Newcastle University School of Chemical Engineering and Advanced Materials


(A gentle introduction to noise filtering)


Exponentially Weighted Moving Average Filter

Averaging Filter
Moving Average Filter
Exponentially Weighted Moving Average Filter
1st-order Low-pass filter

Choice of Filter Constants

Frequency Characteristics

The moving average filter regards each data point in the data window to be equally important when calculating the average (filtered) value. In dynamic systems, however, the most current values tend to reflect better the state of the process. A filter that places more emphasis on the most recent data would therefore be more useful. Such a filter can be designed by following the procedure used in developing the moving average filter. As before, the starting point is the mean expressed as:

But in this case, consider also the mean with one additional point

Since, therefore,


By shifting the time index back one time-step, we obtain the corresponding expression for as:

To simplify the notation, let , which implies that . We can write the filter as:

This expression is known as the Exponentially Weighted Moving Average Filter. When used as a filter, the value of is again taken as the filtered value of . Notice that now, calculation of does not require storage of past values of x, and that only 1 addition, 1 subtraction, and 2 multiplication operations are required.

The value of the filter constant, , dictates the degree of filtering, i.e. how strong the filtering action will be. Since , this means that . When a large number of points are being considered, , and . This means that the degree of filtering is so great that the measurement does not play a part in the calculation of the average! On the other extreme, if , then which means that virtually no filtering is being performed.

The Exponentially Weighted Moving Average filter places more importance to more recent data by discounting older data in an exponential manner (hence the name). This characteristic can be illustrated simply by describing the current average value in terms of past data. For example, since






But ,


If we keep on expanding terms on the right hand side, we will see that the contribution of older values of are weighted by increasing powers of . Since is less than 1, the contribution of older values of becomes progressively smaller. The weighting on may be represented graphically by the following plot:

Figure 2. Exponential weighting effect

What this means is that in calculating the filtered value, more emphasis is given to more recent measurements.

The Exponentially Weighted Moving Average filter is arguably the most commonly used noise reduction algorithm in the process industries. However, it is known commonly by a  another name; one that has its roots in electrical circuitry that are used to produce smooth electrical signals.

swot.gif (2080 bytes)
go to first page
go to previous page
go to next page
go to last page

Back to topwpe5.gif (958 bytes)

Back to the Swot Shop

Copyright M.T. Tham (1996-2009)
Please email errors, comments or suggestions to