Demonstrators

Use Case Leader: AED

Involved Partners: AED, Soitec, Fraunhofer

Dedicated to the development of a multi-microphone system mounted in a vehicle based on ultra-low power microphones and a computing platform developed with SOI technology and provided by SOITEC (Dolphin Design).

Two sets of microphones will be mounted on a vehicle, one internally and one externally, with similar functionalities but different aims. Thanks to the use of a centralized computing device with an AI core, implemented by the FDSOI based chip developed by Dolphin, it will be possible to complete the following aims. Concerning the microphones network implemented externally:

• Identify the speaking voice over the noise and not relevant sounds
• Identify spoken commands
• Recognize the registered users, with biometrical footprint
• Recognize the direction of the incoming voice, being compared with the information coming from a localization UWB system as reference

The microphone network implemented internally has the following objectives:

• Seat occupancy identification
• Identify spoken commands

In parallel to the microphones system, a localization system based on Ultra-Wide Band (UWB) technology will be included in the overall setup to have a reference for the position of the user. As the future of car access is based on UWB and BLE localization systems, able to identify the position of a smartphone used as car key, the microphone demonstrator will be fused with the UWB/BLE localization system.

The applications that will be developed for the microphones positioned outside the car include:

• Beamforming to recognize the direction of the voice
• Voice recognition (Voice ID)
• Commandos recognition (Open trunk, open door, lower windows…)
• Car access system through sensor fusion with UWB and BLE
• The commandos may be accepted only if the direction of the voice corresponds to the phone holder position
• Either always ON microphones (ultra-low power microphones) or triggered by BLE wake-up (CCC compliant)

Considering the microphones placed inside the car the following applications will be developed:

• Seat occupation recognition
• Commandos recognition (open window, temperature set, music control…)
• Noise filtering to recognize commandos over the music or children playing etc.

The system will be build including a 10BASE-T1S network and comparing it to the CANXL interface. The ethernet based connection network is a driving technology that is predicted to replace, starting partially and progressively, the CAN Bus in future vehicles. Additionally for external microphones it will be necessary to consider mechanical protection from external damaging sources (e.g. high-pressure water).

Use Case Leader: AED

Involved Partners: AED, Citrobits, ST

Dedicated to implementing a zonal architecture in an automotive environment that includes different data transfer technologies and cutting-edge devices allowing new applications foreseen in the future of the automotive market, such as autonomous driving and smart access systems.

Different communication protocols will be included, 10BASE-T1S, MGBASE-T1 and ASA-ML. One of the key achievements that will be shown by this demonstrator is the (successful implementation) use of the industry’s first standardized, high-speed, asymmetric wireline communication protocol. The achievement is so critical because this wireline protocol is born out of a European ecosystem but is already being adopted by automotive companies all over the world. As automobiles become more connected and the trends toward software-defined vehicles and autonomous driving accelerate, such a standard is becoming even more critical.

Recently, efforts have been undertaken to develop an industry standard high-speed, asymmetric wireline communication standard for automobiles by the Automotive SerDes Alliance (ASA). ASA was established in 2019 by BMW, Continental, Broadcom, Fraunhofer IIS and NXP as founding members. In the short amount of time since its inception, ASA has been growing rapidly and now has more than 130 active member companies. In December 2020, ASA released the “ASA Motion Link” transceiver specification v1.01, which defines a SerDes communication technology for in-vehicle connectivity ranging from 2Gbps to 16Gbps line rate. Key features of ASA Motion Link include:

• Interoperable Asymmetrical SerDes standard including physical, data link & transport layers
• Up to 15m Coaxial and 10m SDP channels with up to 4 in-line connectors
• Downlink line rates up to 16Gbps
• Uplink data rates greater than 100Mbps
• Deterministic delivery framework
• Comprehensive security framework including an efficient key exchange mechanism
• Application Stream Encapsulation Protocol (ASEPs) for Video, I2C, Ethernet packets
• Robust EMC immunity and EMI emission profiles

AED, as system integrator, performs the following tasks:

• Perform the integration of the communication chips, including the development of the PCB(s), external FPGA interfacing and necessary software
• Development of a camera module based on commercially available imager and ASA-ML PHY
• Development of a Display Module
• Development of an ECU with ASA-ML interface and an FPGA for data transmission to automotive ethernet
• Improvement of TSN switch module to be included as gateway
• Support in the installation and testing of the system in a car

Use Case Leader: Bosch

Involved Partners: Bosch, TU/e, AntenneX, NAM, GreenWaves

The automotive wide-band radar device is a radar system that observes the entire vehicle for safety and comfort. The device detects and classifies living beings, objects and vehicles around a vehicle.

