Palur Palanivelu, DevanandDevanandPalur PalaniveluOispuu, MarcMarcOispuuKoch, WolfgangWolfgangKochDallmann, ThomasThomasDallmann2024-10-012024-10-012024https://publica.fraunhofer.de/handle/publica/47594410.1109/SAM60225.2024.106366852-s2.0-85203384015Source localization from raw array data by repa-rameterizing the array steering vector by the source position is a fundamental principle of direct position determination (DPD). In this paper, we propose a DPD method to estimate the position of locally scattered sources using an unsynchronized array sensor network. Here, we use the well-known generalized array manifold (GAM) model which approximates the steering vector using its first-order gradient in order to characterize the local scattering effect. The proposed method is compared with the conventional two-stage bearings-only localization (BOL) approach. Simulation results reveal that the proposed DPD position estimates asymptotically attain the derived Cramér-Rae Bound (CRB) for high SNR values and an improved localization accuracy is achieved by exploiting the local scattering parameters in the localization.enCramér-Rao BoundDirect Position DeterminationGeneralized Array Manifold modelLocal ScatteringLocalization Accuracyconference paper