Digital video processing demands have and will continue to grow at unprecedented rates. Growth comes from ever increasing volume of data, demand for higher resolution, higher frame rates, and the need for high capacity communications. Moreover, economic realities force continued reductions in size, weight and power requirements. The ever-changing needs and complexities associated with effective video processing systems leads to the consideration of dynamically reconfigurable systems. The goal of this dissertation research was to develop and demonstrate the viability of integrated parallel processing system that effectively and efficiently apply pre-optimized hardware cores for processing video streamed data. Digital video is decomposed into packets which are then distributed over a group of parallel video processing cores. Real time processing requires an effective task scheduler that distributes video packets efficiently to any of the reconfigurable distributed processing nodes across the framework, with the nodes running on FPGA reconfigurable logic in an inherently Virtual' mode. The developed framework, coupled with the use of hardware techniques for dynamic processing optimization achieves an optimal cost/power/performance realization for video processing applications. The system is evaluated by testing processor utilization relative to I/O bandwidth and algorithm latency using a separable 2-D FIR filtering system, and a dynamic pixel processor. For these applications, the system can achieve performance of hundreds of 640x480 video frames per second across an eight lane Gen I PCIe bus. Overall, optimal performance is achieved in the sense that video data is processed at the maximum possible rate that can be streamed through the processing cores. This performance, coupled with inherent ability to dynamically add new algorithms to the described dynamically reconfigurable distributed processing framework, creates new opportunities for realizable and economic hardware virtualization.'
FPGA, Dynamic Partial Reconfiguration, DPR, Video Processing, FPGA Framework, Reconfigurable Computing, Hardware Virtualization, Virtualization, Firmware Virtualization
Level of Degree
Electrical and Computer Engineering
First Committee Member (Chair)
Second Committee Member
Third Committee Member
Hoffman, John. "A Dynamically Reconfigurable Parallel Processing Framework with Application to High-Performance Video Processing." (2013). https://digitalrepository.unm.edu/ece_etds/119