Daytime Project

GSM Base Stations

GSM Base Station Alarms and Performance Metrics


Global System for Mobile Communications (GSM) technology, which is based on the cellular network, is highly familiar to us, as we all are users/subscribers of one or more such network. Wikipedia defines the cellular network as: “A cellular network is a network of handheld mobile phones (cell phones) in which each phone communicates with the telephone network by radio waves through a local antenna at a cellular base station (cell site). The coverage area in which service is provided is divided into a mosaic of small geographical areas called ‘cells’, each served by a separate low power multichannel transceiver and antenna at a base station. All the cell phones within a cell communicate with the system through that cell's antenna, on separate frequency channels assigned by the base station from a common pool of frequencies used by the system.”


figure-1

The cell sites, a.k.a the Base Transceiver Station (BTS), are the front-end nodes to the subscribers of the GSM network. Mobile devices are connected to the network via communication with the nearest cell site, and upon successful connection to the network, they communicate with other mobile or fixed devices through the cell cites. Therefore the cell cites, or base stations, are key components of GSM network coverage and communication. Their physical locations and distributions over the country are determined according to the sizes of local population and network covering goals.



figure-1

Today, possibly because of high-quality devices and very good operation experience, the up rates of GSM base stations are quite high, usually more than 95% of the time. On the other hand, ever-increasing need to data transfer in today’s world decreases the tolerance to any interruptions, even 0.5% downtime. Therefore, this use-case challenges to achieve an almost perfect systems with near-zero downtime by successfully predicting the possible interruptions.


Although the cell sites are fixed, their users are mobile and moreover, the user characteristics is dynamic. Therefore, traditional methodologies are possibly inadequate for making good predictions and providing even longer-term foresight. The short term predictions are vital for the target of 100% uptime. The longer term foresights are also important for investing/establishing new stations or replacing the existing ones.



Main Innovation Challenge


Fusion of different data sources, especially from two main streams, namely cell site data (alarms, metrics, KPIs, for both voice and data transfer), and customer complaints; then, feeding these data into an LSTM-based digital twin model.

Contact Information:

Mail: ahmet.cezayirli@turkcell.com.tr