Differential Pulse Code Modulation

The Differential Pulse Code Modulation (DPCM ) is a pulse modulation method that converts a time-discrete signal into a time and value discrete digital signal. It is an extension of the pulse code modulation (PCM ), and is a precursor to the Adaptive Differential Pulse Code Modulation (ADPCM ). In DPCM difference values ​​of successive samples are formed, resulting in bursts of high autocorrelation as, for example, digital audio signals to a data reduction. DPCM 1950 was developed at Bell Laboratories by C. Chapin Cutler.


A system consists of a DPCM encoder ( modulator ) and a decoder ( demodulator ) with a channel therebetween. Designated to be transferred from the sequence of the data source in the adjacent figure with x [ n], the difference sequence e [n ] is first formed:

Is zugefürt the quantizer Q forming the digital signal therefrom. This is transmitted both through the channel to the decoder, as well as an addition stage fed, which

Forms. The value is stored for the duration of a sampling period, shown in the figure as the z -1. The output of the memory element (English Tap) is supplied to the input stage for forming the difference.

Alternatively, there is also a further DPCM method, which on the input directly the difference between two input values ​​in the form of

Forms. This form of DPCM does not have to compensate for the disadvantage of quantization noise and plays only a minor role in practical applications.

The decoder is to reconstruct from the received sequence using a tap and an addition stage the original signal sequence by the following relationship:

This form of DPCM with only one memory element in the encoder or decoder provides the original and simple DPCM system dar. As an extension of predictive coding, a sequence of p samples are used for the difference formation. The differential signal is then formed by

Represent the prediction coefficients which weighting factors. Depending on how these prediction coefficients are selected in relation to the course of the input signal sequence, the difference signal can be within the scope of, at constant information content, reduce.