Reliable health monitoring: a commercial off-the-shelf and a field programmable hardware approach

Commercial off-the-shelf (COTS);Field-Programmable Gate Array (FPGA);Electroencephalography (EEG);Seizure;Microcontroller;MSP430;Leon3;Advanced Microcontroller Bus Architecture (AMBA)

Cheriyan, Ajay M. ; Iyer ; Ravishankar K.

关键词: Commercial off-the-shelf (COTS);    Field-Programmable Gate Array (FPGA);    Electroencephalography (EEG);    Seizure;    Microcontroller;    MSP430;    Leon3;    Advanced Microcontroller Bus Architecture (AMBA);   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/16175/1_cheriyan_ajay.pdf?sequence=2&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

With the tremendous advancements in low cost, power-efficient hardware and the recentinterest in biomedical embedded systems, numerous traditional biomedical systems can bereplaced with smaller and faster embedded systems that perform real-time analysis to providebio-feedback to the users. This thesis takes a look at two hardware implementations – one using commercial off-the-shelf (COTS) components and the other using field programmable logic.The focus of the design was to ensure a portable, inexpensive, power-efficient and robustdevice that could perform analysis of physiological signals, which would in turn help alert theuser in the event of an abnormality. The COTS hardware implementation provided theframework using a microcontroller as the processing element for a reliable health monitoringdevice with a seizure detection directly embedded in it.The field programmable gate array (FPGA) platform based implementation was proposedand simulated to overcome the two disadvantages of the COTS approach – the inability tosupport customization of the device to suit the end-user’s monitoring requirements and complexdetection schemes requiring significant processing capability. The FPGA platform was simulatedfirst as a standalone module and later as part of an SoC design. The novel algorithm included afeature extraction phase and a machine learning based seizure detection phase. Simulation basedtesting of the device showed a detection accuracy of 99.2 %.