A distance measure approximating the mean square metric

The method of least squares is perhaps the oldest estimation procedure. The justification of choosing this metric is often based on heuristics and has been found to work successfully in many situations. It renders the subsequent problem solving in a mathematically tractable format. This is one of the somewhat stronger reasons for the method attracting diverse applications. However, computation for this estimation is often unnecessarily intensive and the accuracy we are forced to accept, are frequently more than we need. The paper presents a distance measure, which, while keeping the effect of the basic properties of mean square computation somewhat intact, is computationally much simpler and conveniently hardware realizable with faster speed and in general, requiring less real estate on a VLSI chip. In this type of square error measure, we use an appropriate estimate which is a function of position i of mismatch and an m bit difference starting from bit position i of the two numbers. The paper shows how this function is arrived at. The measure proposed, should lead to a hardware implementation ranging from a simple sequential process to a faster highly parallel hardware realization