`a
`3-D lattice formulation is
`The
`of
`the
`2-D
`Burg
`algorithm and,
`extension
`analogously, a'set of 3-D reflection coefficients
`{p(n,m,u)}can be derived. Again,
`the reflection
`coefficients
`are
`calculated directl
`from the
`'n ut data sam les.
`This nethod is basically a
`%_D AR model fitting algorithm.
`
`3-D Burg
`algorithm is essentially a
`This
`algorithm which given an optimal MMSE estimate for
`stationary processes when the image extent is much
`greater
`than
`the
`extent
`of
`the
`filter
`H
`(z ,z ,2 ).
`Signal
`flow diagrams
`for
`the
`3UBMFlR an
`IFR prediction filters are similar to
`2-D
`lattice
`filters with
`additional
`delay
`elements.
`
`0, Results
`
`The most significant results to arising out
`these
`investigations
`can
`be
`summarized
`as
`of
`follows:
`1) Reflection Coefficients - Calculated by
`the harmonic mean method are extremely good means
`of characterizing 2-D imagery data
`in a MMSE
`sense.
`As
`demonstrated experimentally herein,
`very low residual errors remain after only a few
`stages of lattice filtering.
`the 2-D
`2)
`Ex erimentall Verified — That
`algorithm as
`implemented
`here
`with
`Burg
`appropriate
`support
`can
`be
`used
`for
`imagery
`compression
`purposes
`in
`the
`spatial
`domain.
`Heretofore,
`no
`2-D LPC imagery coding of
`this
`nature
`had
`been
`investigated.
`This
`paper
`described
`a
`valuable
`new spatial
`compression
`technique.
`It
`is
`suitable
`for
`real-time
`applications yielding impressive
`5 -9
`percent
`image compression factors with,
`in most cases,
`imperceptible image fidelity deterioration. when
`2-D LPC-induced distortion occurs,
`it
`can
`be
`removed by simple image enhancement type filtering
`techniques.
`3)
`‘A New 2-D In-Place Method - For real-time
`computation of the 2-D lattice filter coefficients
`and_ subsequent
`imagery data filtering has been
`derived and verified experimentally with actual
`imagery data.
`This method provides intermediate
`results at _any filter stage and is amenable to
`multidimensional MMSE digital filtering in excess
`of 2-D.
`This
`new method of
`in-place lattice
`forward and _reverse filtering was utilized for
`display of
`intermediate imagery results display
`after each’ stage of
`the 2-D LPC process. with
`this algorithm, it is possible to form ‘imagery’
`composed from only selected stages of residuals.
`In
`other ‘words,
`imagery
`from which
`selected
`nearest-neighbor
`pixel
`correlations
`have
`been
`removed can be generated. This methodology is, in
`effect. capable of selective decoupling,
`in a MMSE
`Sense, multivariate
`function
`variables.
`This
`ehavior
`is
`afforded by
`to
`the Gram-Schmidt
`orthogonalization realized by the lattice filter
`structure itself.
`4) Extension of 2-D LPC Technigues - To 3-D
`demonstrating that wit
`t e appropriate ordering
`Qf multidimehsional digital
`signal points
`into
`P35?
`and
`future‘,
`the resulting LPC filters
`retain _
`the
`same
`stability
`and
`analytical
`groperties attributed to 2-D and experimentally
`dgmonstrated herein.
`These
`2-D lattice filter
`a Slgns can readily be extended to N dimensions,
`5 well
`as
`to n stages in each dimension.
`The
`Pggdiction error results for each dimension can be
`ependently obtained and the final prediction
`
`root—sum-square composite.
`a
`value calculated as
`This methodology can be utilized effectively to
`realize a multivariate
`function analysis MMSE
`prediction capability.
`
`-
`
`include
`To
`forward
`and
`
`As a result of these investigations, several
`areas remain prime candidates for further work.
`These areas include:
`Stabilit
`1)
`2-D Filter
`for
`additional
`algorithms
`both
`backwgld pMgS1C§n8n'Automated
`2-D Filter Desi n
`Methods
`-
`Investigations of
`the performance of
`reflection coefficient determination algorithms in
`addition to the 2-D Burg algorithm harmonic-mean
`method.
`3) Determination of a Quantitative Fidelity
`Criterion
`- More
`representative
`of
`imagery
`utility.
`4) Development of Precise Methods - Imagery
`testing, evaluation, and performance verification.
`5) Utilization of Evolvin
`State-of-the—Art
`Hardware
`- Laser disks, VHSIC/VLSI
`integration,
`memories, CPU interfaces, etc.
`6) Practical Extension of these LPC Methods-
`correlation, experimenta
`behavior,
`and adaptive
`prediction in a MMSE sense. This 2-D formalism is
`also
`suitable
`for
`analysis
`of other
`2-D or
`multi-dimensional stochastic processes.
`7)
`Com arison with
`the Discrete Cosine
`Transform
`DCT
`-
`2-D
`ata
`compression
`app ications via a suitable quantitative measure.
