An efficient heuristic to identify threshold logic functions

Abstract

CarbondaleA fast method to identify the given Boolean function as a threshold function with weight assignment is introduced. It characterizes the function based on the parameters that have been defined in the literature. The proposed method is capable to quickly characterize all functions that have less than eight inputs and has been shown to operate fast for functions with as many as forty inputs. Furthermore, comparisons with other existing heuristic methods show huge increase in the number of threshold functions identified, and drastic reduction in time and complexity.

ATPG for Delay Defects in Current Mode Threshold Logic Circuits

“…Thus, the tool of [7] was modified in order to assign weights of each TLG. The TL function identification method proposed in [20] was employed to check whether a particular Boolean gate cluster can be implemented as a TLG. The fan-in bound of a TLG was set to eight.…”

Nanopipelined threshold network synthesis

“…By performing reductions in this area, the algorithm proposed typically achieved over 20% fewer clusters, and 40% fewer buffers. Future research will explore using more robust methods for identifying threshold functions such as Gowda et al [2011]; Palaniswamy and Tragoudas [2012], and Palaniswamy et al [2010], in conjunction with balancing paths using the algorithm proposed in this article.…”

Improved Threshold Logic Synthesis Using Implicant-Implicit Algorithms

“…Some examples of TLG nano-implementations are MOBILE and neuMOS [Celinski et al 2000;Beiu et al 2003;Quintana and Avedillo 2008]. This has renewed interest in identification and optimization of TFs Palaniswamy et al 2009Palaniswamy et al , 2010Palaniswamy and Tragoudas 2012a].…”

“…The objective of the methods in Palaniswamy et al [2010] and its extended version [Palaniswamy and Tragoudas 2012a] is to determine very quickly the percentage of n input functions that are TFs. The advantage of these methods is that TF identification calculations involve only modified Chow's parameters instead of using linear programming (LP).…”

“…The presented approach in Palaniswamy et al [2010] is pessimistic, that is, some of the functions termed non-TFs may in fact be TFs. Since the total number of functions grows tremendously as n increases, this approach is practically limited to very small values of n. The method in Palaniswamy and Tragoudas [2012a] is less pessimistic. For n-input functions with n > 10, calculating the quantities required for TF identification of each n-input function is a relatively time-consuming task.…”