PhD thesis on a low power "time-to-first-spike" event sensor

Title: Event-based Image Sensor for low-power

Author: Mohamed AKRARAI (Universite Grenoble Alpes)

Abstract: In the framework of the OCEAN 12 European project, this PhD achieved the design, the implementation, the testing of an event based image sensor, and the publication of several scientific papers in international conferences, including renowned ones like the International Symposium on Asynchronous Circuits and Systems (ASYNC). The design of event-based image sensors, which are frameless, require a dedicated architecture and an asynchronous logic reacting to events. First, this PhD gives an overview of architectures based on a hybrid pixel matrix including TFS and DVS pixels. Indeed, this two kind of pixels are able to manage the spatial redundancy and the temporal redundancy respectively. One of the main achievement of this work is to take advantage of having both pixels inside an imager in order to reduce its output bitstream and its power consumption. Then, the design of the pixels and readout in FDSOI 28 nm technology from STMicroelectronics is detailed. Finally, two image sensors have been implemented in a testchip and tested.

Link: https://theses.hal.science/tel-04213080v1/file/AKRARAI_2023_archivage.pdf