`The
`above
`candidate
`research
`areas
`all
`constitute
`promising topics for
`future work.
`The
`full
`benefit
`of
`MMSE
`LPC
`analysis
`is
`applicable to other than just the multidimensional
`digital data compression problem.
`References
`
`‘A Study of
`1. Harlick, R. and Klein, R.
`Adaptive
`Image
`Compression
`Techniques,‘
`University of Kansas, AD-A094678.
`2. Marzetta, T. L.
`‘A Linear Prediction Approach
`to Two-Dimensional Spectral Factorization and
`Spectral Estimation,‘ Ph.D. Dissertation
`Massachusetts Institute of Technology. 1978.
`
`3.
`
`4.
`
`5.
`
`‘Data Compression of Imagery
`Poehler, P.L.
`Using Linear Predictive Coding Techniques,‘
`Ph.D. Dissertation, Florida Institute of
`Technology.
`l983.
`
`‘Lattice
`Parker, S. R. and Kayran, A. H.
`Parameter Autoregressive Modeling of Two-
`Dimensional Fields,‘ submitted to the IEEE
`Transactions on Acoustics, Speech, and Signal
`Processing, May,
`l983.
`
`Chen, T. C. and Defigueiredo, R. J. P.
`Image Transform Coding Scheme Based on
`IEEE
`Spatial
`Domain
`Consideration,‘
`Transactions on Pattern Analysis and Machine
`Intelligence, Vol. PAM I-5, No.3, May
`1983.
`
`‘An
`
`‘Stable and Efficient Methods for
`6. Makhoul, J.
`Linear Prediction,‘
`IEEE Transactions on
`Acoustics, Speech, and Signal Processing,
`October, 1977.
`
`7.
`
`‘Quantizing for Minimum Distortion,‘
`Max, J.
`IRE Transactions on Information Theory, Vol.
`IT-6, pp 7-12, March 1960.
`
`555
`
`BOHNG
`
`Ex.1031,p.626
`
`BOEING
`Ex. 1031, p. 626
`
`
`
`WORST CASE
`BEST CASE
`
`
`
`‘I
`
`2
`
`‘ID
`9
`B
`7
`E
`5
`A
`3
`NUMBER OF LATTICE FILTER STAGES
`
`11
`
`12 '"7"‘
`
`-5
`.5
`
`.4
`
`.3
`-2
`.1
`
`8.
`
`‘Reduction of
`Reeve, H. C. and Lim, J. 5.
`Image
`Coding,‘
`Blocking
`Effects
`in
`Proceedings of the IEEE Internationai
`Conference on Acoustics, S eech, and Si nai
`
`Processing, April 1983.
`
`IEFLEEYIDI
`cotrncmns
`DVIIKEIII
`I nu‘!!! stmunr
`Smrolu DAM
` ~D FIR
`
`UIIUIIIL
`LAYTICE
`
`
`EOMPRESSION
`mm: min
`mien
`
`I:
`
`O1ccW
`
`E
`—
`*5
`0HJ
`_
`E
`ca
`
`I Eg
`
`3
`
`2
`
`.)
`
`Forward I
`
`resslun Process
`
`nvucnun calrrmnln
`a nun SUPPURY
`
`LATYICE
`L0 M
`EXPANSION
`FILTEFI
`
`'"l'"'W“
`
`mu“ D‘Y‘
`
`
`
`
`
`unvenst
`IVAIIIIZATWI
`
`
`
`IIYIOVV
`DKCDDIIC
`
`
`h) Normalized Prediction Error vs. Number of Lattice Filter
`Figure 1
`Stages
`
`Figure 3
`
`~,i
`
`
`
`5) Direct Realization of FIR Least Squares Error Filter
`(
`= Filter Coefficients)
`N
`
`Em
`
`“N
`
`3.0
`
`2-O
`
`N‘
`
`Order
`
`Subimage
`mock
`Entropy
`(bits/pi'xe1)
`
`H“ --1/N1‘: Mai) 1092 H31.)
`
`N--oc
`Hm H
`
`n
`
`x H
`
`Reaiizeq uigh 2-D LPC Coding, Max 4.3a;
`\ Quantization and Entropy Coding
`\ ~ “ \ _
`\
`‘
`_
`‘ ~ ~ __ \
`I
`.5 biz:/pixei
`Lower Bound
`‘
`
`
`
`D) Lfitticfi Iflvlementation of s FIR Least Squares Error
`Filter (--k1 = Reflection Coeffxcxencs)
`“SW2 3
`
`Image Compression Efficiency Achieved
`
`Figure t
`
`556
`
`BOHNG
`
`Ex.1031,p.627
`
`BOEING
`Ex. 1031, p. 627
`
`
`
`A TERRAIN DATA/DIGITAL MAP SYSTEM FOR LHX
`
`G. 0. Burnham, C. Benning, R. Rivard
`
`Texas Instruments Incorporated
`P.0. Box 405, M/S 3407
`75067
`Lewisville, Texas
`
`84-2723
`
`Abstract
`
`is a new aerial platform
`The Light Helicopter (LHX)
`designed to perform vertical lift missions for Army
`aviation in the air/land battle of 1995 and beyond.
