Integrated Radar are widely used today, with a specific focus on automotive applications, but with new use-cases foreseen in industrial context with shuttles and unmanned aerial vehicles/ automated guided vehicles (UAV/UGV) operated autonomously. Current Radar solutions mainly use frequency modulated continuous wave (FMCW) modulation scheme that enables a very efficient and easy-to-implement detection as the distance and speed of the targets stands in the frequency difference between the transmitted and received signals. This enables the use of low bandwidth and high-resolution ADCs.

In the past, 24GHz was used, and it now turns to be the 77GHz-79GHz band, and then the 140 GHz in long-term perspective. This will improve sensor integration, resolution in distance, speed, azimuth and elevation. As a matter of fact, this massive use in a given band also questions the coexistence issues and its relevance in even-more connected mobility environment.

One of the important figure of merit an efficient Radar system lays in its ability in detecting near targets as well as far ones (near-far problem), as this requires a very high dynamic capability in the receiver, in other words discriminating low level signals (far and/or from small targets) from high level ones (near and/or from large targets).

The other one is related to the necessary digital signal processing that is constrained by the size of the raw data flow, the algorithmic complexity and needed operational memory.

The proposed architecture involves a frequency modulated pulsed waveform that combines the FMCW waveform used in numerous automotive Radar systems and an impulse radio-ultra wide band (IR-UWB) waveform used in localization systems. This combined waveform is processed on the receiver side using a parallel acquisition receiver allowing an overall wideband capture and narrower frequency band selection converted on each ADC. This improves greatly the flexibility by its ability to operate in dual-mode, interleaving or binary orthogonal base projection.

The proposed implementation mixes RF and analog front-end together with highly digital process and reconfigurability. 16 lanes in parallel over 2 paths requires to tightly consider the physical implementation and associated surface of the IC and assess its impact. Using an FD-SOI technology is thereby of interest thanks to the high-level performance, its high usable frequency for digitally oriented architectures, scalability of digital processing, extra trimming capability and finally low power consumption. As a baseline, GF 22FDX will demonstrate FD-SOI capacity, and further considerations on 10nm FDSOI will be assessed.     

Demonstrating FD-SOI advances for such a mixed-signal approach compared to bulk technology is a proof on how RF applications can benefit from a low power, high density CMOS technology. The proposed modulated scheme and its implemented receiver architecture will set a new state-of-the-art by employing a new approach for the (de)modulation scheme. This shall allow a better handling of the near-far problem, and in addition, improve the co-existence with other devices thanks to an associated frequency-hopping (FH) mechanism. The implementation of orthogonal codes on the frequency basis is also helpful for sub-bands selection and filtering. Finally, time-interleaved process is applied on the converted signal to recover the full information. Such advances will support FAMES in demonstrating the interest of highly dense CMOS implementation.

The potential use-cases of this Radar systems will include automotive, but also any kind of environment scanning thanks to its flexibility due to the highly digital-oriented architecture, improving the robustness to the near-far problem and avoiding collisions with different signals operated on the same bands. Implementation of the single-chip on small area of FD-SOI technology will be essential for further massive antenna arrays enabling beam-forming.

access to equipment, collaboration with new partners, visibility, funding, expertise sharing, etc. …

The preliminary implementation of this innovative Radar architecture is done on GF 22FDX technology. It paves the way to future consideration on 10nm FD-SOI. Indeed, it is expected to use the proposed IC connected to a patch antenna. Considering that a Radar solution needs to be able to operate beam-forming to scan a complete scene, it is expected that each IC can be very close to its antenna patch, thereby requiring compacity in order to comply with a full 16 Tx/Rx or 32 Tx/Rx system, especially when considering operation in D-band/140GHz for which large antenna arrays will be developed. As part of the digital process will be on the IC, reducing transistor size from 22nm to 10nm will help reducing the IC size and its ability to comply with the expected surface.

This implementation of frequency modulated pulsed waveform establishes a new state-of-the-art and demonstrates the interest of low-power, reduced transistor technology node size for future technology providing with a high frequency operation up to D-band / 140GHz, dense implementation and low power consumption.

The full transceiver including the publication described receiver and its associated transmitter will be taped out enabling the validation of the full chain, from signal generation and transmission to receiver amplification, filtering, and processing.