Inter-sensor propagation delay estimation using sources of opportunity

TL;DR

This paper investigates acoustic propagation delay estimation between sensors using uncontrolled, passive sources of opportunity, extending models to account for spectral characteristics and proposing bounds based on source properties.

Contribution

It introduces an extended observation model for passive delay estimation and a geometrical analogy to derive a lower bound considering source spectral and spatial properties.

Findings

Extended models account for spectral characteristics of sources.

Time-filtered sources affect delay retrieval accuracy.

Derived lower bounds on estimation variance considering source properties.

Abstract

Propagation delays are intensively used for Structural Health Monitoring or Sensor Network Localization. In this paper, we study the performances of acoustic propagation delay estimation between two sensors, using sources of opportunity only. Such sources are defined as being uncontrolled by the user (activation time, location, spectral content in time and space), thus preventing the direct estimation with classical active approaches, such as TDOA, RSSI and AOA. Observation models are extended from the literature to account for the spectral characteristics of the sources in this passive context and we show how time-filtered sources of opportunity impact the retrieval of the propagation delay between two sensors. A geometrical analogy is then proposed that leads to a lower bound on the variance of the propagation delay estimation that accounts for both the temporal and the spatial properties of the sources field.