`This paper briefly describes the avionics functions
`required to support LHX missions,
`including mission
`management and digital map data retrieval.
`Pres-
`ented are some preliminary ideas
`on terrain data
`system functions and a discussion of how implement-
`ing these functions in the LHX integrated architec-
`ture would impact the overall avionics suite.
`
`Overview
`
`The Air/Land Battle 2000 concept employs Army avia-
`tion in whole or in part
`in all mission areas and
`battlefield tasks.
`Army aviation must
`rapidly de-
`ploy and engage
`enenw forces; conduct
`reconnais-
`sance and airmobile and logistic operations;
`and
`provide C31, airborne fire direction,
`intelligence
`and electronic warfare operations, and battlefield
`interdiction.
`The
`threat
`is real, both quantita-
`tively and qualitatively.
`In facing this projected
`enemy threat the Army Aviation Mission Area (AAMA)
`analysis identified many deficiencies in their cur-
`rent aircraft fleet,
`including aging and obsoles-
`cent aircraft
`that cannot be upgraded satisfacto-
`rily to meet
`tactical challenges.
`The Light Heli-
`copter (LHX)
`family
`is
`envisioned
`as
`the Army
`aircraft that will
`overcome this.
`It
`is
`a
`new
`aerial platform designed to perform vertical
`lift
`missions for Army aviation in the air/land battle
`of 1995 and beyond. Depending on mission require-
`ments and cost-effectiveness analyses,
`it may or
`may not be
`a nmltiple—version aircraft.
`If more
`than one version (e.g.,
`scout attack and utility/
`observation) evolves,
`all
`versions will
`employ
`many common
`subsystems
`such
`as
`engines,
`rotors,
`drive trains, and core avionics.
`
`Avionic processing equipment has been traditionally
`designed for specific subsystems
`in which it was
`intended to operate. This approach has led to dif-
`ferent architectures, hardware, and software sup-
`port within the same system. Providing a low-cost,
`sustainable, mission-effective avionics system re-
`quires fault—tolerant, modular processing
`archi-
`tecture and processing modules. Essential
`in this
`development
`is
`the timely and cost—effective in-
`sertion of
`very-high-speed
`integrated
`circuit
`(VHSIC) technology.
`with
`full-scale engineering
`development of
`LHX scheduled for FY86, it is im-
`perative to initiate development of
`a VHSIC-based
`avionics architecture to ensure feasibility of an
`avionics system capable of meeting the processing
`requirements dictated by
`the missions
`and
`func-
`tions of
`LHX and to avoid early obsolescence or
`costly retrofitting.
`
`One of the functions to be implemented into the LHX
`avionics is a terrain data system capable of work-
`ing in conjunction with the navigation system to
`provide
`accurate
`terrain correlation navigation
`updates as well
`as various displays critical
`to
`low-level NOE flights.
`
`LHX System
`
`required to
`functions
`lists the avionics
`Table 1
`support LHX missions.
`The LHX processing architec-
`ture has
`been defined to provide the signal
`and
`data processing necessary to implement
`these func-
`tional subsystems.
`The processing which they re-
`quire, coupled with stringent physical constraints
`of an
`advanced rotorcraft, dictate an integrated
`architecture
`approach
`that
`crosses
`functional
`boundaries to
`permit
`the
`sharing of
`processing
`resources.
`The processing architecture implements
`the sensor fusion algorithms that combine the data
`of multiple subsystems to create a highly synergis-
`tic system;
`however,
`the control
`structure also
`
`Function
`
`Target acquisition
`Fire control
`Aircraft survivability
`equipment
`Terrain data
`Navigation
`Communication
`Mission management
`Flight control (3)
`
`Total
`
`TABLE 1.
`Preprocessing
`(MOPS)
`
`LHX PROCESSING REQUIREMENTS
`Array Processing
`Data Processing
`Program Memory
`(MOPS)
`(MIPS)
`(KWDS)
`
`Data Memory
`(MBITS)
`
`978
`—
`70
`
`—
`—
`—
`——
`—
`
`1.048
`
`125
`-—
`2
`
`343
`10
`138
`0
`O
`
`618
`
`20
`0.75
`5
`
`7.5
`9
`3.5
`24
`2.3
`
`72
`
`150
`139
`40
`
`120
`96
`l27
`I48
`48
`
`868
`
`33
`1.5
`0.2
`
`81
`4
`4
`27
`—
`
`I51
`
`(391?) right © American Institute of Aeronautics and
`Astronautics, lnc., 1984. All rights reserved.
`
`557
`
`BOHNG
`
`Ex.1031,p.628
`
`BOEING
`Ex. 1031, p. 628
`
`
`
`allows autonomous operation of the subsystems while
`maintaining deterministic
`processing time
`lines.
