Improving Mobile Network Benchmarking using Physical Layer Indicators
Maintaining, troubleshooting and performance evaluation in live networks is mainly accomplished with drive-tests. For this purpose dedicated vehicles are used which have the drawback of high costs. An alternative are autonomous systems which are built into several kinds of vehicles (e.g. cabs) that are roughly driving randomly through a given area of interest. In both cases, the number of calls are for the most part a critical economic factor in terms of cost effectiveness and should be as low as possible for a given statistical accuracy of the measured values.
In this work methods should be found and developed which are utilizing the knowledge of previous measurement campaigns to estimate quantities that can be statistically described by binomial distributions, e.g., call drop rates, service availability or network accessibility ratios as defined by ETSI. These methods should deliver results which are very close to the ones obtained by common drive-tests. The most important advantage will be that even for a very low number of test calls there will be the possibility to give an estimate of a binomial quantity (e.g. the call drop rate). A starting point for this work will be an analysis of physical layer measurements of GSM and UMTS networks.
Work items include:
- Become familiar with UMTS/GSM live network performance measurements
- Statistical evaluation of measurements
- Find correlations between measurements and KPI (Key Performance Indicators)
- Optimization of benchmarking methods
|Subject area||UMTS, GSM, QoS, Benchmarking|
|Type of work||60% Theory, 40% Software|
|Supervisors||Oliver Nipp, Dr. Marc Kuhn|
|Professor||Prof. Dr. Armin Wittneben|