TIME SYNCHRONIZATION THRESHOLDS FOR FUSION LEVELS IN MULTISENSOR RADIO MONITORING

Abstract

The paper analyses how time synchronization accuracy between sensor nodes affects the probability of erroneous decision in a multisensor radio monitoring system. The fusion level of sensor streams (primary observations, features, or local decisions) is shown to be a structural parameter that determines which time scale governs the tolerable synchronization error.

A two-sensor model of distributed reception with independent Gaussian timing offsets is constructed. For each of the three fusion levels, the dependence of the probability of error on the synchronization-error standard deviation is derived analytically; the primary-level expression is obtained as an expectation over the Gaussian phase-jitter distribution. It is shown that at the local-decision and feature levels the relevant time scale is the symbol duration or the integration window, while at the primary level it becomes the carrier period, which is several orders of magnitude shorter at radio frequencies.

The derived primary-level expression is verified by Monte Carlo simulation of the same two-sensor signal model under Gaussian phase jitter. The analytical dependence and the simulation results agree across the full range of synchronization-error values. The simulation is repeated for three values of the per-sensor signal-to-noise ratio (0, 5, 10 dB) and for three representative carrier frequencies (100 MHz, 433 MHz, 2.4 GHz), confirming that the location of the transition region is independent of the signal-to-noise ratio and scales as 1/(2π · fc) with the carrier frequency.

The accuracy regimes of established time-distribution standards (NTP, IEEE 1588 PTP in software and hardware variants, GNSS-disciplined oscillators, White Rabbit) are qualitatively mapped onto the fusion levels for which these standards may be applicable, with the actual limit depending on the carrier frequency. The results enable joint selection of synchronization accuracy and fusion algorithm at the design stage of a multisensor system.