Advanced Multidisciplinary System Engineering or How I learned to think outside of MY box!

Transcription

1 Advanced Multidisciplinary System Engineering or How I learned to think outside of MY box! Dr. Joseph R. Guerci Director DARPA/SPO Joe.Guerci@DARPA.MIL All material cleared for Public Release 1

2 Outline Breakthrough systems/technologies are almost always multidisciplinary Arise from cross-fertilization Cross-fertilization occurs in someone s mind Thinking outside the box = Thinking outside your box Examples: KASSPER HISS New Trend in Multidisciplinary Systems Engineering Level 1: System = Interconnected set of single-purpose subsystems Level 2: System = Interconnected set of multi-purpose subsystems Level 3: System = Embedded multi-purpose subsystems w/o clear boundaries Example: ISIS Summary Sample SPO Projects (A Multidisciplinary Systems Technology Office) Revolutionary Space and Near-Space Technologies Advanced Intelligent Signal Processing & Embedded Systems Next Generation Chem/Bio Sensors & Protection PRODUCT Nicking Enzyme IDA Pathogen DNA DNA Polymerase PRODUCT IRSG RNA Polymerase Pathogen RNA IAR PRODUCT Toxin RNA Polymerase 2

3 md Adaptive Signal Processing Example: Space-Time Adaptive Processing (STAP) Space-Time Adaptive Beamformer Ideal Adapted Pattern Optimum Solution Weiner-Hopf (Optimum space-time beamformer weights) w = R 1s (Desired signal steering vector ) (Inverse of total interference covariance matrix) w, s C NM R C NM NM NM ~ ' s ' s 3

4 Covariance Estimation Problem Practical implementation example and real data example (White Sands DARPA Mountain Top Radar) Sample Covariance Estimation Ideal (Stationary) Data Measured Data Rˆ = xi x i i Ω Welcome to the Real-World! Extremely suboptimal radar performance can occur if one or more of the following occurs: (High false alarm rates and/or low Pd) Heterogeneous Clutter Rapidly varying terrain Mountainous (rapid elevation/reflectivity variation) Rapid land cover variations (e.g., littoral) Dense Target Backgrounds Moving Clutter Military/civilian vehicles Large Discretes and Spiky Clutter Urban clutter Power lines, towers, steep mountainous terrain Range-Varying (Nonstationary) Clutter Loci Bi/Multistatics Nonlinear array geometries (e.g., circular arrays) One or More of the Above is Almost Always Present in Real-World Ops! 4

5 Serious Performance Impacts!! (KASSPER 2 Data Cube & APTI Data Set) SINR Loss High False Alarm Rates GMTI Range Doppler Data (db thermal) 6 5 Range Bin # 2 o SINR/SNR (db) rang bin 24 (38.6km) 3 4 optimal PCI 4 MWF 4 post Doppler (3 bin) Doppler (m/s) Doppler (Fraction PRF).3.4 AMF Exceedance STAP Only STAP w/ Pre Whitening Rx 1 5 Fraction Exceeding Value y distance (km) > (latitude) (35.73,118.5 ) x distance (km) > (longitude) 15 Pixel SINR (db) Knowledge-Aided Sensor Signal Processing & Expert Reasoning (KASSPER) Radar Environmental Knowledge Bases (DTED/DFAD/LCLU, SAR, etc.) Measured Bald Earth Predicted (DTED Level-1) Range (DARPA Mtn Top) Doppler KASSPER Clutter Knowledge Base {γ i } RKA = Clutter Steering Vectors {Ci } Sensor Characteristics 2 2 γ i Ci Ci + σ I -.. x3 x M. ] T Physical Real-Time Database EM Modeling Tools R M Pulses T Array Snapshots x2 2 2nd Stage Conventional Filter KA... T. 2 X. [ x1 -.5 Nonstationary Clutter (plus Signal) KA Pre-Filter N Elements M Pulses T Sum Over Clutter Cells 1st Stage Knowledge-Aided Pre-Filter Response Physical Databases GPS/INS Clutter Cell Returns Reduced-Rank Conventional Filter Y = R KA2 X -6 1 Rˆ SMI Z = Rˆ SMI2 Y -6.5 HPEC Detector

