Multi-target localization, warranted in emerging applications like autonomous driving, requires targets to be perfectly detected in the distributed nodes with accurate range measurements. This implies that high range resolution is crucial in distributed localization in the considered scenario. This work proposes a new framework for multi-target localization, addressing the demand for the high range resolution in automotive applications without increasing the required bandwidth. In particular, it employs sparse stepped frequency waveform and infers the target ranges by exploiting sparsity in target scene. The range measurements are then sent to a fusion center where direction of arrival estimation is undertaken. Numerical results illustrate the impact of range resolution on multi-target localization and the performance improvement arising from the proposed algorithm in such scenarios.