Decentralized phase synchronization scheme for collaborative beamforming in wireless sensor networks

In this paper, we propose a phase synchronization scheme suitable for collaborative beamforming with wireless sensor networks. The scheme does not require the BS to coordinate the allocation of sensors to the training timeslots or poll them individually (which can be burdensome for large networks): sensors randomly choose their respective training timeslots. In that sense, the synchronization scheme is decentralized. In this context, we ask ourselves whether there exists an optimal number of training timeslots, and about the optimal split for the training and data transmission periods. To answer this question, we analytically derive upper bounds of the resulting beamforming gain in two scenarios of interest: ideal, and noisy phase shift estimation. Computer simulation results are mainly given in terms of (normalized) beamforming gain and achievable throughput.