ALTERNATIVE SAMPLING STRATEGIES Many novel ADC sampling strategies have emerged over the last decade, targeting a significant reduction in sampling energy consumption compared to traditional Nyquist rate conversion [Figure S2(a)], by exploiting a priori signal information. To this end, compressed sensing [Figure S2(b)] and innovation rate sampling [Figure S2(c)] try to reduce the sampling bandwidth as closely as possible to the signal's information rate. Feature-sampling ADCs [Figure S2(d)] reduce the dimensionality of the waveform through analog analytics to retain only applicationrelevant signal features, with the intention of classifying these features instead of reconstructing the original waveform. Nyquist ADC Analog Signal Bandwidth = W F_Sample = 2W (a) Compression Sequences SubNyquist ADC Modulator Integrator Analog Signal Bandwidth = W Average Pulse Rate = P F_Sample = M ⋅P < 2W (b) SubNyquist ADC Smoothing Filter Analog Signal Bandwidth = W Average Pulse Rate = P Physical Bandwidth Signal Reconstruction F_Sample = 2 P << 2W (c) Feature Enhancing Filter Analog Signal Bandwidth = W Average Pulse Rate = P Signal Reconstruction e.g., Feature = Pulse Amplitude (d) SubNyquist ADC Feature Processing/ Classifier F_Sample =P <<< 2W Nyquist Sampling Compressed Sensing Sampling Innovation Rate Sampling Signal Information Rate Feature Sampling Application Feature Rate (e) Figure S2: Comparing sampling architectures: (a) Nyquist rate sampling, (b) compressed sensing sampling, (c) innovation rate sampling, and (d) feature sampling using analog analytics. (e) Evolution of the physical bandwidth along the signal chain for the architectures in (a)-(d). IEEE SOLID-STATE CIRCUITS MAGAZINE su m m e r 2 0 15 69