Efficient Symmetrical Imprecise 1-Bit Full Adder Cells Using CNFET Technology for Image Processing Applications

Nowadays energy consumption in mobile electronic consumers is a serious concern than ever. These devices exploit digital signal processing (DSP) blocks in their structure to perform multimedia algorithms. Since in most cases the output of these blocks is used for humans with limited vision perception, it is feasible to utilize approximate computation methods to enhance circuit parameters such as latency, power consumption, area, etc. In fact, the circuit parameters are enhanced at the expense of making some outputs imprecise. In this paper using the remarkable benefits of carbon nanotube field effect transistors (CNFETs) along with approximate computing method, two novel imprecise Full Adder cells are presented. Extensive simulations at both application and switching levels confirm the supremacy of the proposed cells against their conventional and state-of-the-art counterparts. Moreover, to study the robustness of the proposed cells against process variations, Monte Carlo transient analysis in the presence of carbon nanotube (CNT) diameter variations is performed. Simulation results indicate the robustness of the proposed cells.