A method for identifying the memory deficiency i.e. Alzheimer disease is presented. In this work, an electroencephalogram signal is taken from the scalp of human brain & it is analyzed by means of different signal processing tools. In present work, EEG signals from various electrodes are obtained, filtered by preamplifier circuit & are then are fed to Analog to digital converter for digitization. The signals are then given as an input to microcomputer for analysis. The microcomputer performs the analysis of signals & expected result is displayed through display unit. The same system is also interfaced with Computer & printing unit to print the obtained results. In this work, different features of EEG signals comprising of both Normal & Alzheimer affected patient are calculated. It is observed that EEG signal of Alzheimer patients is less complex as compared to that of the Normal patients. In present work, different complexity measures such as Spectral Entropy, Spectral Centroid, Spectral Energy & Zero Crossing Rate are used as different features to verify whether these features plays a significant role in differentiating between AD affected & Normal subjects. The classification in the above method is done by use of K Nearest neighbor method to classify the data between the two groups.