Professional    Learning    Academy

From the experts in Telecommunication, Cyber Security, Big Data and IoT

icon icon

Learn Cellular Wireless communication technology, its architecture and management

icon icon


Learn construct of Information security management system in a simple way

icon icon Learn application of Big data analytics in Internet of things domain

Big Data Analytics in Telecom Tower operations

As you would have known, Base Transceiver Sub-System(BTS), Base Station Sub-system(BSC), Node-B/eNode-B are the distributed system in GSM, 3G, 4G communication which handles the mobile communication at last mile with mobile phones, handsets and stations at the last mile in a telecommunication service provider domain. A typical tower site looks like the following:



You can see it is a combination of all the engineering. Right?  It is all distributed across the geography and mobility is achieved. The passive elements like tower, electricity, air conditioner, battery bank, power sources, Intelligent power units, weather proof cabinet are all available to support the operation of all active units that is either BTS, BSC, NodeB, eNodeB. 

The service uptime has become so critical that the mobile service provider has sought the help of eco-system vendors to support the operations of these towers i.e. passive assets to keep the mobile service up.


So, let’s understand the passive element that needs to be monitored to keep the active mobile network element up. These are:


  1. Remote site security that means these sites will have access control device and cameras.
  2. Environment control that is temperature sensors, humidity control sensors, Air conditioners and free cooling units. What is free cooling units? These are the fans in the weather proof cabinets. If outside temperature is less than inside temperature, throw the air inside given the humidity level and vice versa.
  3. Energy control units those include DG Generator run hours and fuel, Battery Bank Voltage, Charge/Discharge Current and Temperature, mains/grid availability, consumption, phase and frequency. If there are solar energy sources, then other list of sensors and control.

The following diagram shows, how the data from remote site can be brought to the main data centres from those sensors.

In this blog, we will try to understand the intelligence that big data analytics brings to optimize the site operation. You can think of thousands of such site available across an operator  or multiple operators putting their active components (BTS,BSC…) on same passive infrastructure to provide the communication services both voice and data.


The Big Data Characteristic Map




Unstructured data


  1. All sensors data is captured at a set timer frequency, generally less than a minute.
  2. The main power from grid will have a digital meter and sensor on it. It provides the power factor, voltage and current at an instance of time. The electricity meter has different registers that accumulates the data and that can be read. Mind it, these accumulators provides the total aggregated consumption from inception at that instance of time.
  3. Battery banks’ sensor provide the data of charge/discharge level of each cell in the bank.
  4. Diesel generator provides the temperature, fuel level, run hours through either manual read or through sensors.
  5. Solar power will have like main power parameters coming from site and also the health of each photo-voltaic cells.
  6. Intelligent power unit will have a data on which time of the day what was the power source. For example, grid power voltage has gone down, hence either battery power or DG set has been made on. These data will show the power source at an instance of time.


Structured data

  1. There will be a system where all the past data like last meter reading is kept for operation cost calculation and settlement purposes.
  2. Hence, to find out the energy consumption during a period, need to subtract the most recent reading with last read reading.
  3. There will be data for equipment operating standards, for example, Diesel consumption per hour(CPH) like 10lltrs per hour, Power rating like 25KVA etc.

Semi structured data

  1. Most of the countries of the world still use fossil fuel for redundant power source like Diesel generator. So, most of the service provider has a tie up with fuel distribution agency to fill the DG genset at a frequency. They follow a beat plan. These agencies send the fuel filling data to the maintenance agencies for reconciliation with actual consumption.
  2. The electricity distribution company has a data exchange facility for invoice details for electricity consumed with the operators.

And SO ON….



Each sensor on the site creates data every second. With thousands of sites and hundreds of sensors at one site, the velocity of data is huge. Although, data size may be small, but it needs to be brought. All those data must be brought to the main data centres


With the above figure, you can assume the volume, if data has to be correlated and analysed for a multiple months and year. Sometime, reconciliations of different party’s cost take month and years in case of dispute. These data are required to stored and processed to derive intelligence.




Sometime, you want to take social media data on recent incidents in the area and want to provide the extra protection of the site in order to reduce any probability of site bringing down by the violent mobs. The moment, this is brought into the system, the veracity of big data analytics starts playing.


On top of these, the active elements data that is usually monitored through network operation centers of Service providers,  can be clubbed to provide a much needed full operation view and key performance indicator in real time to the service providers.

What Problem can we solve with Big data analytics


With so much of data at such speed coming to your system, the following key performance can be achieved.

  1. Site Uptime and operational excellence in real time
  2. Equipment health monitoring and Preventive maintenance of equipment
  3. Analytics and Data based decisions
  4. Continuous improvement in energy source selection based on current and past data
  5. Field behaviour analysis to stop fossil fuel pilferage
  6. Financial budget allocation and cost control


Big data solution

Following solutions can be implemented to provide a real time data feed for operational excellence as well as business intelligence for cost based decision making and organization wide policy formulation.


  1. Use map reduce function to derive data the equipment failures. For example, temperature of DG is shooting high from its normal range now. Schedule a preventive maintenance independently.
  2. It’s hot summer at 45oC and cabinet temperature is 32oC which means, raise alarm and do a breakdown maintenance.
  3. Analyze the battery recharge level and raise alarm for non-charging cells.
  4. scheduling the preventive and breakdown maintenance.
  5. performance of people, process and technology.

 And many more for operations excellence.


Some of the Analytics can be done to analyse the outliers in energy consumption particularly the fossil fuel consumption. For example,  the difference in predicted value from actual and standard fuel consumption per hour(CPH). Ideally, all the value should be zero but it never happens. The outliers should be seen carefully and acted upon.

Similarly, tower sites with multi operator tenancy with standard CPH can be analysed and decision can be made to review those sites’ operational and people performance to streamline operation.



Telecommunication Tower site management is a perfect case for utilizing big data analytics based on IOT devices fitted on tower sites deployed in far reach corners of service provider’s operating area. The operations excellence can be achieved only when all the data can be analysed in real-time using big data analytics tools and technology. Use of Apache SPARK with Python and using Python Pandas for plotting the graph can be termed as a set of technological implementation for big data analytics. Needless to say, you need to use the python statistical library  for regression and trend analytics.






Go Back


Blog Search


Blog Archive