Dimensioning a mobile network benchmarking campaign

It starts with the questions:

• How many measurement systems and teams are needed?
• How many measurement days should be planned or how many measurement km driven?

Since these questions do not have precise answers, we can provide certain guidelines and methods. The R&S white paper: “How to dimension a mobile network benchmarking campaign“ describes the dimensioning of quarterly benchmark assessments in Germany using a population-centric approach. This entry summarizes the main ideas in the white paper.

Demographic information for Germany is needed as an input – this can be the publicly available population statistics from Wikipedia. The NPS methodology sets minimum sample requirements and selects and weights geographical categories. This allows us to assume that the measurement samples are appropriately distributed. But let us take a step-by-step look:

1. Demographic information

Publicly available information shows the total German population and its distribution among major cities. To cover 25% of the population (a common target for the telecommunications industry), the top 14 cities in Germany are selected and the time and distance needed to reach all these cities are calculated.

2. Geographical categories and weighting

When using the NPS methodology suitable geographical categories must be selected that are representative for the target country. The relevance of different categories is defined and weighted for the NPS calculation. In our example we defined the following categories and weighting:

3. Sample requirements

The NPS methodology requires1000 samples for statistical relevance. This requirement applies to each NPS subset in the final report. We want a statistically relevant result for each of the three categories. We also assume the samples are distributed in line with the category weighting above. This is not a hard NPS requirement but is handy for dimensioning and has the advantage of generating KPI results for the various categories with the same weighting as the NPS score. Please see the Rohde & Schwarz white paper: "Network Performance Score (NPS)” for more information.

4. Calculations

These factors need to be combined and some basic assumptions applied to determine the dimensioning for our benchmarking assessment.

A total distance of ~2300 km is needed to cover the cities. Since other cities and towns will be selected, an estimate is added to determine the total kilometers and hours in the road category and the total number of samples collected.

The duration calculation uses the distance in km and average estimated measurement speed for the category (50km/h).

The total number of samples can be calculated for the road category in line with the initial required sample distribution using the NPS weighting. The estimated samples for the road category represent 15% of all NPS samples. Knowing the total number of samples now allows all the other figures to be calculated along with the category weighting and estimated driving speeds. These can be seen in the table below:

To calculate samples, duration and distance use the following formulas (calculations can work in both directions as needed):

• duration = #samples / 18 (18 samples per hour based on the call window of 180 seconds and general experience)
• distance = duration/average duration (per geographical category)

Since the total measurement scope for the remaining city and town categories have been defined and must be distributed among actual cities, the preferred method is to distribute by relative city population. While the list of 14 major cities generally remains fixed, the selection of other cities and towns can vary from quarter to quarter for improved the coverage throughout the year.

Conclusion

Quarterly benchmarking can be done with two measurement teams in ~30days. The assessment delivers a relevant and statistically stable NPS score for Germany and allows for the measurement of quarterly trends.

The full detailed guideline can be found in the Rohde & Schwarz white paper: “How to dimension a mobile network benchmarking campaign”.