6 Space-Time Advanced and CACFAR IF Sidelobe Fully Adaptive Adaptive (STAP) Real-Time AGC, etc. Canceler Array Radar STAP 5 s 6 s 7 s 8 s 9 s First Gen Savant High-speed, single function Intelligent Adaptive Radars Statistical Signal Processing Knowledge Multi-function, slow access speeds Real-world nonstationarity does NOT support conventional adaptivity Real-time knowledge-aided Radar returns Old Classic Reinventing Adaptive Radar s s KASSPER FLOPS/Throughput + Data type/mbytes New KASSPER True Intelligent Processing SAR Roads VMAP Discrete Conventional vs. KASSPER HPEC Processing Space-Time Snapshot Vector Test Cell Guard Cells Range Cells Conventional Space-Time Filtering. x x x x i+ 2 i+ 1 i x. i 1 i 2 Ω Rˆ = x x Ω i Ω k k Highly Parallel Systolic Array Implementation (Achieves s to s of GFLOPS) KASSPER HPEC Challenge: Optimizing adaptation by injecting environmental knowledge intelligently into the front-end signal flow First Gen Real-Time KASSPER HPEC Intelligent Signal Processing w = R 1 s Clutter Knowledge Base QR Factorization w/ Back substitution (from Antenna-Based Signal Processing Techniques for Radar, A. Farina, Artech House) KASSPER KASSPER requires requires memory memory access access interrupts interrupts Optimal Optimal interrupt interrupt scheduling scheduling Optimized Optimized ISP ISP Look-Ahead scheduling Look-Ahead scheduling 6

7 Capacity Access Time CPU Registers s Bytes <1s ns Cache s-s K Bytes 1- ns Main Memory M Bytes -3 ns Disk s G Bytes ms Tape Infinite sec-min Look-Ahead Scheduling Addresses Memory Latency Issues Problem: Registers Cache Memory Disk Tape Instr. Operands Blocks Pages Files Staging Transfer Unit Prog./Compiler 1-8 Bytes Cache Controller Bytes OS 512-4K Bytes User/Operator MBytes Source: Dave Patterson, Graduate Computer Architecture Course, University of California, Berkeley, Spring, 21 Faster Speed Higher Cost Larger Size Lower Cost t + t Solution: KASSPER Look-Ahead Interrupt Scheduling Predictor Clutter Knowledge Base t Next-Gen KASSPER HPEC Testbed Architecture: Base computer and I/O cards purchase order completed Lab computer configuration complete Various processing concepts in review PDR planned for late June 3 Demonstration at DARPATech 4 Parallel Vector Library (PVL) chosen for open standards programming language LL reviewing initial KASSPER algorithms for library impacts Coding started on basic radar signal processing components (pulse compression, data retrieval, etc.) Algorithm developers will program the hardware MP-5 mercury processing Open standards for realtime processing Application Code Vendor Software Vendor Hardware Multiple highspeed RAID drives ASIC high-speed cache memory devices Application Code Portable Library Maps Vendor Software Vendor Hardware Upgrades restricted to hardware remapping & new features 7

8 Pre-filtering Followed by Conventional STAP Adaptive Matched Filter GMTI AMF Output (db thermal) 3 6 Pre-Filtering Reduces The Tail of the Exceedance Function Range Bin # AMF Exceedance STAP Only STAP w/ Pre Whitening 2 1 Range Bin # Doppler (Fraction PRF) Without Prefiltering GMTI AMF Output after Whitening (db thermal) Doppler (Fraction PRF) With Prefiltering Fraction Exceeding Value Better Behaved Tail 13 db! Pixel SINR (db) KASSPER: It s an Architecture, NOT an Algorithm KASSPER is is an an architecture for for real-time adaptation of of multidimensional sensor systems in in real-world environments KASSPER Architecture Environmental context context is is key key to to efficient adaptation Sensors, Sensors, like like humans, humans, benefit benefit from from context! context! Key Key enablers: look-ahead scheduling and and resource allocation Multiresolution philosophy: blurring blurring the the boundaries between SAR SAR and and GMTI GMTI KASSPER as as a modern manifestation of of the the Bayesian method! KA-STAP KA-STAP Bayesian Bayesian STAP STAP The The DARPA KASSPER Challenge: Creatively explore the the possibilities Re-examine entire entire adaptive signal signal processing paradigm with with an an eye eye towards maximizing knowledge-aided robust methods Robust Robust STAP STAP algorithms AND AND KASSPER architecture Environmental Environmental knowledge knowledge base base as as read/write read/write scratch scratch memory memory What What is is implementable? 2? 2? 22? 22? Environmentally aware aware sensors have have a future! future! 8

9 Emerging Field Special Issue of IEEE Signal Processing Magazine Handheld Isothermal Silver Standard Sensor (HISSS) The goal of the HISSS program is to develop a handheld sensor that is capable of identifying biological threats including bacteria, viruses and toxins. How to shrink into a handheld? Order-of-mag faster! At least as accurate! DNA detection RNA detection RNA readout Fluid handling Protein detection Protein readout Polymerase Chain Reaction (PCR) Machine Notional Sensor DNA readout Notional Sample Cartridge System check 9

