High-Efficient Approximate Subtractors for Change Detection Applications

Approximate full subtractors (FSs) are used in unsigned restoring dividers (Drs) for change detection in image processing. In this letter, two new single-bit approximate FSs with eight and six transistors are presented which first circuit uses XNOR-F2 gates while the second one is designed based on multiplexer-F2 gates. The circuits are named XNF2 and MF2. The designs are implemented by the gate diffusion input (GDI) technique and 32-nm carbon nanotube field-effect transistor (CNTFET) technology. The XNF2 improves the delay and area with a favorable error rate (ER), with only two errors. The MF2 creates a suitable tradeoff between power, area, and accuracy and free of error distance (ED) > 1 that leads to a mean relative ED (MRED) < 0.375 and only 3 errors. Different figures of merits (FoMs) confirm the reliability of the cells for error-resilient applications, like change detection.