Compact low power CMOS radar features on-chip antennas

Radars are extremely promising as sensors for contactless, non-intrusive interaction in IoT applications such as people detection and classification, vital signs monitoring and gesture interfacing.

A wide adoption will only be possible if radars achieve a higher resolution, become much smaller, more power efficient to run, and cheaper to produce and to buy. This is what imec’s research on 140GHz radar technology targets.

This low power 140GHz radar solution comprises an imec proprietary two antenna SISO (Single Input Single Output) radar transceiver chip and a frequency modulated continuous wave phase-locked loop (FMCW PLL), off-the shelf ADCs and FPGA and a Matlab chain.

The transceiver features on-chip antennas achieving a gain close to 3dBi. The radar link budgets are supported thanks to the transmitter Effective Isotropic Radiated Power (EIRP)  that exceeds 9dBm and a receiver noise figure below 6.4dB. The total power consumption for transmitter and receiver remains below 500mW, which can be further reduced by duty cycling.

The FMCW PLL enables fast slopes up to 500MHz/s over a 10GHz bandwidth around 140GHz with a slope linearity error below 0.5% and has a power consumption below 50mW. The FPGA contains real time implementation of basic radar processing functions such as FFTs (Fast Fourier Transforms) and filters, and is complemented by a Matlab chain for detections, CFAR (Constant False Alarm Rate), direction-of-arrival estimation and other advanced radar processing.

“With our prototype radar, we have demonstrated all critical specs for radar technology in 28nm standard CMOS technology,” said Wim Van Thillo, IoT Programme Director at imec.

“We are well advanced in incorporating multiple antenna paths in our most recent generation solution, which will enable a fine angular resolution of 1.5cm in a complete MIMO radar form factor of only a few square centimeters. We expect this prototype in the lab by the end of 2018, at which point our partners can start building their application demonstrators.

"First applications are expected to be person detection and classification for smart buildings, remote car driver vital signs monitoring (as cars evolve towards self-driving vehicles), and gesture recognition for intuitive man-machine interactions. Plenty more innovations will be enabled by this technology, once app developers start working with it.”