On Heuristic Linear Decomposition of Symmetric Index Generation Functions

Decomposition methods of index generation functions gained a lot of interest lately due to their impact on memory size in hardware realizations. In this paper, we focus on a special class of those functions, i.e. symmetric index generation functions. We analyze the properties of this class, and how they influence on searching for a linear decomposition of index generation functions. In particular, we focus on a heuristic algorithm that uses discernibility sets, which efficiently reduces index generation functions. We present experimental results to prove the efficiency of this algorithm. Furthermore, we prove that the properties of symmetric index generation functions lead to better time efficiency of this algorithm, due to significant variables reduction.