Calinda Huang, SRI International; David Keller, SRI International; Mark Grygon, SRI International; Patrick Maeda, SRI International; Namwoong Paik, SRI International
https://doi.org/10.64861/ATDD1306
Keywords: Space domain awareness, faint object detection, radiometric modeling, aperture scaling, optical instrumentation, Neyman-Pearson detection, probability of detection
Abstract:
Closed-loop tip-and-cue for Space Domain Awareness (SpDA) pairs a wide-field-of-view (WFOV) cueing sensor that searches a region of interest (ROI) with a narrow-field-of-view (NFOV) sensor that confirms and characterizes candidates. The loop only closes fast enough to be operationally useful if the cueing sensor can detect a candidate in a small number of frames. Each additional frame the cueing sensor must accumulate to declare a detection drives onboard compute, memory footprint, trigger latency, and — for harder cases where raw ROI thumbnails are downlinked for ground-side confirmation — telemetry volume proportionally higher. This work identifies and quantifies a sharp, nonlinear operating boundary in aperture size across which closed-loop cueing transitions from multi-frame accumulation to single-frame trigger operation.
The analysis formulates aperture sizing, frame-stacking depth, trigger latency, and per-cue onboard processing volume as a single coupled trade. Detection performance is computed using a radiometric and statistical model that maps aperture diameter, integration time, and measured focal plane parameters (SRI next-generation 8k × 8k CMOS visible configuration) into signal-to-noise ratio (SNR). The model includes shot noise, background radiance under representative Geosynchronous Earth Orbit (GEO) conditions (including zodiacal light), dark current, and read noise. Multi-frame accumulation assumes statistically independent frames with a conservative stacking efficiency of 0.80. Probability of detection (Pd) at fixed false-alarm rate (FAR) is computed under a Gaussian Neyman–Pearson formulation, enabling optical parameters to be propagated directly into trigger operating behavior.
Sweeping aperture from 0.10 m to 0.40 m at fixed f/# = 2.286 (matching the SpDA Payload with Onboard Control & Knowledge (SPOCK) wide-field three-mirror anastigmat, TMA, baseline) reveals a sharp sizing boundary. For a faint GEO target (m_AB = +20) at a 7-second per-frame exposure under nominal sky conditions, the boundary lands at D* ≈ 0.34 m. Below the boundary, the cueing sensor requires 2 to 41 stacked frames to reach the same detection confidence (Pd = 0.95 at one false alarm per 100,000 looks). Above it, one frame is sufficient. Above the boundary, aperture growth produces a sharp collapse in required accumulation depth: from 41 frames at D = 0.10 m down to a single frame at D ≥ 0.34 m. This translates to a 41× reduction in both trigger latency (from 287 s down to 7 s) and per-cue onboard processing volume.
The SPOCK baseline aperture (D = 0.35 m) sits 0.01 m above the boundary at the m_AB = +20 stretch threshold, placed there by independent optical-design decisions (wide-field three-mirror anastigmat sized for ESPA Grande payload volume and the SRI next-generation 8k × 8k format) and confirmed by this boundary analysis as the architectural design point for single-frame trigger detection at the threshold of operational interest, while preserving the wide instantaneous field of view (5.51 deg²) needed for efficient GEO belt cueing.
Parameter sweeps over target magnitude, integration time, and background radiance indicate that the sizing boundary persists across the full operational envelope, although the aperture at which the boundary occurs shifts predictably with each parameter. The boundary therefore reflects a structural feature of the coupled radiometric-autonomy geometry rather than an artifact of one background assumption. Combined sensitivity bounds place D* between roughly 0.25 m and 0.40 m across the realistic operating envelope; the nominal SPOCK design point sits centrally within this range..
By embedding radiometric detection theory within a closed-loop cueing architecture and propagating optical size, weight, and power (SWaP) decisions into accumulation depth, latency, and per-cue processing volume, this framework provides mission architects and acquisition decision-makers with a quantitative sizing tool. It inverts the usual sizing conversation — instead of asking “is the SNR margin met?” it asks “on which side of the boundary do we sit?” — and enables the aperture decision to be defended directly against the operational requirement it serves. All results are derived from end-to-end radiometric and statistical modeling anchored on measured focal plane characterization data, with a defined pathway toward hardware-in-the-loop validation.
Date of Conference: September 15-18, 2026
Track: SDA Systems & Instrumentation