`
`required
`the
`of
`Table 1 also presents estimates
`program memory, data memory, and processing. Three
`types of processing are required:
`complex vector,
`real array,
`and scalar.
`Five processing modules
`have been defined: complex vector processing module,
`array processing module, data processing module,
`memory module, and mission computer executive mod-
`dule.
`
`LHX Mission Management Subsystem
`
`subsystem
`the nnssion management
`The objective of
`is to integrate and coordinate all activities with-
`in the overall system.
`These components, Figure 1,
`include a central manager, data base manager, dis-
`play manager,
`and subsystem manager for each sub-
`system.
`It should be understood that the functions
`of the
`central manager, data
`base manager,
`and
`display manager are distributed among the various
`subsystems.
`
`f'7fiEEF'1
`f'7REE"1
`CODWROL
`I
`|
`IIRGEY
`mv“A;;_,cI£,rm_ I ,,,,,,,,,,,,,
`unmask I -LGWIWS
`
`,_-_..__-.._..-_..-_....-..-_..
`
`CDIOUNICAIION
`‘”3fi¥” qlllullln kwww
`
`ewe IIIIIIIII wmww
`I
`I
`I
`I
`IIIIIIIII Mew»
`Iligailiillllllli .mam.
`I
`I
`I
`I
`IEIIAIII DAV
`VERNAIII DATA
`
`suRvIvAaII.IvI I_Il smv-mun_Il “‘°“"""’
`
`I——u—:c——-I———u-——¢——————u-——
`
`F_——_—-————.-—.._____-._.-
`
`.. ....E;3:2:?
`
`DIGIIIL NIP
`
`AIMMII
`s%‘{I‘7¥&'s‘#I
`
`mar
`
`(ENVRIL
`Nmarin
`
`svsmt
`was-enlsi
`
`5'99‘
`0... mg
`
`
`
`
`
`
`
`including any set_
`prioritizes and sequences them,
`up, follow-up, or subtasks.
`This
`is particulariy
`important when some of the tasks must be performed
`by the pilot.
`
`shared centra]
`The data base manager maintains a
`data base.
`The
`subsystems
`initiate requests
`for
`the storage or retrieval of certain data.
`The data
`base manager
`then
`transforms
`all
`references
`to
`actual memory addresses and performs the necessary
`data base functions.
`Since there will be various
`types of memory storage in the system (perhaps with
`different access
`times, word
`lengths, error cor-
`rection or detection capability,
`EMP protection,
`power backup, etc.),
`the data base manager decides
`before and during the mission where and how data
`shall be stored to optimally meet mission require-
`ments.
`It also performs any automatic or periodic
`data functions on its own.
`
`The display manager meets the changing information
`needs of
`the pilot.
`As different
`subsystems pro-
`duce information to be displayed,
`the display man-
`ager must decide the best way to make this inform-
`ation available to the pilot, taking into conside-
`ration the data's priority. Choices include merg-
`ing the new data with the data currently being
`displayed, overwriting part of the display, waiting
`until
`the pilot finishes with the current task be-
`fore interrupting,
`letting the pilot know that the
`data is available but will not be displayed until
`requested, or
`using
`an
`attention-getting device
`such as color, high-intensity or flashing graphics,
`audible tone, or spoken phrase. The display manager
`must keep track of what
`the pilot
`is currently
`viewing as well as what he has just viewed. Using
`this information and a description of the current
`situation,
`the display manager predicts what
`the
`pilot will want
`to see next,
`labels the program-
`mable keys appropriately, and starts computing the
`highest probability displays to minimize the pro-
`cessing time required once the pilot makes a choice.
`
`subsystem man-
`Certain functions are common to all
`agers.
`Each must keep the central manager informed
`of any changes in the subsystem's status, mode, or
`configuration.
`In addition to the shared data base
`
`managed by the data base nmnager, each subsystem
`will also have local memory
`storage for
`its own
`use.
`Each
`subsystem manager
`is
`responsible
`for
`controlling access
`to its local
`storage and any
`data transfer
`to or
`from the shared system data
`base or other subsystems.
`To a large extent, fault
`monitoring and inflight
`testing will be performed
`at the subsystem level
`by each subsystem manager.
`while the final decision on sensor assignments and
`reassignments is made by the central manager, each
`subsystem manager generates requests for the sensors
`it needs
`and controls the sensors
`it has already
`been assigned.
`In addition, certain data areas,
`processors, sensors, or equipment may be dedicated
`to a specific subsystem manager. what procedures
`to call and what parameters to call with them are
`decided at
`the subsystem level.
`If
`a procedure
`needs to communicate with another subsystem, with
`the central manager or pilot, or with the system
`data base,
`the subsystem manager is responsible for
`constructing an
`appropriate symbolic
`request
`and
`addressing it
`to the central manager, data base
`manager, display manager, or the desired subsystem
`manager.
