The Role of Big Data

At the same time there is a need for aggregating data to find patterns that exist across large portions of the network.  This is the role of Big Data.  Time constraints are relaxed and the focus is on pattern recognition.  Once an actionable pattern has been identified, then it can be passed to the EEO system for short-term execution, and to network planning, and other systems for longterm.

For example, let’s look at a variation of the use case described above.  If, instead of a backhaul failure, there is a failure that affects a basestation itself, the EEO system can also move RAN capacity to other basestations in the neighborhood.  This improves EEO in the short-term.  At the same time the Big Data system is looking at failure rates of components and it discovers that, on average, a particular board in a particular basestation model fails in 18 months (a Telco tells me that such a pattern has been found).   So, now both the EEO system and the network planning system track when these particular boards are put in service.  Instead of waiting for them to fail, they are now replaced when they have been in service for 17 months.  With NFV, as Telco networks begin more and more to resemble Clouds, anticipation based on actionable patterns may find more places to add value.

However, once having found such an actionable pattern and created algorithms based on it, the work can’t stop there.  There has to be ongoing work to check and make sure that the patterns have persisted.  For example, has the board vendor made a software or hardware upgrade that extends the mean time before failure?  If so, the Big Data system by itself will not see the new pattern, because all of those boards have been removed.  So there has to be some form of ongoing test.  In this use case, running an ongoing test of a sample of removed boards to see it the failure rate has increased, while having the EEO system report in service outages to see if it has decreased.

Thus it can be seen that there is a symbiotic relationship between EEO and Big Data in QoE Management.  Neither is sufficient by itself to assure the highest possible QoE, but working together they can.