A Beat Signal Processing System with Parabolic Frequency Chirp Radar and Computation-in-Memory

A novel beat signal processing system to overcome the detection distance limitation due to sampling frequency is proposed. The detection range of conventional FMCW radar is limited not only by signal-to-noise ratio of received signal but also by Nyquist frequency of beat signal sampling. The proposed system architecture exploits parabolic frequency chirp for radar signal and under-sampling technology for beat signal processing. To process the beat signal, cross-correlation base signal processing accelerated by Computation-in-Memory (CiM) is proposed. CiM can reduce calculation latency and power consumption compared to conventional CPU base computing. Although CiM is affected by non-idealities of memory device such as quantized conductance and conductance variation, the simulation results indicate that the proposed system shows capability to detect both near target at 50m and far target at 200m with 10MHz sampling rate at the same time without any transition of chirp rate.