`
`The target acquisition subsystem manager's
`tions also
`include
`determining
`the
`area
`
`func-
`to
`be
`
`Figure 1. Mission Management Functions
`
`The central manager manages the various high-level
`functions of the system, e.g., those that are com-
`mon to all
`subsystems or those that either involve
`more than one
`subsystem or
`take more than a
`few
`seconds to perform or require interaction with the
`pilot.
`The central manager must maintain models
`of the system's configuration, the overall mission,
`and the
`status of
`the various aircraft
`systems.
`These models primarily provide the pilot with
`a
`summary and
`tentative evaluation of
`the current
`situation.
`The central manager interacts with the
`pilot to refine its internal models
`and to plan
`appropriate activities (route planning, tactic se-
`lection, weapon handoff, etc.).
`It
`also inter-
`acts with the subsystem managers
`to optimize re-
`source allocation, execute pilot
`requests or plan-
`ned activities, and coordinate sensor assignments.
`At the same time,
`the central manager must monitor
`the
`status
`of
`the
`various
`subsystems,
`perform
`inflight testing/reconfiguration,
`log all mission
`activities, and handle routine situations.
`It
`is
`the central manager's duty to process pilot
`inputs
`(voice commands, keyboard entries, etc.) appropri-
`ately and
`to integrate the
`output of
`subsystem
`data (video displays
`and
`speech output).
`when
`doing this,
`the central manager tries to predict
`the pilot's current
`information needs and to pro-
`vide suitable default
`inputs. Often, when several
`tasks are
`to be performed,
`the
`central manager
`
`558
`
`BOHNG
`
`Ex.1031,p.629
`
`BOEING
`Ex. 1031, p. 629
`
`
`
`correlation navigation functions (e.g., SITAN) and
`to aid in TF/TA schemes, mission planning,
`and
`visual presentation of stored terrain information.
`An LHX TDS must provide digital
`imagery to allow
`the crew to quickly orient itself georgraphically,
`determine and display threat envelopes as a function
`of aircraft position,
`calculate and display the
`optimum flight path between two geographic points,
`automate mission planning,
`and allow precise pre-
`pointing of
`targeting sensors
`by
`supplying
`3-D
`relative position between aircraft and prestored or
`mission-acquired targets.
`Figure
`2 presents
`an
`example of functional
`flow.
`
`Terrain data processing includes three major sec-
`tions: data
`retrieval, processing algorithms, and
`display formatting.
`Data
`retrieval
`involves
`the
`control and interfacing of
`the bulk/cache memory
`section.
`The
`processing algorithms
`contain the
`functional algorithms
`associated with
`a
`terrain
`data system (e.g., SITAN or passive ranging). Dis-
`play formatting the data produces
`information in
`the correct
`form to be displayed.
`
`
`
`W‘
`
`‘
`
`nsxonuuoon
`can an:
`(us rt. um
`rcmnv
` uuunuv unm s
`mum on can! om.
`‘
`mm
`
` vanunc u
`Ima‘
`
`v
`
`a:A‘v"A'
`,.,u"
`
`Iuovamu
`must
`
`Figure 2. Terrain Data Functional Flow
`
`The referenced map data is stored in a compressed
`format
`in a nonvolatile bulk memory
`system that
`contains terrain features, elevations,
`threat
`in-
`formation, and
`cultural
`data.
`This
`information
`will come
`from the Defense Mapping Agency
`(DMA)
`and intelligence missions.
`The resolution of this
`data defines the accuracy of the system; therefore,
`system error is a determining factor in the select-
`ion of the data compression scheme.
`
`Data accessed from the bulk memory undergoes data
`reconstruction (the inverse of
`the data compres-
`sion algorithm) and is stored in small, fast-access
`time cache memories. These cache memories are large
`enough to allow for platform dynamics between re-
`freshes; i.e., the storage area of these memories
`is based on the flight dynamics of the helicopter.
`The memory scroll and control function scrolls data
`into the cache memories when needed and provides
`the overall
`control
`for
`the
`retrieval
`system.
`Scrolling is dependent on vehicle speed and atti-
`tude. Real-time updates of terrain, cultural, and
`threat data acquired during the mission are needed.
`This information may be
`received either offboard
`or onboard and is stored in an update memory that
`is addressed and scrolled into the cache memories
`in a
`fashion similar to that of the bulk system.
`
`High- and low-resolution data algorithms are per-
`formed after data retrieval.
`Passive
`ranging of
`targets from the TAS
`inputs
`is an example of
`a
`high-resolution algorithm.
`The TAS provides FLIR-
`derived target azimuth and elevation information to
`the terrain data subsystem,
`allowing the terrain
`system to trace a line-of-sight radial
`through the
`
`559
`
`BOHNG
`
`Ex.1031,p.630
`
`localing
`sensors,
`subsystem's
`the
`searched with
`targets (azimuth and elevation),
`identifying and
`classifying targets (using data from other subsys-
`tems, digital
`terrain maps, prebriefed data,
`the
`pilot, and external
`sources), and maintaining the
`current target list.
