Efficient Low-Cost Fault-Localization and Self-Repairing Radix-2 Signed-Digit Adders Applying the Self-Dual Concept

摘要

Signed-digit adder, which eliminates carry propagation chain, can execute addition operation independent of the length of operands, in constant time. The confined carry propagation implies remarkable advantage in terms of error detection, localization, and correction. We developed a new low-cost technique, for fault-localization and error-correction, which utilizes the self-dual concept in binary signed-digit adders. The foundational idea of our technique is that the designed method, when fed by the complement of its functional input under the existence of a stuck-at fault, yields a fault-free inverse of the expected output. In this method, after detection of a fault, the method continues with input inversion, recomputation, and desired output inversion. The faulty component will localize by comparison of the faulty and fault-free outputs. Higher reliability with less computational time and reduced hardware-area overhead are the distinct features of this new design. In this method, all of the single stuck-at faults can be localized and corrected, whereas previous approaches were unable to localize and correct with 100 % reliability even with longer time durations and greater hardware cost. The experimental results prove that our design requires 20 %-90 % less time and 20 %-40 % smaller area overhead compared with a previous related work.