1: \begin{abstract}

2: Massive co-located devices require new paradigms to allow proper network

3: connectivity. Internet of things (IoT) is the paradigm that offers

4: a solution for the inter-connectivity of devices, but in dense IoT

5: networks time synchronization is a critical aspect. Further, the scalability

6: is another crucial aspect. This paper focuses on synchronization for

7: uncoordinated dense networks without any external timing reference.

8: Two synchronization methods are proposed and compared: i) conventional

9: synchronization that copes with the high density of nodes by frame

10: collision-avoidance methods (e.g., CSMA/CA) to avoid the superimposition

11: (or collision) of synchronization signals; and ii) distributed synchronization

12: that exploits the frames' collision to drive the network to a global

13: synchronization. 

14: 

15: The distributed synchronization algorithm allows the network to reach

16: a timing synchronization status based on a common beacon with the

17: same signature broadcasted by every device. The superimposition of

18: beacons from all the other devices enables the network synchronization,

19: rather than preventing it. Numerical analysis evaluates the synchronization

20: performance based on the convergence time and synchronization dispersion,

21: both on collision and non-collision scenario, by investigating the

22: scalability of the network. Results prove that in dense network the

23: ensemble of signatures provides remarkable improvements of synchronization

24: performance compared to conventional master-slave reference.

25: \end{abstract}