As a system integrator, the Bosch company has a unique perspective on how FDSOI technology can be applied in real-world applications and combined with other technologies to create complete systems. This expertise allows the company to develop and support system-level demonstrator developments that showcase the benefits and capabilities of FDSOI technology in a practical context, while ensuring that the systems are reliable, efficient, and meet the stringent requirements of real-world applications.

In this task, therefore, Bosch will be in charge of the management of the task (Sourcing of components, time planning of component delivery, validation of shipped components, Demonstrator assembly and functional tests). The main purpose is to emphasize the high capacity of the chip (developed in WP4) by addressing key customers’ requirements listed above. Bosch will create also the link between the demonstrator performances and the AntenneX characterisation results.

TU/e will be in charge of the design and development of a (distributed) antenna array system to be integrated in the system demonstrator.

NAM will be focusing on the design and development of the package

AntenneX will be working on the design and development of easy-to-use Over The Air (OTA) test facility measurement equipment. AntenneX will develop custom test facility for validation of integrated sub-systems and systems (software-defined antenna systems) according to the defined test requirements in WP1. An uncertainty framework will be created for all relevant metrics to be measured. AntenneX will support the actual antenna & device materials measurements.

GreenWaves will supply of GAP10 microcontroller, in 22nm FDSOI technology, for near sensor computing and AI inference. Greenwaves will be also porting and optimising the digital processing and neural networks required for the target use cases, using a dedicated tool suite.

Use Case Leader: STM

The automotive wide-band radar device is a radar system that observes the entire vehicle for safety and comfort. The device detects and classifies living beings, objects and vehicles in and around the vehicle.

To comply with the EU New Car Assessment Programme (NCAP), all new vehicles must have an indirect sensing system from 2023. A direct sensing system is required from 2025 onwards.

The National Highway Traffic Safety Administration (NHTSA) of the United States prepares a Child Presence Detection (CPD) system in all new vehicles from 2026 onwards.

The automotive in-cabin monitoring device targets the following use cases:

• Child Presence Detection and Classification (adult vs. child)
• Occupancy Detection (adults, children, babies)
• Intrusion Alert in vehicle interior
• Vital sign monitoring (life presence detection of humans and pets, respiration rate)

Furthermore, similar mechanisms are used around the vehicle for the following use cases:

• Kick sensing (to open the booth with a kick gesture when both hands are busy)
• Keyless entry (where ultra-wideband signal is used to measure securely the distance between the car and the person wearing / carrying the digital key holding device)

As a supplier of the major components involved in car access, ST microelectronics is investing a lot in UWB systems (on top of NFC controllers and Secure Elements). Leveraging these assets to provide services that will enhance the driver experience and increase the safety of its passenger is a strategic interest for the company.

Thanks to a holistic approach of radio usage, ST will be able to demonstrate how UWB can offer radar services, while maintaining a perfect co-existence with other services sharing the same radio paradigm (for example for ranging) and the same components (to demonstrate that additional value can be generated at no extra cost).

Key characteristics of FDSOI will add to the experience by improving the sensitivity of the radar and enhance the resolution of the vital signs to a cabin where multiple passengers may be sitting in various configurations (eg on 2 raws of seats).

The additional processing power will also allow to run real-time signal processing algorithms to achieve better results. In particular, fast fourrier transforms and IA-accelerated vector processing will enable state of the art radar sensing performances.

In this task, therefore, ST microelectronics will be in charge of the management of the task (Sourcing of components, time planning of component delivery, validation of shipped components, Demonstrator assembly and functional tests).

Use Case Leader: UTIA

Involved Partners: UTIA, STM, GreenWaves, Citrobits, IMA

Dedicated to development of active noise cancellation of acoustic signal sampled by digital microphone. Active acoustic noise cancellation system will be implemented on low power 28 FDSOI device on GP DPU unit in floating point and also in fixed point SW for multiple controllers on single 22 FDSOI device.

SIMD GP DPU (FP32/FP16) developed in WP3 will be integrated in SoC system on low power 28 FDSOI programmable logic device CertusPro-NX. Support for two microphone inputs will be developed and implemented as PCB nodule with open CRUVI standard. Filtered acoustic output will be provided as fixed line Ethernet stream and as analogue output on CRUVI PCB module. The implemented adaptive RLS filter will require DPU implementation with at last 1 GFLOP (FP32/FP16) performance.

Fixed point version of normalised adaptive RLS filter for active noise cancellation will be also ported and optimized for the GreenWaves’ GAP10 microcontroller (22nm FDSOI technology), establishing a reference point for a programmable ASIC implementation, and enabling comparison of its performances and benefits compared to solutions implemented in programmable logic.

STM will provide data for comparison of MFLOPs/W reached on the low power 28 FDSOI demonstrator with SW implementation on the 40 nm CMOS ST devices (STM32H7, STM32H5).

