Through a review of the so-called discriminatory channel estimation (DCE) scheme [1,2], this chapter introduces a way to enhance physical layer secrecy from a channel estimation perspective. The DCE scheme is a novel training scheme that is aimed at providing discrimination between the channel estimation performances at a legitimate receiver (LR) and an unauthorized receiver (UR). This effectively enlarges the difference between the signalto-noise ratios (SNRs) at the two receivers and leaves more room for secrecy coding or transmission schemes in the data transmission phase. A key feature of DCE designs is the insertion of artificial noise (AN) in the training signal to degrade the channel estimation performance at UR. To do so, AN must be placed in a carefully chosen subspace to minimize its effect on LR. However, this requires preliminary knowledge of the channel at the transmitter, which can be difficult to achieve without benefiting the channel estimation at UR as well. To achieve this task, two DCE schemes were proposed in the literature, namely, the feedback-and-retraining [1] and the two-way training [2] based DCE schemes. Both of these schemes require a preliminary training stage to provide the transmitter with a rough estimate of LR’s channel and an AN-assisted training stage to enable channel estimation at LR while degrading channel estimation at UR. These schemes as well as methods to optimally allocate power among pilot signals and AN are reviewed in this chapter.