1: \begin{abstract}

2:   % Oscillator synchronization is an instrumental component of many

3:   % engineering applications. For example, it can provide networked

4:   % devices with a common temporal reference necessary for coordinating

5:   % actions or decoding transmitted messages.  In this paper, we study

6:   % the problem of achieving both phase and frequency synchronization on

7:   % a network of heterogeneous oscillators using only local

8:   % measurements.  Most current solutions suffer from phase differences

9:   % in steady state due to frequency heterogeneity; others provide a

10:   % convergence analysis which is valid only locally, only under

11:   % synchronous adaptation, or only under a regular graph structure.  In

12:   % contrast, our solutions can exhibit little or no steady-state phase

13:   % differences under arbitrary frequency heterogeneity.  Furthermore,

14:   % we provide a global convergence analysis valid on arbitrary

15:   % connected graphs and either in continuous time or under sufficiently

16:   % fast asynchronous updates.

17:   Many network applications rely on the synchronization of coupled

18:   oscillators.  For example, such synchronization can provide

19:   networked devices with a common temporal reference necessary for

20:   coordinating actions or decoding transmitted messages.  In this

21:   paper, we study the problem of using distributed control to achieve

22:   both phase and frequency synchronization of a network of coupled

23:   heterogeneous nonlinear oscillators.  Not only do our controllers

24:   guarantee zero phase error in steady state under arbitrary frequency

25:   heterogeneity, but they also require little knowledge of the

26:   oscillator nonlinearities and network topology.  Furthermore, we

27:   provide a global convergence analysis, in the absence of noise 

28:   and propagation delay, for the resulting nonlinear system whose 

29:   phase vector evolves on the $\bm{n}$-torus.

30: \end{abstract}

31: