Understanding Error Location Analysis (ELA)


Error Location Analysis is a technique of digital bit error testing that provides means to acquire a stream of bit-accurate locations of errors detected during a testing session, and then to perform error location analysis algorithms on this stream to provide useful analysis results.  


The hardware engine that supports Error Location Analysis works in cooperation with other hardware features that maintain synchronization between the incoming channels bit streams and their reference data patterns and makes comparisons between the incoming streams and the reference data patterns.  The result produces a stream of events ranging from errors (mismatches between the incoming and reference data patterns) to synchronization status, in addition to others.  These events are queued and destined for software algorithm analysis.  The memory used to transfer these events from hardware into software algorithms, however, is finite, and since the bit streams being tested generally operate at bit rates orders of magnitude faster than the stream being provided for software analysis, it is mandatory to also manage when the event stream may be interrupted by being too full.  In this case, certain events, such as synchronization events, are considered high-priority and room is always provided for these types of events, while errors may be discarded.  Its important when this happens to group such "squelched" periods together and to indicate with other types of high-priority error location events when these situations begin and end.  This results in an error location event stream that contains synchronization information, error information, and squelch management information, which is used to perform software algorithms that can precisely represent error analysis for only those portions of time when complete error information is available.


Fortunately, it is usually the case that high error rate situations can be diagnosed even when only examining a small proportion of the overall errors from the raw channel under test.


Examples of Error Location Analysis include:


ELA Burst Lengths Histogram Application


The ELA Burst Lengths Histogram application displays the distribution of the lengths of the error bursts encountered during your run session.  A burst is defined as a consecutive grouping of errors. You can use the Burst Separation setting below to select how many consecutive error-free bits will trigger the end of the burst.  When error free bits are selected to be allowed within bursts, the burst length will include these error free bits if they occur inside the burst.  Bursts always begin and end with true errors.


ELA Error Free Intervals Histogram Application


The ELA Error Free Intervals Histogram displays the distribution of the lengths of error free intervals found during your run session.  An error-free distance is defined as the number of bits between detected errors.  This histogram is very useful to detect systematic souces of errors, since errors occurring at a particular cadence will also create error free intervals of specific lengths.  This histogram easily demonstrates if errors are occurring randomly or systematically.


ELA Errors Modulo Histogram Application


The ELA Errors Modulo Histogram displays the location of individual errors relative to a repeating modulo factor.  Different modulo factors can be very useful for analyzing if errors are coming from certain sources within your communications system.  For example, if your system has a 32-bit bus, then seeing a histogram of errors modulo 32 will demonstrate if one particular bit of your bus is more-prone to error than the others.  


ELA Block Errors Histogram Application


The ELA Block Errors Histogram displays the distribution of the number of errors found in a specified block size.  Blocks are defined by selecting a number of bits in each symbol as the Symbol Size in the ELA Console settings, and by selecting a number of symbols in each block as the Block Size in the settings below.  If you select the Symbol Size to be 1, then the Block Size will represent a number of bits and the histogram will represent the number of bit errors in each consecutive block of the specified block size.


ELA T-Strength Strip Chart Application


The ELA T-Strength Strip Chart Application displays a running log of the maximum number of symbol errors encountered in all the blocks analyzed in each interval of measurement.  Blocks are defined by selecting a number of bits in each symbol as the Symbol Size in the ELA Console settings, and by selecting a number of symbols in each block as the Block Size in the settings below. The measurement intervals is a number of seconds defined in the settings below.  Each vertical bar on the display represents one interval of measurement, and the height of the bar indicates how many symbol errors occurred during that interval.