Network Time Synchronization Using OFDM-Based Cooperative Consensus (CoCo) Algorithm

We investigate the convergence behavior of the novel "CoCo" network time synchronization protocol, which compensates both offset and skew, for a network of two tight clusters of nodes, taking fading, quantization error consistent with practical OFDM modulation, and additive noise into account. CoCo update equations are similar to the consensus-based asynchronous time synchronization (ATS) protocol, but CoCo replaces the "one-by-one" packet exchanges with neighbors in ATS with the approximate sample mean provided by the semi- cooperative spectrum fusion (SCSF) technique of concurrent transmissions, in just one packet interval. SCSF requires that all participating nodes share the same upper and lower bounds on the parameter being transmitted. In this paper, two of the three SCSF-transmitted parameters have deterministic bounds; the third parameter uses optimized adaptive bounding, which is not practical, but used in this paper to learn how the spreads of the parameter vary with time and other parameters. Simulation results show a rapid convergence to very low rms time errors, followed by a slow divergence, suggesting that CoCo should be stopped just after the rapid convergence, for best results.