10 PCR vs. Isothermal Polymerase Chain Reaction Starting the process: Polymerase Primers Pathogen DNA Trigger template Polymerase Nicking enzyme Anneal at 55ºC Denature at 95ºC Isothermal Starting the process: Pathogen DNA Cleave Extend t ~ 6 sec t ~ 3 sec Extend at 72ºC Product falls off Polymerase re-binds Products: copies of Pathogen DNA Products: copies of reporter HISS DNA Amplification

11 HISSS Progress Progress: Demonstrated false alarm rates, using ROC curve analysis for HISSS assays that are equal to or better than current DNA, RNA, and protein assays Successfully developed and utilized a flow-through testbed to test all assays DNA ROC Curves RNA ROC Curves 1 1. Protein Toxin ROC Curves PI PIStatic (1:99)(1:99) PII (1:99) PIIFlow (1:99) PCR PCR(1:99) (1:99).6.6 Pd Pd PI (1:82) Static (1:82) PI PII (1:82) Flow (1:82) PII RT-PCR (1:82).8.8 RT-PCR (1:82) Pd Pfa.6.6 PI (1:3) PIStatic (1:3) PII Flow (1:3) PII (1:3) ELISA (1:3) ELISA (1:3) Pfa Pfa New Airship Design Philosophy Capability cannot be added to airship after development ISIS requires integration of sensor and airship MDA Airship Conventional Airship Payload bay Payload: ~2% of system mass Payload: 3-4% of system mass Turn a disadvantage (large size) into an advantage (large antenna)! 11

12 The First ISIS? Echo 1 Most Powerful Airborne GMTI/AMTI Radar & Comms Ever Conceived Simultaneous AMTI/GMTI Operation via Dual Band (UHF/X-Band) Aperture Cruise Missile Defense Long-range AMTI/GMTI/COMM FOPEN GMTI Detect/Track Dismounts Steep Grazing Angles Extremely High Capacity Comms Near Zero Platform Speed No In-Theater Ground Support 99% on station availability for 1+ years 6km radar horizon at 7kft operational altitude 12

13 Unprecedented Radar Performance Global Hawk Platform Carries the Antenna 1. Relative Search Capability (PA/R2) 1. Relative Track Capability (PA2/λ2/R4) Joint STARS X X AWACS S S ISAT X Antenna Is the Platform X 24 5, 3,3 15, , VHF 3,, X ISIS Sustained Operations Logistics Aircraft-based ISR Requires Local air base Multiple aircraft to keep 1 flying Air crews Ground crews Fuel supplies Maintenance facilities ISIS Unmanned Deploys worldwide from U.S. base Regenerative Fuel Sources One-year continuous ISR capability 13

14 Wind Conditions Drive Propulsion Power Needs Where ρ = air density at altitude V = volume of airship v = relative velocity of air η = efficiency of propellers ρ Cd V 2 / 3 v 3 P= 2η Propulsion Propulsion Power Power for for V V == m m (C (Cdd=.22) =.22) Power Required Required Power (kw) (kw) m/s 2,5 2,5 2, 2, 1,5 1,5 1, 1, Max winds drive power system requirements Wind Wind Speed Speed (m/sec) (m/sec) Station Keeping ISIS Objective: 99% on-station availability for 1 year Function of airship speed (sustained and sprint) and available energy (regenerative and stored fuel) Need operational algorithms for maximizing availability Maximum Sprint Speed Latitude (degress) Managing airship energy ala satellite delta-v Wind Speed (m/s) Mean Wind Speed Average 99 Percentile 14

15 M P Power Mass 2/3 P V Volume Requires Large Mass Reductions ISIS designs are mass-centric V M Lifting gas has reached the maximum limit:.61kg per 1m 3 of 21km.66kg per 1m 3 of H 21km ISIS focusing on: Removing mass from largest contributors Integration, INTEGRATION, INTEGRATION! M displacedair = M liftinggas + M structure Integration + M radar + M power + M propulsion + M avionics M ISIS = ρ V + c ρ V gas h hull 2/3 + ρ Components aperture ρ A+ η power power P radar ρairc dv + 2η 2/3 propulsion 3 v + M propulsion + M avionics Summary Breakthrough systems/technologies are almost always multidisciplinary System engineers need to be continually learning about new technologies and methods across ALL disciplines Be an annoying know-it-all! Tactic: Can the thermal engineer give the flight control engineer s briefing? Often cross fertilization can occur even if with only a 1 st or 2 nd order understanding of multiple disciplines Balance of depth and breadth How should engineering programs be structured in light of above? Undergraduate programs typically have the breadth, but don t seem to close the deal Example: Senior class semester devoted to dissecting a complex system Emergence of a Level 3 systems integration Multidisciplinary from its inception! 15