We address the sensor network localization problem given noisy range measurements between pairs of nodes. We approach the non-convex maximum-likelihood formulation via a known simple convex relaxation. We exploit its favorable optimization properties to the full to obtain an approach that: is completely distributed, has a simple implementation at each node, and capitalizes on an optimal gradient method to attain fast convergence. We offer a parallel but also an asynchronous flavor, both with theoretical convergence guarantees and itera- tion complexity analysis. Experimental results establish leading performance. Our algorithms top the accuracy of a comparable state of the art method by one order of magnitude, using one order of magnitude fewer communications.