Digital Filtering With Unreliable Molecular Electronics

Molecular electronics such as silicon nanowires (NWs) and carbon nanotubes (CNTs) have emerged as the potential fabrics for building future nanocomputing systems. However, the non-ideal fabrication process and the consequent excessive defect rates make reliable molecular computing a challenging task. Existing solutions addressing this problem aim at achieving the absolute correctness, which becomes increasingly infeasible due to the prohibitive cost involved. It is noticed that in the context of signal processing applications, absolute correctness is generally unnecessary. This could relax the design constraints on defect tolerance in molecular-scale integration. In this paper, we propose a new approach based on defect-insensitive signal processing. By allowing molecular-scale DSP architectures to bear defects, the proposed framework achieves reliable signal processing while ensures the effectiveness of defect tolerance.