Rolling element bearings are widely used in rotating machinery. Bearing health condition monitoring plays a vital role in predictive maintenance to recognize bearing faults at an early stage to prevent machinery performance degradation, improve operation quality, and reduce maintenance costs. Although many signal processing techniques have been proposed in literature for bearing fault diagnosis, reliable bearing fault detection remains challenging. This study aims to develop an online condition monitoring system and a signal processing technique for bearing fault detection. Firstly, a Zigbee-based smart sensor data acquisition system is developed for wireless vibration signal collection. An enhanced Teager-Huang transform (eTHT) technique is proposed for bearing fault detection. The eTHT takes the several processing steps:​ Firstly, a generalized Teager-Kaiser spectrum analysis method is suggested to recognize the most representative intrinsic mode functions as a reference. Secondly, a characteristic relation function is constructed by using cross-correlation. Thirdly, a denoising filter is adopted to improve the signal-to-noise-ratio. Finally, the average generalized Teager-Kaiser spectrum analysis is undertaken to identify the bearing characteristic signatures for bearing fault detection. The effectiveness of the proposed eTHT technique is examined by experimental tests corresponding to different bearing conditions. Its robustness in bearing fault detection is examined by the use of the data sets from a different experimental setup.