Approximate Adders with Configurable Input Wiring: A Quantum-dot Cellular Automata Nanocomputing Perspective

In this work, two novel low-hardware-cost approximate adder designs utilizing quantum dot cellular automata (QCA) technology are introduced. The designs exploit the configuration of input bits and carry generation from the inexact lower-part (LP) to the exact upper-part (UP) of the adder, yielding efficient performance in terms of area and delay. When benchmarked against truncation-based adders (trun-0 and trun-1), superior performance is demonstrated, while significant area reductions are achieved in comparison to exact QCA-based adder designs. Evaluation metrics, including design area, propagation delay, mean error rate, and performance within an image processing application, are employed to provide a comprehensive analysis. The experimental results suggest that the proposed designs achieve over a 60% reduction in the area-delay product relative to exact adders, exhibiting substantial benefits in terms of minimized area requirements, decreased propagation delay, and improved error performance compared to previous approximate adder designs. Specifically, the proposed 8-bit adder achieves area savings of 46% and 18% compared to 8-bit EXA and AXA designs, respectively. For image processing applications using 3×3 low-pass filtering, proposed designs demonstrate robust performance with average PSNR values of 16.8 dB and 18.1 dB across multiple test images proving a comprehensive evaluation of the adders' efficacy.