3D passive source localization by a multi-array network: Noncoherent vs. coherent processing

Abstract

This paper investigates the three-dimensional passive source localization problem by a system of multiple antenna arrays. It is well-known that centralized and coherent processing of all sensor antenna outputs offers a performance benefit in comparison to decentralized processing, where the latter is easier to implement. We compare the Cramér-Rao bound on the source locations for both cases, review the decentralized direct position determination approach and outline the near-field source localization approach generalized to the three-dimensional problem. Finally, we discuss the trade-off between performance benefit and implementation complexity

Direct Wideband Coherent Localization by Distributed Antenna Arrays

“…Even though the criterion function (12) has a similar form as the original MUSIC in [26], the differences of the SCM-MUSIC algorithm relative to the ones in the literature are as follows. In our approach, the unknown parameters (the location of the Tx) are contained in the covariance matrix (this is the steered covariance matrix approach), whereas in [5,26], they were contained in the steering vector, which limits the application to narrowband signals. We model bandpass signals, unlike [23,25], where the signals are in the baseband ([23] is for acoustic signals), and our antenna array is distributed, unlike [23,26].…”

“…With spatial coherence, the accuracy of localization can be much improved. The authors in [5,6] analyzed the accuracy gains of coherent localization in passive systems and multiple-input multiple-output (MIMO) radars (active localization), respectively. Unlike [6], in which the receiving system was synchronized to the transmitters that are at known locations (radar/active localization), in our paper, the receiving system is not synchronized to a transmitter, and the transmitter is at an unknown location (passive localization).…”

“…Some practical implementations of passive localization were described in [7,8]. It is important to point out that the methods in [5,6,7,8] relied on spatial coherence for increased accuracy, but are suited to narrowband signals.…”

“…The receiving channels are phase and time synchronized, but the alignment of the transmitter’s time axis to the one of the receiving system is not required (unlike for ToA-based methods such as in [6]). In contrast to, for example [5,6], our signal model is both frequency-wideband and spatial-wideband. The effects of spatial-wideband and frequency-wideband signals are discussed in detail in [21,22].…”

A glance on geographical positioning

“…One step approach has been successfully proposed for moving sensors context [36], [37] and to the GNSS [38]. Passive localization of radio transmitters has also received great interest [39][40][41]. Note that such one step strategies assume that all signals of all base stations are directly transmitted to the CPU, they suffer from practical implementation issues.…”