A Scalable and Instantaneously Wideband RF Correlator Based on Margin Computing

Kareem Rashed, Aswin Undavalli, Shantanu Chakrabartty, Aravind Nagulu, Arun Natarajan

Research output: Contribution to journalArticlepeer-review

Abstract

Correlation is a fundamental operation in radar/ communication signal processing. Enabling efficient reconfigurable correlation at high frequencies with wide bandwidths and high dynamic range is a fundamental challenge. This article presents a direct-RF wideband analog correlator that utilizes a novel margin computation paradigm replacing traditional multiply-and-accumulate (MAC) with analog addition and thresholding to enable energy-efficient analog computation. A high-efficiency charge-domain realization is proposed and implemented in 65-nm CMOS occupying 0.97 mm2. The analog correlator IC supports 5-GS/s inputs, a large correlation length of 1024, and 8-bit computing accuracy with a high energy efficiency of 152 TOPS/W. Practical applications of typical high-speed correlation are also demonstrated through system-level measurements such as radar signal detection, and code-domain processing.

Original languageEnglish
Pages (from-to)3612-3626
Number of pages15
JournalIEEE Journal of Solid-State Circuits
Volume59
Issue number11
DOIs
StatePublished - 2024

Keywords

  • Analog computing
  • approximate computing
  • code-domain radars
  • correlation
  • inner-product
  • multiplier-free
  • RF sensing
  • spread spectrum communication

Fingerprint

Dive into the research topics of 'A Scalable and Instantaneously Wideband RF Correlator Based on Margin Computing'. Together they form a unique fingerprint.

Cite this