Why SON never picked up and would die Soon?
First, there is nothing like Real time SON, only automated parameter tuning.
Most of the SON vendors are dependent on Performance (PM) counters which at-least take 15 minutes as aggregation period and processing time over it. Although there are few vendors supporting very few Performance Metrics near real time but those come at cost of heavy processing and again changes made in the network are not real time.
Second, Reactive approach.
SON was one of the most hyped feature of cellular networks, came inbuilt with LTE network as Distributed SON which has limited view of the network and works only within local eNodeBs. Then there are vendors with Centralized SON, which depends on PM counters from EMS/OSS and tune parameter reactively. RF network is very dynamic and co-related and hence both of these approaches are able to deliver as promise.
Third missing the most important information i.e. Customer experience metrics.
The most important KPI for any network operator customer experience but no SON solution takes that into account, Oh Yes MDT feature do, But how many deployments we have seen of that?
Forth, No real value with current SON solution, another way to make KPI sheet more Greener.
Traffic steering from one radio access technology other is projected as big time SON feature, again reactive approach. Platform wait for KPI to degrade, Users to suffer and then offload traffic forcefully to achieve only Green KPIs but not the desired results.
In reality, Only Two features which works and are commonly deployed are Plug n Play and ANR (Automatic neighbor relation).
Fifth, Zero Co-relation between Network Elements
SON, so called as Self optimizing network, works only on Radio Network. Doesn’t even consider or co-relate the various elements which could be the choking or de-grading factors.
Sixth, Absolutely not a chance of current solutions for future network
With current solutions and static SON algorithms, it is not possible to optimize current networks, leave the thought of future hyper-dense networks.
One feature for all networks around geographies: Static SON algorithm
One of the biggest issue with the current, so called SON solution is they fixed and static solution/algorithms for networks, all geographies and all use-cases. Pretending dynamic solutions, vendors give few input only based on input thresholds.
Future of Network Optimization:
The solution is not SON, Solution is Machine Learning and Artificial Intelligence based Platform which not only is Pro-active, Predictive but also takes into account all Co-relation of network elements, customer experience metrics and support future networks. Another important feature is to have Dynamic Algorithm builder rather than having static threshold based algorithms
We have created an End to End Network optimization platform which has below features and is future proof unlike SON.
First, there is nothing like Real time SON, only automated parameter tuning. There is Machine learning and Artificial Intelligence based on big data analytics and data mining.
Real time optimization is not possible as we need to know if Network is degrading but what if It has a pattern, what if during certain time of the day the Cell site XYZ is congested for few hours. The machine can learn based on data mining and analytics and take Predictive and Real time measures.
Second, Reactive approach. Proactive approach
With all the data and understanding, solution will take pro-active decisions, such as change handover parameters to make one way handover faster for 8-9 am in morning and hence we come out of reactive approach.
Third missing the most important information i.e. Customer experience metrics. Customer Experience metrics recorded, Geolocated
In the end, it doesn’t matter if your customer has bad experience and you have all Green KPI Report. Hence it is very important to capture customer experience to understand the issues, concerns and areas.
Forth, No real value with current SON solution, another way to make KPI sheet more Greener. Optimization adds value if the algorithm kicks in even before the de-gradation happens or going a step ahead doesn’t let degradation of KPIs and Customer Experience happens, what is the fun of optimizing when you already had observed call drops >10%.
Fifth, Zero Co-relation between Network Elements Co-relation is the Key
Co-relating not only Performance with Configuration and Alarms but multiple layers and elements of networks. It might happen one of your network element is down which affects the overall experience.
Sixth, absolutely not a chance of current solutions for future network Future optimization requires a lot more than tuning parameters.
Networks are getting dense, not only with numbers but with number of vendors and technologies. The Customer expects a seamless experience. There is no Hybrid SON or Centralized SON which would be able to help future networks without Big Data analytics/Data Mining/Co-relation. The solution needs to have predictive analysis.
One feature for all networks around geographies: Static SON algorithm Bring your own Algorithm
Every Network is different, there is no such One Solution Fit All networks and geographies. Hence, the base platform shall support algorithm builder based on each unique use-case and network.
Would be happy to share more information and schedule Demo.Please reach out.
Product manager / Co-Founder, Author Techplayon