ent, Limited (DIL) Networks Approaches and Considera2ons Ma< Fisher Nik Keapproth Progeny Systems March 29, 2012 Part No: P006683, Rev: -

2 Typical Problem Areas for Afloat C 4 I Dynamic and Ad- Hoc inter- plaxorm informa2on exchanges over Disconnected, Intermi<ent, and Limited (DIL) network (SIPRnet) Evolving topology and mission è Highly variable bandwidth demands Data genera2on can exceed available bandwidth è Cannot guarantee adequate resources for all generated data to be exchanged Unreliable connec2ons è No implicit assump@on that shared data will be fully replicated as intended Informa2on objects have varying degrees of relevance to the user at different 2mesè Systems should act to focus communica@ons resources where they are most needed Decision- makers rely upon the quality and completeness of shared tac2cal informa2on è Systems should provide assurance of the integrity of the shared tac@cal informa@on Unmanaged Data Poses Risk to Mission Success 2

3 Undersea Warfare Decision Support System The Common Tactical Picture (CTP) is defined as providing situational awareness with such timeliness and accuracy that would facilitate an overt action on the part of a commander up to and including engagement with weapons. References: ASW Command and Control (C2 Study) Naval Warfare publica2on (NWP &NTTP) 32-1 Fleet Exercise Results (VS- 06, VS- 07, USWEX) Submarine combat Systems APB- 06 Capabili2es Le<er An2- Submarine Warfare (ASW) Ini2al Capabili2es Document (ICD) Global An2- Submarine Warfare (ASW ) Concept of Opera2ons (CONOPS) OPNAV N6 Moderniza2on Requirements Le<er for AN/UYQ- 100 USW- DSS CAN Study Improving Fleet ASW: Common Opera2ng Picture and Tac2cal Decision Aids, Nov Operational Need Provide an ASW Common Tactical Picture Across the ASW Enterprise which: Increases Situational Awareness Reduces Operator Workload Enables Improved Mission Planning, Search Execution, ASW Track Management and ASW Battle Management

4 ASW C 4 I Game Changer Time Late Labor Intensive Inaccurate Joint Undersea Superiority Study An2 Submarine Warfare Ini2al Capability Document (C4 for ASW) Metrics Based Planning and Execu2on Improved Sensor Search Effects Real Time CSG Synchroniza2on TDZ and LLOA around CVN SK centered on CVN ASW Master Tac2cal Plot IAW NTTP 3.21 Time Late Lacks Environmental Effects Lacks ASW Sensor Inputs and Fusions GCCS- M Carrier Strike Group Display Current Opera2ons ASW C 4 I Transforma2on CNO Vision for ASW Superiority, May Ac2ve and Passive SPP overlays using in- situ measured noise show significant differences in this dynamic environment Carrier Strike Group Screen Kilo Quick visual SA showing Threat INSIDE of CVN LLOA but OUTSIDE TDZ Automated ASW Sensor Data Exchange ASW Sensor Contact Correla2on and Fusion Shared Tac2cal Decision Aids on USW Common Tac2cal Picture USW DSS Improvements

5 USW- DSS Data Exchanges Data exchanged between DDGs, CGs, CVNs, IUSS*, and ashore command sites: Track / Contact Data Tac2cal Decision Aids (TDAs) Sensor Performance Data (measured noise and reverb) Search Plans Tac2cal Status and Orders Administra2ve Data (Force Management and Versioning) Comms via SIPRnet today; other paths under evalua2on for future systems * IUSS Informa2on exchanges are a subset of those listed 5 here

6 USW- DSS Current Approach JCIDS- level requirements for USW- DSS B2 to establish and maintain a common USW tac8cal picture between connected plaxorms Topology is managed hierarchically and implemented automa2cally to provide consistency Data is exchanged peer- to- peer to maximize reliability and minimize cri2cal nodes USW- DSS B2 employs priori8za8on of data by type to cope with DIL network When bandwidth limits are reached, priority is given en masse to certain data types over others Some data exchanges use periodic synchroniza8on mechanisms Reconcilia2on s2ll dependent on data- type priority hierarchy 6

7 Selected Pla>orm Details Roles and Status of Composi2on Membership Geographic Extent of Shared Data Filters out extraneous geospa2al data 7

8 Priori8za8on by Topic High/Med/Low Priority and BW Caps for each Bandwidth Cap Total BW limit BW Plots over Time Color Coded by Type Instantaneous Values Pie charts and values 8

9 Data Exchange Challenges - Causes Applica2ons contribute to the overall bandwidth, but: Applica2ons may be unaware of the total available bandwidth for the system Applica2ons may be unaware of the bandwidth incurred by other applica2ons Applica2ons may be unaware of the bandwidth incurred by their own data Total system bandwidth can vary based on comms performance and other systems BW limits/schemes that do not account for all factors will subop2mize the system Different data components from any one source are generally not uniformly important to the human user, and can vary in importance based on: Mission priori2es Absolute rela2onships (loca2on, threat category, 2me, speed, etc) Rela2ve rela2onships / comparisons with other data objects (range, LLOA, etc) What is present on a peer (acceptable latency or thresholds for differences in shared informa2on) Is Every Update of Every Object in One Type Always More Important than Any Update of Any Object in Another Type? 9