`
`subsystem manager maintains and
`The communications
`uses prebriefed frequency and call sign data; con-
`trols the message conditioning functions; manages
`target handoff,
`radio navigation, and IFF communi-
`cations;
`is responsible for distributing the sys-
`tem's data to the outside world;
`and routes
`in-
`coming data (such as inflight updates).
`
`is the primary
`subsystem manager
`The fire control
`when using certain weapons,
`it
`weapon manager.
`determines what maneuvers are required to properly
`align the aircraft before launch.
`It contributes
`recommendations for weapon and platform assignments
`and processes target designations and fire commands.
`when target data is sent or received via communica-
`tions, this subsystem is responsible for the mess-
`age content.
`
`The navigation subsystem manager maintains status
`and control of the Kalman filters and controls and
`updates the state vectors.
`It calculates the opti-
`mal trajectory to follow,
`taking into account
`the
`automatic flight and obstacle avoidance algorithms.
`It is
`in charge of
`the passive ranging and the
`offboard or non-line-of-sight
`target
`cueing func-
`tions.
`It controls the primary vehicle reference
`system and manages coordinate transformations.
`
`the
`The terrain data subsystem manager maintains
`aircraft's current position relative to the digital
`map data base during all maneuvers.
`It
`resolves
`about line-of-sight conditions.
`It must handle re-
`generation of
`the appropriate map resolution from
`the compressed data, combine map data with any over-
`lay data, and provide smooth map scrolling through
`the data
`base
`for both
`calculations
`and display
`functions. This manager fine-tunes route plans to
`take maximum advantage
`of
`the
`terrain.
`During
`internal or external updates or temporary loss of
`external navigation data,
`this system selects and
`orients the digital map
`from prominent
`local
`ter-
`rain features to maintain the aircraft's position
`at all
`times.
`
`The survivability subsystem manager constantly mon-
`itors and categorizes threats, maintains the cur-
`rent threat list, warns the pilot of any threats,
`recommends and sets
`up appropriate maneuvers
`and
`countermeasures, and
`initiates
`any
`automatic
`or
`requested countermeasures or
`threat surveillance.
`
`the pilot communicates with the system
`In summary,
`via the central manager. The central manager coor-
`dinates (sets priorities,
`resolves conflicts, as-
`signs modes, etc.)
`the
`interactions
`between
`the
`subsystem managers.
`The data manager supports sym-
`bolic data
`access
`and maintains
`the
`system data
`base. Each subsystem manager controls all activity
`internal
`to its own subsystem,
`including management
`of any assigned sensors or memory storage, proces-
`sing, testing, and reconfiguration.
`
`Terrain Data Subsystem (TDS)
`
`The TDS is a digital map data retrieval system that
`conditions stored digital map data into a form use-
`ful for avionics system algorithms. Modern avion-
`ics systems
`use digital data to perform terrain
`
`BOEING
`Ex. 1031, p. 630
`
`
`
`range and the local
`map data to determine target
`slope of
`the terrain at
`the target.
`The target
`range is determined by the intersection of
`a
`ray
`from the platform, defined by azimuth and elevation
`to the target, with the terrain along the line-of-
`sight vector.
`The
`local
`slope is useful
`in many
`target identification algortihms.
`In
`an
`active
`radar mode,
`the TAS could provide target
`azimuth
`and range to the terrain system, which would then
`give local slope and target elevation.
`
`High-resolution data is needed when creating real-
`istic digital
`imagery.
`A perspective
`image
`of
`encountered terrain requires three steps:
`radial
`addressing,
`hidden-line
`computations,
`and
`scene
`generation. Radials of
`sight defined by display
`resolution, field of view, and sight distance are
`addressed in the cache memories.
`The elevations
`of the radial
`samples are scaled with respect
`to
`distance from the observation point
`and compared
`to see which points are hidden.
`A projection of
`the points not hidden is used to generate the scene
`to be displayed.
`A plan view image is created by
`displaying the elevations, features, and/or cultur-
`al data
`from a
`rectangularly addressed array of
`points about a reference observation point. Other
`examples of high-resolution algorithms
`are
`data
`TABLE 2.
`
`conditioning for
`of display images.
`
`SITAN updates
`
`and data overlay
`
`An example of a low-resolution data algorithm is 3
`route-planning algorithm that uses stored threats
`targets, terrain,
`and cultural data with a coarse
`mission route and mission parameters to plan opti-
`mal mission routes.
`These
`routes will
`take into
`account terrain variations and the line of sight of
`the identified threats.
`Given the
`start
`and end
`points of
`a mission,
`approved waypoints
`(points
`between start and end that are safe to travel) are
`determined and all possible routes
`through these
`points plotted. Deterministic rules are then used
`to pick the optimal
`route, with mission-acquired
`information being used to update and
`change the
`route when necessary.
