Embedded Lossless and Near-Lossless Real-Time Compression of sEMG Signals
keywords EMBEDDED SIGNAL PROCESSING, FIRMWARE AND SOFTWARE DEVELOPMENT, NEAR-LOSSLESS AND LOSSLESS COMPRESSION, REAL-TIME COMPRESSION, SURFACE ELECTROMYOGRAPHY, WEARABLE AND IOT DEVICES, WIRELESS BODY AREA NETWORK (WBAN)
Reference persons DANILO DEMARCHI
External reference persons Paolo Motto Ros (firstname.lastname@example.org)
Research Groups MiNES (Micro&Nano Electronic Systems)
Thesis type EXPERIMENTAL
Description Optimal data handling in bio-signal acquisition wearable (IoT) systems is still an open issue, heavily depending on the final, specific, application. From a medical diagnostic point of view, for example, ideally no information should be lost, but this requirement can be challenging for the data storage and communication subsystems. On the opposite side, a (highly) lossy compression/information synthesis approach could greatly relax the constraints on the power consumption and hardware resources, at the expense of significantly lowering the average performance in terms of “data quality”.
Major aim of this thesis is to design and implement a lossless and near-lossless (i.e., exactly directly controlling the maximum, and not the average or RMS, error) compression algorithm that sits between the two briefly mentioned above common options. Starting from the investigation of both compression algorithms and surface ElectroMyoGraphy (sEMG) acquisition/processing, through the design and development of signal processing/compression algorithms in the C programming language, the final goal is the implementation of the algorithm, operating in real-time, in a wearable embedded system, featuring a sEMG Analog Front-End (AFE) subsytem, an ARM Cortex-M4F MCU, an external Flash memory for data storage, and a Bluetooth Low-Energy (BLE) wireless transceiver.
Required skills C programming; MCU firmware development; Software development
Deadline 30/09/2022 PROPONI LA TUA CANDIDATURA