GreenWaves will provide the 22nm FDSOI GAP10 microcontroller and associated tools to port, optimize and validate complex digital processing and support in porting of demonstration application onto GAP10, and performance assessment.

UTIA will work on integration of DPU on the low power 28 FDSOI programmable logic device, development and integration of microphone input and output on CRUVI PCBs, implementation of LMS echo cancellation algorithm in FP32 and FP16 arithmetic and Implementation of Ethernet.

Citrobits will design PCB module with small footprint (of ~20x20mm) with CertusPro-NX device and a carrier board with CRUVI connectors, that will drastically reduce the cost of the demonstrator and will allow scaling to multiple units if needed. As Lattice Partners, Citrobits has deep understanding of Lattice technology and toolset.

IMA will provide testing and user feedback to UTIA and GreenWaves related to use in automotive (in-cabin active noise cancelation).

Use Case Leader: IMA

Involved Partners: UTIA, STM

Dedicated to the development of a deep edge sensor data processing unit with low power consumption and efficient computational power combining the traditional deterministic sensor data processing algorithms with HW accelerated AI based algorithms. Demonstrator is aiming on the Automotive sector use-cases mainly. The data processing is supposed to be performed as close to the sensor as possible to avoid high data traffic. The target is to provide only pre-processed condensed data to the domain controller and to perform basic decision making on the edge to conserve energy and progress end-point autonomy.

IMA will exploit the benefits of embedded Flash memory with fast and energy efficient low-leakage SRAM memory enabled by advanced FDSOI technology. The FP16 and FP32 acceleration with SIMD processing unit architecture will be exploited in order to benefit from the fast and efficient data processing.

The HW component base for the Edge processing unit demonstrator will be based on a platform developed in WP4. The platform will be centred around a low-power MCU with efficient FP32 MAC (Multiply-Accumulate) capabilities for the AI inference algorithms to be performed on a sensor data in real-time. The unit architecture will be designed followed by the development of the HW platform based on selected core computational component - MCU. As the multi-layer NN architectures require larger memory size, powerful MCU including very efficient power management, we intend to integrate new SOIL FDSOI MCU components for the final demonstrator.

The target is to provide only pre-processed condensed sensor data to the domain controller only when needed and to perform basic decision making on the edge to conserve energy and progress end-point autonomy. Fast and efficient switching to the low power sleep mode is also one of the key points for long term low power operation. Demonstrator integrated in WP5 is first aiming on the Automotive sector use-cases. In a second step the usage of developed modules and IPs will be considered for access control for critical infrastructure use-case.

Use Case Leader: ISD

Involved Partners: AED, ST

Dedicated to the demonstration of the capability of the HPDP80 processor which is theoretically 12x more performant than the current HPDP40 solution. On real silicon and associated evaluation board, ISD will update the associated SW development tool kit and will port and optimize the S/W on HPDP80 target. The different modules for encoding hyperspectral pictures with CCSDS 123 are:

• Pre-processing,
• Spectral decorrelation,
• Spectral decorrelation,
• Quantization,
• Entropy coding,
• Header generation.

All these SW modules will have to be optimized on the new HPDP80 with the updated SW development tool kit.

Use Case Leader: CEA-LETI

Involved Partners: AED, STM

Dedicated to the development of a so-called “zero-energy” communication demonstrator compatible with the existing 5G cellular communication network. The aim is to promote the development of Green-IoT in the future 6G communication networks and to facilitate the development of the new application. It will be also key to promote the capability of SOI technologies to reduce the power consumption of communicating systems.

The following picture gives a brief description of the demonstrator functionality and context. The TAG developed in the scope of the SOIL project is depicted by the grey form surrounded in red. It contains some information (depicted as a gift) stored (message-security) or collected by some sensor.

This information can be either collected by the base station when the mobile phone is transmitting some message to the base station or it can be collected by the mobile phone when it is receiving a message. In none of these configurations the TAG is generated RF power, as a consequence, the TAG is almost passive and can be only powered by a solar panel.

The concept of crowd-detectable zero-energy-device (CD ZED), based on ambient backscatters in mobile networks is under standardisation process by Orange and in under development for 6G. These results will contribute also to the standardisation

The main objective of the demonstrator is to reach a TRL between 5 to 6. It will be a demonstration prototype, which will be operating in real environment. The objective will be to evaluate the interest of the communication technology in different use cases

CEA-LETI will be in charge of providing the demonstrator specifications, to develop the overall system architecture and to design the board. The board will be sub-contracted. AED will review the specifications as well as the architecture and board. AED will also proceed with the design of the package and will be in charge of the assembling and packaging of the overall TAG. The validation of the demonstrator will be operated by CEA-LETI in collaboration with external partners. STM will review the specification and consider potential interaction with the developed micro-controller.