A major challenge for Space
Domain Awareness (SDA) and CounterSpace
Operations (CsOps) is that the number of Resident
Space Objects (RSO) – satellites as well as debris – is increasing
exponentially while their physical size is decreasing. In
particular, this is a challenge for the Data Fusion (DF) functions
of detecting, tracking, classifying, and characterizing Space Potential
Threat Events (SPTE) fast enough and with enough certainty to allow
effective and timely CsOps. It is infeasible to
continuously track all RSO because there are magnitudes more RSO than senors and EOIR sensors cannot “see” RSO if obscured by
Earth or non-illuminated by the Sun or Moon. However, most RSO are in Keplerian orbits
so their kinematic state (position and velocity) can be predicted once
their ephemeris is known. Once an
RSO’s ephemeris is known, it is declared to be in “custody”.
For many
reasons, achieving custody of new object -- called an Uncorrelated Track
(UCT) -- can take days. (A UCT can
arise for many reasons such as newly launched object, an object separated
from a mother RSO, an RSO conducting an unanticipated maneuver, collision
fragments, or stale or otherwise inaccurate ephemeris parameters on the
RSO.) Unfortunately, there are cases
where the SPTE threat to own asset RSO may require CsOps
to be executed more rapidly to be effective. (CsOps could be
defensive such as maneuvering or self-destructing own asset or offensive
action such as jamming, other EW, destruction, or disabling of the SPTE
object.)
Quick Upstream
Anomaly Detection (QUAD) is an innovative alternative approach that can
provide Quick Reaction (QR) Indications and Warning (I&W) of SPTE. It
is an Upstream Data Fusion (DFU) approach based on comparing
EOIR space sensor observational data with synthetically generated imagery
to detect unexpected observations that may be indicative of SPTE. The
first stage of the analysis is to compute the differences – called residuals
-- between predicted synthetic and real image frames. The second
stage links or associates the residuals across frames to screen out
spurious residuals. The final stage then screens SPTE hypotheses
based on their classification, threat evaluation, and a decision processes
to determine if an SPTE alert should be generated. If an SPTE alert
is decided, an imagery snippet is made that is efficiently short but
provides sufficient evidence along with the SPTE annotations for an
operator to decide if CsOps mitigations should be
taken, e.g., maneuver, Electronic Warfare (EW), or interception. Four
things make all of these operations feasible:
1.
Leveraging of existing technologies:
synthetic imagery generation, image change detection, and fast Linear
Assignment Program (LAP).
2.
A frequently updated and comprehensive
catalog of ephemerides of all RSO in custody along with well-established
propagation algorithm repositories such as for Simplified General
Perturbations (SGP4) to estimate the expected RSO state at observation time
and thereby the image position in the focal plane of the sensor.
3.
A Space Sensor Network (SSN) cloud to which
all space sensor data is published and is accessible via an API. An
example of an element of the SSN on the unclassified side is Unified Data
Library (UDL).
4.
Operators have stated that SPTE are “easily
identified visually in calibrated imagery.”
This DFU
doesn’t replace but precedes and enhances the conventional Downstream Data
Fusion (DFD) of filtering, correlating, classifying, achieving
custody. DFU augments DFD so that the overall DF
system meets DF performance requirements such as probabilities of detection
and false alarm while also satisfying mission requirements for timely,
sometimes rapid, response.
|