The overall basic approach is to use piezoelectric MEMS sensors and passive voltage gains from piezoelectrics MEMS transformers to detect the physical assets or RF signals, respectively, followed by a low power integrated circuit (the “nanoWatt classifier”) to do the signal processing to classify the detected signals. For physical asset detection and classification, passive piezoelectric MEMS resonant seismic sensors will be utilized to produce output signals suitable for direct use by the nanoWatt classifier. The narrow-band resonant sensors also perform spectral decomposition of the signatures to aid classification by having a bank of sensors, each designed to a different resonant frequency selected to provide maximum differentiation of the a physical assets being detected. To achieve 10nW power consumption, the signal processing electronics will use a combination of translinear computational circuits operating deep in the weak inversion regime and non-volatile analog memories to classify seismic signals at extremely low energy cost. Progressive power-gating of higher order correlators will ensure that only the simplest and lowest-power sub-circuits are enabled continuously, while making available more sophisticated algorithms for the infrequent events that are more difficult to classify (Figure 1).