Wideband coherent sources localization based on a two-node distributed sensor networks

• A Teoplitz matrix based on two sensors can de-correlate multiple wideband sources.
• Exist noise subspace orthogonal to signal subspace even matrix is non-Hermitian.
• The prior geometrical information can be helpful to eliminate false intersection.
• Combine all the ideas above, the localization of wideband sources can be achieved.

In conventional research of source localization using sensor networks, usually many sensors were required, and only a small number of sources could be localized. Meanwhile it is still very hard to deal with coherent sources for conventional methods. In this paper, a new approach is proposed to localize wideband coherent sources based on a distributed sensors networks. For this sensor networks, only two node-arrays are required, and each node-array consists of only two sensors. Firstly, direction-of-arrivals (DOAs) estimation is done at each node-array by employing a new algorithm proposed in this paper, which can estimate DOAs of multiple wideband coherent sources only using two sensors. Next, combining the pattern matching idea and the prior geometrical information of sources, a cost function is constructed, and the localization problem is converted to obtain an optimal solution that minimizes the cost function. Combining the cost function and the estimated DOAs results of each node-array, multiple wideband coherent sources can be localized. Numerical examples are provided to demonstrate effectiveness of the proposed approach.