`
`The display formatting function converts the pro-
`cessed data into a display format.
`Examples are
`the overlaying of
`threat
`and
`target
`information
`onto scenes and perspective radial combinations in
`which all radials in the field of view are combined
`and filtered to create a pleasing scene.
`A stabil-
`ization algorithm is
`required to take
`roll
`and
`pitch into account
`in image displays.
`
`0 AREA COVERAGE
`
`0 DATASOURCE
`0 I/O
`
`0 [/0 DATA QUALITY
`0 DISPLAY
`0 NAVIGATION
`0 TF/TA
`
`LHX TERRAIN DATA SYSTEM
`REQUIREMENTS PRELIMINARY
`100 KM X 100 KM, EXPANDABLE TO
`1,000 KM BY 1,000 KM
`DMA LEVEL l
`
`ELEVATION DATA TO TFfI'A, NAV,
`WEAPON DELIVERY, THREAT AVOIDANCE
`
`I DISPLAY LEVEL MSE
`100 METER CEP FOR SITAN
`NO UNDERESTIMATES OF TERRAIN
`ELEVATION
`
`I DISPLAY MODES (256 X 256 PIXELS)
`0 PLAN VIEW
`0 LINE-OF-SIGHT
`0 PERSPECTIVE VIEW (OPTIONAL)
`0 AIRCRAFT VELOCITY
`0 DISPLAY UPDATE RATES
`0 HQAME
`0 COMPLETE SCENE
`0 SIZE. WEIGHT, POWER
`
`450 KNOTS
`
`60 FRAMES/SECOND
`10 TO 30 SECONDS
`1.5 ATR, 45 POUNDS, 450 WATTS
`
`Baseline Re uirements. Table 2 summarizes the base-
`line requirements for the TDS.
`It can be seen that
`the minimum area to be covered is 100 km x 100 km
`expandable to larger coverage by adding more memory
`or by exchanging memory modules.
`The source data
`will consist of DMA Level-I Digital Terrain Ele-
`vation Data
`(DTED), Digital Feature Analysis Data
`(DFA), plus
`imagery
`and
`information
`from other
`sources such as Landsat
`and intelligence sources.
`The TDS data will be distributed in a
`raw or pre-
`processed form; for example, information sent to the
`TF/TA processor
`could be
`an
`array of elevation
`values and/or vertical obstructions and heights, or
`range and angle information required to generate
`TF/TA commands.
`
`requirements,
`storage
`large data
`the
`Because of
`some form of data compression will be required to
`minimize memory
`requirements.
`For
`this purpose,
`techniques
`are
`suitable, e.g., discrete
`cosine
`
`Hadamard
`Block Truncation Coding,
`transforms,
`transforms, and polynomial
`surface coding.
`Texas
`Instruments is conducting studies as part of
`the
`Air Forces
`Integrated Terrain Access and Retrieval
`System (ITARS) program to determine the impact of
`data compression reconstruction techniques and the
`performance of TF/TA, navigation, display,
`threat
`avoidance and weapon delivery algorithms on system
`hardware and software implementation.
`As
`a result
`of the studies,
`it will
`be possible to match the
`data compression technique to the application and
`select the
`optimal
`implementation.
`Examples
`of
`criteria for
`selecting data
`compression/recon-
`struction algorithms appear
`in Table
`2 under
`the
`heading I/0 data quality.
`The TDS will
`be able
`to provide
`plan
`views
`and direct
`line-of-sight
`views and be capable of generating real-time per-
`spective views if required.
`Present design goals are to have a TDS that will
`operate for aircraft velocities up to 450 knots,
`
`560
`
`BOHNG
`
`Ex.1031,p.631
`
`BOEING
`Ex. 1031, p. 631
`
`
`
`BEE
`
`blocks
`on
`performed
`Initally, functions
`stores
`data.
`The magnetic-tape
`loader
`(MTL)
`Universal Transverse Mercator (UTM) blocks of digi-
`tal terrain data.
`Each UTM block represents 12.5
`x 12.5 kilometers.
`The intermediate memory acts as
`a cache memory,
`feeding data to the elevation and
`cultural reconstruction processors.
`
`of
`81
`
`is
`a Discrete
`reconstructed with
`Elevation data
`Cosine Transform/Differential Pulse Code Modulation
`method.
`A footprint method is used to reconstruct
`features.
`is
`then
`Reconstructed
`data
`cultural
`stored in the scene memory, which acts as another
`cache memory,
`feeding data to the per-frame func-
`tions.
`Functions are then performed each display
`390
`380
`570
`360
`550
`5190
`330
`520
`3 I 0
`500
`290
`230
`270
`260
`250
`2150
`250
`220
`2l0
`
`300
`
`£00
`
`|2.5 M
`
`I00 H
`
`200
`KNOTS
`
`100
`
`favorable
`have
`and
`frames/second
`60
`display at
`size, weight, and power characteristics as shown in
`Table 2.
