The problem of passive localization is commonly solved by independently measuring intermediate parameters (such as angles of arrival (AOA), times of arrival (TOA)...) on several multiple sensors base stations in a first step. In a second step, the transmitted parameters are then used to estimate the position. Recently, studies proposed new promising one step algorithms based on stacked multiple station observations vectors gathering all received signals of all base stations. This strategy leads naturally to a wideband estimation problem, solved in this paper by an alternative space-time processing. The proposed algorithm is compared to existing techniques and the corresponding Cramer-rao bound.