SVM-based Fuzzy Inference System (SVM-FIS) for frequency calibration in wireless networks

To guarantee the connectivity in wireless networks is an important issue. For this purpose, we have to provide precise frequency for the network such that it can meet the various requirements of future applications. In this paper, we not only propose a novel frequency calibration methodology based on SVM-based fuzzy inference system (SVM-FIS) in wireless networks, but also against temporary failure of wireless connection. There are two modes - normal mode and handover mode - in the calibration system. In normal mode, the primary clock can periodically delivers the controlling signals to steer the remote station's oven-controlled crystal oscillator (OCXO) through the wireless connection. Once the wireless connection was temporarily crashed, the station switches to holdover mode which the SVM-FIS controller takes over sending the controlling signals from primary clock to steer its OCXO. Experimental results indicate that performance of frequency stability of the OCXO can be improved from a few parts in 10⁹ to 10¹³ for averaging times of one day, and the performance can be maintained within a few parts in 10¹² over a day in holdover mode.