Print This Page

Sub Working Group 3.1 - Measurements Based Optimisation for Mobile Wireless Nets

Chairman: Dr. Maciej J. Nawrocki, Wrocław University of Technology, Poland

The SWG3.1 focuses on mobile wireless network optimisation aspects with special attention to the data used as an input to optimisation process. Most of automated RF network optimisation tools available on the market (with some recent exceptions of GSM automatic frequency allocations tools) are based on extensive simulations where predictions become the major input to optimisation routines. Operator requirements seem to be much higher than the quality offered by those kinds of tools. Network models together with simulated environment models (incl. propagation predictions) are the major limiting factors. The natural solution is to substitute artificially generated data by the measured data taken from the real system.

Measured data can be taken from network management system in form of mobile terminal aggregated reports as well as directly from RNC (with use of appropriate protocol analysers). Measurements can be also taken by external equipment e.g. drive tests tools.

Above process comes with advantage of high accuracy of input data however also with some disadvantage which is low data diversity compared to predictions since virtually every parameter and characteristic can be simulated but not always measured. Therefore, research within SWG3.1 concentrates on identification and selection of proper measurements. The main idea can be expressed in one sentence: "It is better to have simple but accurate optimisation than sophisticated and inaccurate", however some measurement oriented optimisation-like activities are surprisingly advanced (e.g. autotuning).

It seems to be reasonable not to concentrate on one system only but be able to conduct research for several different technologies. Thus, this SWG concentrates on cellular networks (UMTS, HSPA, GERAN), WLANs as well as WMANs since all B3G systems will be increasingly complicated and this kind of automatism (i.e. accurate) will be really required to make networks fully operational.

The SWG3.1 pillars of activities include:

1.   Measurement data acquisition, definition and processing:
-    network performance criteria development,
-    measurement data acquisition (definition, sources, availability, techniques, soundness etc),
-    data filtering and processing (noise reduction, long-term effects, classification, correlation, grouping), model-based filtering,
-    measured data separation techniques.

2.   Optimised parameters:
-    optimization parameters definition and selection,
-    parameter correlation with counters/quality indicators.

3.   Algorithms, models and tools:
-    network modelling with use of measured data,
-    measurement based network model tuning,
-    optimization methods (including problems with no response in optimization loop),
-    algorithms implementation (tools).

4.   Measurement vs. model based optimisation and associated reference scenarios:
-    comparison between measurement and simulation based optimisation approaches,
-    reference scenarios definition for comparison process,
-    field tests - verification with practical trials.