Designing a Realtime Test Environment using Model Continuity

1
Designing a Real-time Test Environment using
Model Continuity
2
Problem Measure the effectiveness of a track
correlation system
A particular track correlation system can 1)
observe a set of aircraft with its own sensor
system, and 2) receive information about a set of
aircraft from other sensor systems. How
effectively can the correlation system determine
which aircraft that are observed locally
correspond to the aircraft being reported from
other sensor systems?
3
Real trajectory Observed trajectory Reported
trajectory
4
Relevant issues

• Reports are provided periodically (typically
  every six or twelve seconds). Reports can be
  lost or delayed in the communication network.
• There can be any number of reporters for a given
  object.
• There can be any number of objects being reported
  on and observed simultaneously.
• Reported positions and observed positions will
  deviate from true position by some error whose
  upper bound is known.
• Not all observed objects are reported on and not
  all reported objects are being observed.

5
Objectives

• We have been given the correlation system and a
  set of aircraft and missile trajectories.
• Want to produce a simulator that can be used to
  create a modeled environment to experiment with
  the algorithms of the correlation system.
• Want to re-use the same models in this simulator
  to create a synthetic environment that operates
  in real time to experiment with the real
  correlation system.

6
Basic model components and their coupling
Remote report model
Air and missile trajectories
Correlation Model
Observation model
7
Remote report model

• For a first try, make some simplifying
  assumptions
• All objects are reported on.
• Each object has one reporter.
• An object is reported on with a fixed period.
• Reports are error free (reported position matches
  exact position).
• Reports are transmitted instantaneously.
• No reports are lost.

8
Remote track model
Trajectory sampler
Remote model contains one trajectory sampler for
each trajectory. A special component, the
dynamic structure executive, creates trajectory
sampling models as new objects are discovered.
9
Observation model

• Again, make some simplifying assumptions
• Errors in the observed trajectories are due to
  misalignment of the sensor
• Results in a translated and rotated world view
• Sensor has an unlimited field of view

10
Observation model
Observation model
Observation model produces one observation per
object. Observations are produced by translating
and rotating the object trajectory.
11
Discrete Event Simulator working in logical
time to test correlation algorithms
Discrete event simulator
Modeled environment
Correlation model
Remote report model
Air and missile trajectories
Observation model
12
Real-time Federation to test real correlation
system
outputs translated to emulator input format
Radio terminal emulator
hardware interface
Correlation system
Re-using model components, replace the simulator
by a real-time executor
Sensor emulator
outputs translated to HLA interactions
internally integrated coupling
proprietary software with HLA interface