Comparative Analysis for Conversion of Double Base Number System (DBNS) from Conventional Number System Using Binary Search Algorithm by Changing Base Patterns for ANN-Based Applications

Efficient execution of high-performance arithmetic operations is demanded by compute-intensive signal processing algorithms as well as neural network-based computations. The attention of many researchers is being garnered by the double base number system (DBNS) because of its high-performance arithmetic units. However, the advantage of DBNS cannot be exploited due to the large conversion time required from binary to DBNS. The main focus of this paper is to make a comparative analysis on conversion of DBNS from conventional number systems (CNS) with bases 2 & 3 and 2 & 5. DBNS is a non-conventional number system known for its efficient multiplication capabilities. The paper addresses this challenge by presenting a new transformation scheme specifically designed for converting CNS into a DBNS representation with two different sets of bases like 2 & 3 and 2 & 5. In this context, it is pertinent to mention that the set of bases can be termed as single index pair. The key contributions of the paper are to establish the novelty by making a comparative study by incorporating different search algorithms in this conversion process, making DBNS more accessible and useful for efficient multiplication operations.