Constant factor delta modulation - a new instantaneously adaptive delta modulation system

A new model of an instantaneously adaptive delta modulator called hero a "CONSTANT FACTOR DELTA MODULATOR", (abbreviated C.F.D.M.) has been developed and the selection of suitable constant factors (adaptation constants) with its adaptation logic has been described. The basic delta modulator has been adapted to give an improved performance by introducing a small memory and prediction method in the feedback loop thereby enabling the coder to adapt to the instantaneous variations in the analogue input signal. This C.F.D.M. model of adaptive system adapts its step size, at every sampling instant r, as a result of the detection of the four possible binary groups formed from the last three binary values transmitted. The adaptation constant which is the ratio of-the present step size mr, to the previous step size mr-l, can have, at any sampling instant, one of four values with a magnitude of Ai; A2, A3 and A4, corresponding to the four different possible groups formed. The polarity of the present step size is the same as the present binary value Lr. The effect of this C.F.D.M. system is that for a given decoded signal to noise ratio, the necessary bandwidth of the transmission channel is reduced. The C.F.D.M. described here gives an improved overall coding characteristic and removes an objectionable hunting characteristic compared to the one-bit memory adaptive DM by JAYANT(13). It offers wider dynamic range for the bandlimited Gaussian Input. The results are compared with other similar schemes on adaptive delta modulators and computer plotted graphs are presented whenever necessary. From these results and responses, the C.F.D.M. seems to be promising for encoding video-signals. Several computer simulations have been made for the design of the Constant Factor delta modulator, JAYANT's CODER, WINKLER's H.I.D.M. coder and linear delta modulator. The performances of these coders have been compared. A considerable number of computer simulation results are presented which relate to digital low-pass filter and the estimation of power spectra.