Spaceborne Distributed Array Processing for Precision Source Localization

ABSTRACT

Satellite‐based signals intelligence and spectrum sensing increasingly rely on low‐latency signal localization. However, operational processing chains often remain ground‐based, stressing feeder links and limiting scalability as constellations grow. Moreover, widely used time‐difference‐of‐arrival/frequency‐difference‐of‐arrival (TFDoA) geolocation is constrained by signal‐to‐noise ratio (SNR) and effective signal bandwidth, which limit accuracy for a given observation time. Single‐platform direction‐of‐arrival (DoA) estimation is likewise aperture‐limited, particularly in bands where spacecraft antennas are electrically small relative to the wavelength. Sparse long intersatellite baselines, however, enable sharp nominal angular resolution but induce manifold grating lobes, yielding direction ambiguities. This paper studies multiplatform precision localization by combining two complementary enablers for distributed satellite systems: (i) aperture fractionation to synthesize a large virtual array and (ii) distributed algorithms to avoid feeder‐link bottlenecks and the burden of centralizing raw sample streams. Accordingly, the proposed framework couples long‐baseline processing with ambiguity‐mitigation mechanisms and coherence‐aware modeling. System‐scale stress metrics quantify when narrowband and far‐field (plane‐wave) assumptions break down across the synthesized aperture. Building on these models, a two‐stage, decentralized DoA processing chain is developed. Platform‐level subarray processing provides low‐complexity, locally unambiguous intermediate statistics, while coherent system‐level processing exploits long baselines to refine the bearing estimates; the network‐wide quantities are computed in a distributed fashion through average consensus, avoiding the centralization of raw sample vectors. Communication overhead and computational complexity are characterized under inter‐satellite link constraints, and DoA‐based performance is evaluated under sparse‐baseline ambiguity and inter‐platform impairments. The approach is demonstrated in a satellite‐based Automatic Identification System (AIS) scenario and shows significant accuracy gains under realistic constellation conditions.