IMPLEMENTATION OF LNS BASED APPROXIMATE LOGARITHMIC MULTIPLIER USING OPERAND DECOMPOSITION METHOD

Electronics and communication engineering

Generally multiplication consumes more power, area and are complex. Multiplication plays a vital role in implementation of many Digital Signal Processing (DSP) applications. One feasible alternative solution to simplify multiplication in the use of Logarithmic Number System, instead of binary. For exact fixed-point multiplication a matrix multiplier is usually used. The LNS based Approximate Logarithmic Multiplier implemented using Mitchell’s algorithm significantly reduces area, power and time delay while having low worst-case error which are used in the application of Convolutional neural networks. Mitchell algorithm is done by taking logarithm for the two binary inputs, adding them and finally taking antilogarithm for the added result. Thus, the multiplication is converted to addition. The proposed system, operand decomposition shows a significant increase in the amount of accuracy over the existing method when applied to the logarithmic multiplication. Operand Decomposition performs Mitchell algorithm twice according to the method, in order to improve the accuracy.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.