1: \begin{abstract}

2:   We address the sensor network localization problem given noisy range

3:   measurements between pairs of nodes. We approach the nonconvex

4:   maximum-likelihood formulation via a known simple convex

5:   relaxation. We exploit its favorable optimization properties to the

6:   full to obtain an approach that: is completely distributed, has a

7:   simple implementation at each node, and capitalizes on an optimal

8:   gradient method to attain fast convergence.  We offer a parallel but

9:   also an asynchronous flavor, both with theoretical convergence

10:   guarantees and iteration complexity analysis. Experimental results

11:   establish leading performance. Our algorithms top the accuracy of a

12:   comparable state of the art method by one order of magnitude, using

13:   one order of magnitude fewer communications.

14: \end{abstract}

15: