What Type of Analysis Should You Use?

The BitAlyzer offers many types of error location analysis. Each type is designed to show one perspective of the precise statistics of how errors occur during a measurement. The different analysis methods include:

• Error Statistics — A tabular display of bit and burst error counts and rates.
• Strip Chart — A strip chart graph of bit and burst error rates.
• Burst Length — A histogram of the number of occurrences of errors of different lengths.
• Burst Length — A histogram of the number of occurrences of errors of different lengths.
• Error Free Interval — A histogram of the number of occurrences of different error free intervals.
• Correlation — A histogram showing how error locations correlate to user-set block sizes or external Marker signal inputs.
• Pattern Sensitivity — A histogram of the number of errors at each position of the bit sequence used as the test pattern.
• Block Errors — A histogram showing the number of occurrences of data intervals (of a user-set block size) with varying numbers of errors in them.
• 2-D Error Map — A two-dimensional map of the positions of your errors in the data stream.

What to do first?

A good place to start is with an Error Statistics display.

This BitAlyzer view will show you the first type of error analysis, where the total errors are separated into burst and non-burst categories. This simple separation can help to isolate errors that may be random, or small, from error bursts, or errors that are due to a major interference. You will want to understand more about Integration Period, Burst Error Free Threshold, and Minimum Burst Length.

A good next step is to view a Strip Chart of these error rates. The Strip Chart illustrates the occurrence of errors over time. You will be able to see if the errors occur in any regular pattern or cycle. Panning and zooming this graph and using the cursors allow you to make measurements from the graph.

Now you begin to have choices. Depending on what type of errors you have uncovered, you can choose particular types of analysis. A common next step would be to look at the burst lengths present during your measurement. Look for anomalies from what you would expect. Burst lengths can quickly distinguish many errors that are truly random from ones that are somehow correlated. Burst lengths communicate a lot about an underlying error mechanism. For instance, Gaussian random errors are typically one or two bits long, while interference is typically more than one bit long.

You can learn more about what different types of Burst Length Analysis are telling you in Interpreting a Burst Length Histogram.

The next step is to isolate your errors. Systematic errors typically leave behind signatures that can easily be identified in a histogram of error free intervals. This histogram shows the number of occurrences of bit spacings between errors. Any spacing that occurs suspiciously more likely than any other spacing implies a non-random behavior. The size of the spacing indicates the frequency of the systematic error.

For help in understanding what an Error Free Interval analysis has uncovered, you can read more on Interpreting an Error Free Interval Histogram.

Congratulations!

Once you’ve gotten this far, you are well on your way to understanding the uses of the added power of error location analysis in your work.