ALTERNATIVE ALGORITHM FOR OPTIMIZATION OF REED-MULLER UNIVERSAL LOGICMODULE NETWORKS

Reed-Muller universal logic modules (RM-ULMs) can be used as individua l modules or as networks to implement RM functions of any number of va riables. For a large RM expansion, it is generally more efficient to r ealise it using cascades of low-order RM-ULMs. A programmed algorithm for the optimisation of the number of modules at the sub-system level has already been published. Here an alternative algorithm is presented which performs similar optimisation of fixed-polarity generalised RM expansions but without the need to maintain a saved-branch counter and an input record. Also, whenever the number of piterms having the spec ific control-tuple-state exceeds two, it is not necessary for the algo rithm to perform separate variable examinations. Consequently, the sav ing in computation time will be especially significant in RM expansion s with large number of variables. The resulting algorithm is simple in structure and can be easily implemented using high-level languages su ch as C or Fortran.