Towards real time implementation of reconstructive signal processing algorithms using systolic arrays coprocessors

Reconstructive signal processing algorithms encompass a broad spectrum of computational methods. Fortunately, most of the methods fall into the classes of the matrix algebraic calculations, convolution, or transform type algorithms. These algorithms possess common properties such as regularity, locality and recursiveness. Considering such general class of reconstructive signal processing (SP) techniques, in this paper we propose a new Hardware/Software (HW/SW) co-design paradigm for the implementation of reconstructive SP algorithms via efficient systolic arrays integrated as digital SP coprocessors units. In particular, the selected matrix–matrix and matrix–vector multiplication algorithms are implemented in a systolic computing fashion that meets the real time SP system requirements when employing the developed Hardware/Software Co-Design method oriented at the use of a Xilinx Field Programmable Gate Array (FPGA) XC4VSX35-10ff668.     

Investigation of directional effects on the electrical conductivity and piezoresistivity of carbon nanotube/polypropylene composites obtained by extrusion

The electrical conductivity and piezoresistivity of multiwall carbon nanotube (MWCNT)/polypropylene (PP) composites obtained by extrusion are investigated, with particular attention to the possible directional effects generated during the extrusion process. This is accomplished by investigating the electrical and electromechanical responses of the nanocomposites at three MWCNT weight concentrations (3, 4 and 5 wt%) in three directions, viz. the extrusion direction, transverse to extrusion (in-plane) and through thickness. Higher electrical conductivity in the extrusion direction was more evident for the lowest MWCNT content. However, the piezoresistive sensitivity was similar in all directions. Films with 4 wt% showed the highest piezoresistive sensitivity, reaching gage factors of?~?4.5 for strains between 0 and 0.8%, and?~?10.2 for strains between 1 and 3%. After an initial drop in the electrical resistance, concomitant with stress relaxation, the changes in electrical resistance showed large reproducibility. Digital image correlation conducted during cyclic piezoresistive testing at 0.8% strain indicates small accumulation of local plasticity as the number of cycles increases, especially in zones near the electrodes. These irreversible changes in the material are expected to trigger the permanent changes in the electrical resistance measured.

Graphical abstract