Skip to Main content Skip to Navigation

Conception et intégration d'un convertisseur analogique-paramètres flexible pour les capteurs intelligents

Abstract : The Internet of Things (IoT) is currently experiencing huge developments. IoT includes lots of different devices such as Wireless Sensors Networks (WSN) or wearable electronics that rely on wireless communications. These networks need to understand the context in which they are used. This mean that the system must know what is happening around it, i.e. sense the environment, and understands the needs of the user. This requires always-on sensing on many sensors while being small, cheap, reliable and having a lifetime of several years. Analog-to-Feature (A2F) conversion is a new acquisition method that was thought for IoT devices. The converter aims at extracting useful features directly on the analog signal. By carefully choosing a set of features, it is possible to acquire only the relevant information for a given task. The proposed converter is based on the Non-Uniform Wavelet Sampling (NUWS) architecture. The architecture mixes the analog signal with tunable wavelets prior to integration and digital conversion. The aim of the thesis is to propose a method to design a generic A2F converter based on the NUWS. It includes the definition of the wavelet parameters in order to acquire a broad range of low frequency signals (ECG, EMG, EEG, speech …). This step requires the use of feature selection algorithms and machine learning algorithms for selecting the best set of wavelets for a given application and should be used to define the specifications for the converter. The feature selection step must be aware of physical implementation constraints to optimize energy consumption as much as possible. A feature selection algorithm is proposed to choose wavelets for a given application, in order to maximize classification accuracy while decreasing power consumption, through a power model designed in CMOS 0.18μm.
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Wednesday, March 30, 2022 - 11:17:14 AM
Last modification on : Thursday, March 31, 2022 - 3:09:29 AM


Version validated by the jury (STAR)


  • HAL Id : tel-03624373, version 1


Antoine Back. Conception et intégration d'un convertisseur analogique-paramètres flexible pour les capteurs intelligents. Micro et nanotechnologies/Microélectronique. Institut Polytechnique de Paris, 2022. Français. ⟨NNT : 2022IPPAT020⟩. ⟨tel-03624373⟩



Record views


Files downloads