D-modulation implementation

[yiannakop]

Hi. I want to implement an algorithm (in matlab or C) for delta modulation. I don't know much on the issue, but I've read that delta modulation is a one-bit code: 1 is transmitted if the current sample is greater than the previous sample and 0 otherwise. I don't understand how can this code work, since the difference between 2 samples is not standard?
Thanx,
Theodore

[Joe Nellis]

At a high enough sample rate would the delta be reduced to zero or one?

[yiannakop]

I don't excatlly understand your answer. What do u mean by saying that delta would be deruced?
Regards,
Theodore

[Joe Nellis]

The data you are collecting may not fall into this scenario but here is my idea. Suppose we are doing analog to digital conversion. We want the delta from the last reading. If the timeslice is fast enough you may get a reading that is the same as the previous reading (or within a tolerance enough to be considered 'no change') This is your zero state.

Since the sampling rate is so fast, the most the signal can move up or down from the last reading can be considered a discrete unit. Whatever the actual value of this discrete unit, it is considered to be your one state. The idea is to gradually turn up the sampling rate until you are seeing only zero delta or your discrete unit delta.

If the input varies wildly (possibly the input follows a function which is not continuous) then this method probably won't work.