Implementation alternatives of a DBNS adder

In this paper the implementation issues of a double-base number system (DBNS) adder are discussed. Two alternative implementation strategies, dedicated analog hardware and analog CNN processor are considered and their differences compared. The principle of including multiplication feature to the DBNS adder is discussed. Moreover, it is discussed what kind of problems arise from the finite size of the array to be used to add numbers in a DBNS form. Finally we show some simulation results.