Massive co-located devices require new paradigms to allow proper network connectivity. Internet of things (IoT) is the paradigm that offers a solution for the inter-connectivity of devices, but in dense IoT networks time synchronization is a critical aspect. Further, the scalability is another crucial aspect. This paper focuses on synchronization for uncoordinated dense networks without any external timing reference. Two synchronization methods are proposed and compared: i) conventional synchronization that copes with the high density of nodes by frame collision-avoidance methods (e.g., CSMA/CA) to avoid the superimposition (or collision) of synchronization signals; and ii) distributed synchronization that exploits the frames' collision to drive the network to a global synchronization. The distributed synchronization algorithm allows the network to reach a timing synchronization status based on a common beacon with the same signature broadcasted by every device. The superimposition of beacons from all the other devices enables the network synchronization, rather than preventing it. Numerical analysis evaluates the synchronization performance based on the convergence time and synchronization dispersion, both on collision and non-collision scenario, by investigating the scalability of the network. Results prove that in dense network the ensemble of signatures provides remarkable improvements of synchronization performance compared to conventional master-slave reference.