Compressed and Deep Sensing: Models, Structures, and Tasks

Información del proyecto

CoDeS

Identificador del acuerdo de subvención: 101000967

DOI

10.3030/101000967

Fecha de la firma de la CE 30 Abril 2021

Fecha de inicio 1 Julio 2021

Fecha de finalización 30 Junio 2026

Financiado con arreglo a

• EXCELLENT SCIENCE - European Research Council (ERC)

Coste total

€ 3 000 000,00

Aportación de la UE

€ 3 000 000,00

Coordinado por

WEIZMANN INSTITUTE OF SCIENCE
Israel

Periodic Reporting for period 2 - CoDeS (Compressed and Deep Sensing: Models, Structures, and Tasks)

Período documentado: 2023-01-01 hasta 2024-06-30

Resumen del contexto y de los objetivos generales del proyecto

In this project we are exploring how we can build systems that are far more efficient than those existing today by designing them in an end-to-end holistic fashion. Throughout their design we take into account ay structure in the input or system, and we also consider the final system task. We have been exploring ways in which this approach can make various types of systems more efficient including communication systems, radar, medical devices and more. In all of these areas we consider the basic question: Can we measure only what is necessary to perform our desired system task by jointly optimizing the entire signal acquisition and processing chain?

Trabajo realizado desde el comienzo del proyecto hasta el final del período abarcado por el informe y los principales resultados hasta la fecha

In our work so far we have been developing fundamental trade-offs between sampling rate, quantization resolution, and task fidelity. We have also developed new results on the bounds of model-based deep learning that takes models into account within the learning process.
On the engineering side we have developed several new ADC devise based on this principle like modulo-based sampling that can accommodate high dynamic range, and time-based sampling that allows removing the system clock all together. On the application side we have been considering 6G networks, joint sensing and communication systems, efficient vital sign monitoring, ECG monitoring and more.

Avances que van más allá del estado de la técnica e impacto potencial esperado (incluida la repercusión socioeconómica y las implicaciones sociales más amplias del proyecto hasta la fecha)

On the engineering side we have developed several new ADC devise based on this principle like modulo-based sampling that can accommodate high dynamic range, and time-based sampling that allows removing the system clock all together. On the application side we have been considering 6G networks, joint sensing and communication systems, efficient vital sign monitoring, ECG monitoring and more.