Correlators are fundamental building blocks in radar/communication signal processing and analog-to-information (A-to-I) applications such as spectrum sensing [1]. Typically, correlation, which is equivalent to an inner (dot) product, is performed using digital multiply-and-accumulate (MAC) operations (Fig. 12.3.1), with power consumption scaling with frequency, compute, and ADC power. Analog correlators eliminate ADCs but suffer from high power/area and/or small correlation lengths (~10 samples) [2], [3]. While matrix multiplication in compute-in-memory (CiM) cores can be used for correlation, such blocks typically operate with lower speeds for multi-bit inputs, require new materials/memory IP and DACs and ADCs at the input/output [4], [5].
No takes yet. Share an insight, caveat, or question.
Rashed et al. (2024) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: