Image compression and energy harvesting for energy constrained sensors
Metadata[+] Show full item record
The advances in complementary metal-oxide-semiconductor (CMOS) technology have led to the integration of all components of electronic system into a single integrated circuit. Ultra-low power circuit techniques have reduced the power consumption of circuits. Moreover, solar cells with improved efficiency can be integrated on chip to harvest energy from sunlight. As a result of all the above, a new class of miniaturized electronic systems known as self-powered system on a chip has emerged. There is an increasing research interest in the area of self-powered devices which provide cost-effective solutions especially when these devices are used in the areas that changing or replacing batteries is too costly. Therefore, image compression and energy harvesting are studied in this dissertation. The integration of energy harvesting, image compression, and an image sensor on the same chip provides the energy source to charge a battery, reduces the data rate, and improves the performance of wireless image sensors. Integrated circuits of image compression, solar energy harvesting, and image sensors are studied, designed, and analyzed in this work. In this dissertation, a hybrid image sensor that can perform the tasks of sensing and energy harvesting is presented. Photodiodes of hybrid image sensor can be programmed as image sensors or energy harvesting cells. The hybrid image sensor can harvest energy in between frames, in sleep mode, and even when it is taking images. When sensing images and harvesting energy are both needed at the same time, some pixels have to work as sensing pixels, and the others have to work as solar cells. Since some pixels are devoted to harvest energy, the resolution of the image will be reduced. To preserve the resolution or to keep the fair resolution when a lot of energy collection is needed, image reconstruction algorithms and compressive sensing theory provide solutions to achieve a good image quality. On the other hand, when the battery has enough charge, image compression comes into the picture. Multiresolution decomposition image compression provides a way to compress image data in order to reduce the energy need from data transmission. The solution provided in this dissertation not only harvests energy but also saves energy resulting long lasting wireless sensors. The problem was first studied at the system level to identify the best system-level configuration which was then implemented on silicon. As a proof of concept, a 32 x 32 array of hybrid image sensor, a 32 x 32 array of image sensor with multiresolution decomposition compression, and a compressive sensing converter have been designed and fabricated in a standard 0.5 [micrometer] CMOS process. Printed circuit broads also have been designed to test and verify the proposed and fabricated chips. VHDL and Matlab codes were written to generate the proper signals to control, and read out data from chips. Image processing and recovery were carried out in Matlab. DC-DC converters were designed to boost the inherently low voltage output of the photodiodes. The DC-DC converter has also been improved to increase the efficiency of power transformation.
Table of Contents
Introduction -- Hybrid imager system and circuit design -- Hybrid imager energy harvesting and image acquisition results and discussion -- Detailed description and mathematical analysis for a circuit of energy harvesting using on-chip solar cells -- Multiresolution decomposition for lossless and near-lossless compression -- An incremental [sigma-delta] converter for compressive sensing -- Detailed description of a sigma-delta random demodulator converter architecture for compressive sensing applications -- Conclusion -- Appendix A. Chip pin-out -- Appendix B. Schematics -- Appendix C. Pictures of custom PCB