We can call the concept of Self Organizing Networks as Artificial Intelligence in telecommunication networks. Self adaptability and self organization have been famous topics since the beginning of humanity because organism which can adapt to its current environment survives.
This concept has not been changed since then. Networks, computers, devices which can adapt to unique properties, or functions or features of its operating environment (user base) survive or will survive. What does Survive mean in this context? It means these systems will serve its users in a much better way and will acquire more users.
If we go back to telecommunication networks, I want to talk about two important key performance indicators;1- Capacity and 2- Coverage. Capacity means the speed which bits delivered to a user in telecommunication network. It also represents the total number of users that can connect to the network simultaneously. Coverage means the total area where users (cell phones) can connect to the network, or it is the total area where our phones receive sufficient amount of signal power so that they can connect to the network.
All that being said, in 2G and 3G telecommunication networks, capacity and coverage were almost fixed. That is, network was designed to meet particular capacity and coverage targets. Once network was designed, it stayed as it was for a long time. Service providers deployed more base stations to provide more capacity and coverage if there was enough user demand. However, there was not any dynamic network capacity or coverage adjustment.
In 4G networks, there is a new concept introduced, called self organizing networks. This basically means that base stations can change their coverage and capacity based on the changing needs/demand in the network. If demand increases, base station increases its capacity. If more capacity is needed at a particular location, base station shrinks its coverage to serve those specific locations with more capacity.
4G base stations collect data about the current coverage and capacity performance. Based on the past performance of the network, they also predict future performance with some small error. So, base stations know the expected coverage and capacity levels at a certain time in the future, and adjust their antenna power, direction, antenna tilt, and other available resources to achieve the predicted coverage and capacity levels. (Certain optimization algorithms are used for this purpose; but this is not the topic of this writing)
There is one very important critical point that we need to pay attention here. As we know, networks in general, and 4G networks specifically are designed to cover a large geographical area. There are hundreds and sometimes thousands of base stations to cover a particular market (geographical area). So, once the network is designed, if group of base stations start changing their settings based on the demanded coverage and capacity, this will create problem for other base stations nearby (interference). Therefore, all base stations serving particular area need to act together to improve the performance of the overall network. For this to happen, base stations need to exchange information with each other (through their X2 interface) about their current load levels, predicted capacity & coverage levels in the future, etc. Once all base stations have this information available, each base station finds the optimum parameter settings to meet user demand, and at the same time to keep the negative impact on the overall network performance very low. Thus, network starts optimizing itself to serve its users in a much better way.
I expect that self organization will be implemented in much better way in 5G networks that it was implemented in 4G networks.
To be continued
