An instantaneous frequency estimation algorithm is developed for multi-component signals acquired through multiple sensors. The proposed method estimates the mixing matrix for the strongest source in the given multi-sensor recording. The information of the mixing matrix is used to extract the corresponding source through spatial filtering. The spatially filtered signal is then analyzed using adaptive directional time-frequency distribution. The instantaneous frequencies of the extracted sources are then estimated through the Random sample consensus (RANSAC) algorithm that involves the detection of peaks, fitting of the instantaneous frequency curve using Fourier series approximation, and evaluating the instantaneous frequency estimate using an objective function. Both the mixing matrix and instantaneous frequency information is used to remove the corresponding source from the mixture signal by employing time-frequency and spatial filtering. The process mentioned above is then repeated till the instantaneous frequencies of all the sources have been estimated. Once the instantaneous frequencies of all the sources have been estimated, each source's instantaneous frequency is re-estimated by removing all the other components using time-frequency and spatial filtering. Till convergence, the process is iterated. Through numerical results, it is shown that the proposed method performs better than the relevant methods in the state of the art.