Distributed Synchronization for Ad-Hoc Acoustic Sensor Networks

Aleksej Chinaev and Gerald Enzner
Carl-von-Ossietzky University of Oldenburg

With the concept of signal processing in wireless acoustic sensor networks, network-wide synchronization is required to exhaust their potential in signal enhancement. Our demo presents a gossiping approach for distributed synchronization using waveform-based estimation of sampling rate offset (SRO) by the double-cross-correlation processor (DXCP). To this end, a Jupyter notebook is implemented on asynchronous microphone signals of a simulated Sound Interface to the Swarm (SINS) environment shown in Fig. a). The simulution uses the Acoustic Sensor Network SIgnal Generator (ASNSIG) framework presented separately in demo D1. As shown for a possible topology in Fig. b), each node receives a local reference signal one-way either from a global reference node (here node 0) or from a parent node (here node 9) on a rooted-tree topology. The proposed method then operates closed-loop with DXCP-based SRO estimation and resampling to the local reference. We demonstrate how the synchronization gossip for SRO estimation propagates network-wide and converges to the ground truth for various configurations of the topology, as shown for instance in Fig. c).