Continuous-Time DSP Using Configurable Devices

GIRD Systems, Inc., and EchoICs propose to design and build a compact low-SWaPC Spectrum Situational Awareness (SSA) package for small swarm UAVs using continuous-time digital signal processing (CT-DSP) techniques. This design will integrate a custom integrated circuit pipelined level-crossing ADC, CT-DSP spectrum analysis algorithms in reconfigurable logic, and machine learning / neural network (NN) classifier.ĀCT-DSP reformulates conventional sampled DSP to operate on continuous-time digital signals.Ā The input analog signal is quantized by a level-crossing ADC, which produces a continuous time-varying multi-bit digital amplitude.Ā Our objective is a 9-bit quantizer with at least 1 GHz bandwidth.Ā This digital signal can be processed in various forms, including unit-distance Gray code, delta signals, or place-value binary.ĀFourier and Walsh transforms are used to project the continuous-time waveforms into a frequency space by using a tapped delay line to provide current and past values of the waveform.ĀBecause of the linearity of the transforms, computations can be simplified by processing each place-value bit as a separate continuous waveform, and combining the partial results with power-of-two weights to produce the net result.Ā All these actions are conducted in continuous time using specially-designed logic components.ĀĀClassification of the electromagnetic environment is performed with an artificial neural network operating in continuous time on the outputs of the spectrum analysis algorithms.ĀThe structure of the NN leverages the fundamentally continuous time nature of the lattice, which performs scaling and adding multiple signals, and combining them through activation functions.ĀThe result is made available to a discrete-time consumer, nominally a supervisor processor in the system.ĀThe design is expected to produce a lower SWaPC cost for performance that will be a technological leap over purely discrete time solutions, and maintain the tactical advantage over adversaries.