`‘
`
`Again, it must be stated that Table 2 represents a
`baseline set of
`requirements which will evolve as
`new technology and new user requirements are ident-
`ified and defined.
`
`Sev-
`Terrain Data Subs stem hn lementation Issues.
`eral areas inherent
`in the defined IDS require in-
`novative design to allow implementation of the sub-
`system within the physical constraints of LHX.
`The
`bulk memory must be nonvolatile and lightweight and
`have reasonable access times.
`The leading techno-
`logies in this area are bubble memories and medium-
`density magnetic-tape technology. Data compression
`techniques currently in use may need to be improved
`to reduce the required storage to a more manageable
`level. Real-time
`generation
`of digital
`imagery
`requires very-high-speed address and graphics gen-
`eration. This
`task
`seems
`to be
`ideally suited
`for VHSIC processors.
`Two architectures currently
`being used for
`terrain processing are the (DMG)
`and the alternative DMG.
`
`The DMG which was developed under contract with the
`U.S. Army Avionics Research and Development activ-
`ity, reconstructs
`compressed digital map
`inform-
`ation and provides outputs suitable for driving a
`CRT display.
`The
`DMG has
`three major
`sections:
`block-oriented functions, per-frame functions,
`and
`video processor functions.
`
`OPS/SEC(MILLIONS)
`
`Figure 3.
`
`Digital Map Generator
`(Operations Per Second)
`
`DMG THROUGHPUT REQUIREMENTS
`TABLE 3.
`(DMG Operations per Second (12.5 meter data)
`300 Knots
`
`Tasks
`Elevation reconstruction
`Feature reconstruct
`Per-frame tasks
`Totals
`
`Tasks
`Elevation reconstruction
`Feature reconstruct
`Per—frame tasks
`Totals
`
`Adds
`3.60E+ 07
`4.3SE+ O6
`6.29E+07
`9.93E+07
`
`Sin-Cos
`3. 54E + 07
`
`Lookup
`6.83E+03
`3.47E +03
`
`l.03E+04
`
`Compares
`l.O2E+04
`1.l2E+04
`5.60E+07
`5.60E+07
`
`Logical
`6.83E + 03
`l .69E+ 02
`2.45E+07
`2.45E+07
`
`Inv Tan
`
`Divides
`
`Square Root
`
`Shift
`
`Mults
`
`2.7lE+05
`1.74E+04
`
`2.08E+05
`Ex Add
`
`3.55E+O7
`
`3.50E+O6
`3.90E+07
`Ex Mult
`
`6.00E+05
`4.23E+07
`
`6.00E+0l
`6.00E+0l
`
`6.00E +01
`6.00E+0l
`
`6.00E+0l
`6.00E+0l
`
`7.00E +06
`7.00E+06
`
`3.50E+06
`3.50E+O6
`
`Total OPS/second
`Total memory
`
`3.43E+06
`46 MBITS
`
`DMG Operations per Second (100 meter data)
`300 Knots
`
`Tasks
`
`Elevation reconstruction
`Feature reconstruct
`Per-frame tasks
`Totals
`
`Tasks
`Elevation reconstruction
`Feature reconstruct
`Per-frame tasks
`Totals
`
`Adds
`
`l.37E+05
`l.27E+05
`6.29E+ 07
`6.32E+07
`
`Sin-Cos
`l.30E+O5
`
`Lookup
`
`l.09E+03
`1.0 lE+03
`
`3.00E+03
`
`Compares
`
`2.99E+03
`3.26E+O3
`5.60E+07
`5.60E+O7
`
`logical
`
`l.99E+03
`4.94E+0l
`2.45E+07
`2.45E+07
`
`Inv Tan
`
`Divides
`
`Square Root
`
`Shift
`
`Mults
`
`2.0lE+03
`5.06E+03
`
`7.06E+03
`Ex Add
`
`l.30E+05
`
`3.50E+06
`3.63E+06
`Ex Mult
`
`6.00E+06
`7.03E+ 06
`
`6.00E +01
`6.00E+0l
`
`6.00E+0l
`6.00E+0l
`
`6.00E+ Ol
`6.00E+0l
`
`7.00E+O6
`7.00E+O6
`
`3.50E+06
`3.50E+06
`
`Total OPS/second
`Total memory
`
`2. 1 5E+08
`46 MBITS
`
`561
`
`BOHNG
`
`Ex.1031,p.632
`
`BOEING
`Ex. 1031, p. 632
`
`
`
`BOEING
`
`Ex. 1031, p. 633
`
`BOEING
`Ex. 1031, p. 633
`
`
`
`MIL-STD-1553 DUAL REDUNDANT REMOTE TERMINAL SUPERHYBRID
`
`84-2728
`
`BY STEVEN N. FRIEDMAN
`
`BBB ILC DATA DEVICE CORPORATION
`
`105 WILBUR PLACE,BOHEMlA,' NEW YORK 11716
`
`ABSTRACT
`
`the performance,
`paper describes
`This
`physical and electrical characteristics