These days, all the main operators and third parties responsible for the network in the Telecom sector run both preventive maintenance and corrective maintenance models. If you get the right information at the right time, the former—and more disruptive—model can break old trade-offs, and assure lower operation costs and higher network stability to not only electric systems, but also business intelligence systems. In doing so, it also allows efforts and investment to be focused on innovation developments.
Following on from earlier planning, based on defined time intervals or according to pre-established criteria, this approach uses principles of statistical process control to determine at what point in the future maintenance activities will be appropriate. Such techniques are designed to help determine the condition of the equipment and predict corrective maintenance needs, thus reducing the risk of failure or decreased performance of the equipment.
It can, however, reveal inefficiencies because there are still risks of unexpected stops or breaks, requiring intensive labour, which can cause system unavailability, low productivity and revenue losses due to the network maintenance. Once applied, this approach can show that particular equipment has a high probability of suffering damage and needs to be replaced, which forces the Telecom Operator to shut down the network and invest significant resources to correct the issue.
On the other hand, corrective maintenance, also known as breakage maintenance, only intervenes to repair equipment when it stops working. In other words, it is a reactive approach, with potentially higher operating expenses due to the periodicity and the elevated costs of manual labour that are required. Poor activity planning can, in addition, generate the paralysis of several services and impact revenue directly. This strategy may be more cost-effective, but only until a catastrophic failure occurs—and this can happen.
Both types of maintenance strategy have advantages and disadvantages, and both tend to use obsolete working methodologies, which can sometimes lead to a lower quality of service (QoS), a reduction in equipment life, and system unavailability (downtime).
Learning how to get the timing and planning right
The predictive maintenance approach measures historical and real-time data from the Network Elements to understand the process of service degradation before failure. It also predicts which Network Elements are more likely to fail in the upcoming days or hours, using predictive analytics tools and techniques.
It can, therefore, work as a periodic inspection of the equipment’s conditions, but in an automatic and self-learning way—determining in advance the need for maintenance services, increasing the equipment’s availability time, and reducing the amount of unplanned emergency work. It can also increase the confidence level in equipment performance by predicting the probability of failure, as well as the useful lifetime expectancy of the equipment and the conditions needed for maximizing that time. To do that, predictive maintenance demands a solid information platform and network reading systems.
Putting in place an effective predictive maintenance strategy
Predictive maintenance should be able to offer:
- Real-time analysis: Ability to analyze real-time network monitoring data
- Recurrent analysis: Ability to analyze recurring events, fault indicators and their solutions
- Pattern recognition: Ability to analyze historical data and identify patterns associated with failures or other events
- Operations optimization: Advanced analytical foundation for optimizing the operations planning.
To put in place a successful process, Telecom operators need to start collecting all historical and real-time data from operations and maintenance services and processes and transform that data in order to understand important behaviours such as average, mode, standard deviation, % of recurrences, etc.
The next step is to build an advanced analytical engine that can do all the deduction and identification of discriminatory and probabilistic variables by using multiple methodologies and statistical models. Once that is in place, the Predictive Model will be capable of understanding, predicting and avoiding downtimes and flaws in the telecommunication network.
Below is a simplified functional and logical architecture of this methodology:
By applying knowledge gathered about Telecom and Engineering operational reality, we can give some practical examples of possible predictive correlations (non-exhaustive):
- VSWR: Due to connector problems caused by either installation or weather conditions, service failures and equipment damage may occur. This can be predicted by analyzing the broadcasted information from dumps and RNCs, and establishing correlations with throughput decrease
- Temperature: The increase in temperature levels can make equipment crash and damage system components. To predict this event, you must analyze the broadcasted information from dumps and RNCs and establish correlations with processing time increases
- Battery tension: With an established pattern, it is possible to have a good understanding about when battery storage becomes obsolete and needs to be replaced. By measuring the average tension of the database, we can establish a pattern of decreasing energy storing capability
- Bitrate transmission error: Defining error bitrate patterns allows for the setup of patterns that precede a transmission equipment crash. By accessing the transmission equipment’s management, it becomes possible to set the BER variation patterns.
All in all, predictive maintenance techniques have a direct impact in network operation and maintenance, with several operational benefits in main scope areas, such as:
- The system remains active for longer periods because the equipment stays available for a longer time. That leads to an improved service quality for clients and assures higher revenues from a greater system activity period
- System reliability increases due to a higher trust level in equipment performance, allowing maintenance and the resolution of equipment problems to take place much sooner
- It reduces unplanned emergency work because the maintenance is only performed when necessary, establishing diagnoses and performing trend analysis
- It maximizes depreciation by taking advantage of the lifespan of equipment components and preventing the increase of physical damage in the equipment’s electrical components, thus reducing maintenance and offline period costs.
Predictive maintenance can break the trade-offs of the older strategies by enabling telecommunication companies to maximize the useful life of their equipment while avoiding unplanned downtime, minimizing planned downtime, and saving costs.
In essence, predictive maintenance is a disruptive innovation because, in addition to the operating benefits, it also helps the company to act in regions with high client churn rates, and where they suffer with continued falls in the network.
Mathias Hubert is currently Manager for Strategy & Operations projects in Beijaflore Consulting Group in Brazil, and specializes in the Telecommunication Media sector.
Togo Ribeiro is currently Manager for Telecom Engineering projects in Beijaflore Consulting Group in Brazil, and specializes in the Telecommunication Media sector.