10 Data Exchange Challenges - Effects Operator can priori2ze between data types, but can only affect incurred bandwidth by removing shared content or subtrac2ng plaxorms (Orig_rate * num_plaxorms) > bw_avail è Conges2on Under constrained bandwidth, some data will not be exchanged è Reduced Effec2veness Data priority is applied in rigid, over- broad groupings è Stra2fied Simple, Sta2c Priori2za2on Schemes Limit Poten2al 10

11 USW- DSS Data Priori2za2on Research Synchroniza8on State (S S or Data Validity) can be calculated for each remote plaxorm Determined as a func@on of % accuracy, latency characteris@cs, and mission relevance Manage communica@ons to maximize S S Each data object should be able to be priori2zed and managed independently by the system Data objects should be able to co- mingle in priori2es between data types to maximize S S Measure Sync Quality and Maximize with an Objec2ve Func2on 11

12 Key Features of an Improved Approach Affect / control the rate of data produc2on Prevent satura2on Improve system ability to respond to network condi2ons Evaluate data objects individually Quan2fy opera2onal significance at the object level Prevent stra2fica2on by type when BW is constrained Provide synchroniza2on metrics Give users an understanding of (and confidence in) the integrity of the shared tac2cal picture Metrics should be intui2ve, easily comprehensible 12

13 USW- DSS Candidate Approach Centrally manage all cross- plaxorm informa2on exchanges Enables intermingling of data types in priority through a coordinated knowledge of bandwidth availability and need Considera2on: avoid architectural constraints requiring bilateral code changes wherever possible Dynamically priori2ze traffic to maximize an objec2ve func2on Quan2fy the state or quality of the shared situa2onal awareness Manage transmissions (data object + remote endpoint pairs) individually Focus on those transmissions that most contribute to the objec2ve func2on 13

14 USW- DSS Candidate Approach Assign each data object a discrete value that relates each shared object to a mission cri2cality (across all data types) è Tier Value Dynamically update as parameters change Tier profile to be customized by Warfare Commander and disseminated by the system For each locally produced {data object, des@na@on planorm} pair, track the remote state of that object using 2mestamp è State Matrix For each entry in the State Matrix, calculate a (con2nuous) Urgency è Urgency Matrix a func2on of 2er, 2me- lateness, displacement error, categoriza2on error, other criteria Rank- order the Urgency Matrix by highest priority è Transmission Priority Vector (TPV) Remove transmissions for known unreachable units Select highest priority bytes from the TPV to each des2na2on plaxorm Not to exceed outgoing capacity of the source, incoming capacity of des2na2ons BW capacity based on last packet(s), with configurable default values and maxima Provide a parallel periodic synchroniza2on using similar object- level priori2es Under Development using USW- DSS Build 2 as Demo System 14

15 Prototype Objec2ves Demonstrate that a given Synchroniza2on State can be obtained with reduced bandwidth Demonstrate that for a fixed bandwidth, an improved Synchroniza2on State can be obtained Demonstrate that improved operator control over data exchange priori2es is possible Demonstrate that improved operator awareness over the state of shared tac2cal informa2on is possible 15

16 Tradi2onal Data Priori2za2on Data Type 1 Δ Shared Data Set Set Cap & Priority Dest n A Data Type 2 Data Type N Δ Shared Data Set Set Cap & Priority Δ Shared Set Cap & Data Set Priority Available Real- Time Human Ac2ons Output Queue Dest n B Dest n N Effec2ve But Coarse- Grained and Reliant on Human Interven2on to Modify Sezngs when BW is Constrained 16

17 Classic Feedback Loop Reference Input + - Controller System System Output Measurement Enables System Response Faster Than A Human Could Respond 17

18 Intelligent Data Priori2za2on Δ Tier Profile Available Real- Time Commander Ac2ons Data Type 1 Δ Shared Data Set Calculate Urgency Dest n A Data Type 2 Data Type N Δ Shared Data Set Δ Shared Data Set Assign Tier By {object, dest} pairs Consider latency, error, etc Output Queue Dest n B Dest n N Available Real- Time User Ac2ons Es2mate Remote State Real- Time System Ac2ons Transmission Status Sync Query Response Feedback Loop and {Object, Des@na@on}- Level Priori2za2on Enable More Granular, Automa2c, Objec2ve- Oriented Data Priori2za2on 18

19 Conclusion Systems with no data priori2za2on capability risk mission accomplishment or require significant operator interven2on Systems with coarse- grained data priori2za2on approaches provide a certain measure of control and ensure the flow of certain types of data under adverse network condi2ons An approach which emphasizes synchroniza2on goals at a system level and adapts to changes in data content, mission relevance, and network condi2ons can exceed the performance of today s genera2on of C4I systems 19

20 Ques2ons 20