Estimation of the direction-of-arrival (DOA) of signals embedded in noise is a problem of great interest in applications such as RADAR, SONAR, and wireless communication. In many situations of practical interest the signal-to-noise ratio (SNR) at the receiving array of sensors is too low for satisfactory performance of the DOA estimation techniques. The performance may be improved if the SNR at the receiving array of sensors is enhanced prior to performing DOA estimation. In this talk, some nonlinear techniques of SNR enhancement based on wavelet transforms and stochastic resonance will be presented. Simulation results will be presented to illustrate the improvement in terms of mean square estimation error and resolution of closely spaced sources.