Utilizing the time differences of arrival of a signal, impinging on a distributed sensor network, is a well known approach for the location estimation of a radio wave emitter. The high sampling rate necessary for a precise positioning implies a huge amount of data exchange between the sensor nodes. Contrary, the absence of high data rate enabled backbone links, connecting the nodes, restricts the system performance or may even render it dysfunctional in some cases. In order to tackle the problem we propose a novel method for compressed time difference of arrival based localization. Due to the joint spatial sparsity of the underlying problem the amount of exchanged samples can be reduced by applying the compressed sensing methodology. Furthermore, our algorithm provides a direct way of estimating the location, avoiding the necessity of solving a system of nonlinear hyperbolic equations. The performance of the proposed method and the impact on the estimation accuracy are evaluated based on simulations.