Application of artificial neural networks in differential pulse code modulation scheme

In this work an Artificial Neural Network with Radial Basis Function (RBF) is employed to model a predictor, utilized in Differential Pulse Code Modulation (DPCM) scheme. The RBF predictor estimates the magnitude of signal incoming to DPCM, the error between incoming signal and estimated one is applied to quantiser unit. The resultant error contains a few bit word length and is transmitted in data format towards a receiver unit. The output of the RBF predictor is added to error at the receiver part of DPCM to produce actual output. In this study application of RBF predictor in DPCM is explained. The potential offered by DPCM scheme using RBF predictor leads to a considerable amount of reduction in word length of filter coefficients.