On the tradeoff between resolution and ambiguities for non-uniform linear arrays

The performance of most direction of arrival (DoA) estimation algorithms is driven by two different kinds of errors: small, local errors in the main lobe and possibly large errors due to sidelobe peaks ambiguities. Reducing these two kinds of errors simultaneously is not possible and therefore a compromise has to be made. Motivated by automotive applications with small arrays, we consider solving this tradeoff by modifying the positions of the sensors of the array. In this letter, two new criteria are proposed to solve the above mentioned tradeoff. Optimal solutions are derived and illustrations are provided with a 3-element non-uniform linear array.