Abis Network Design This application pack is useful for network design validation and troubleshooting quality and interference problems. It contains a number of reports to help pinpoint the severity and location of poor quality measurements.

The problem Extensive drive-tests are typically used to check the validity and quality of a GSM network design or frequency plan. The advantage here lies in indicating the precise geographical location of problems when they occur. However, the main disadvantage of drive-tests is that they give a snapshot view of the network at a very specific point in time (while the network may not be operating at its normal usage level), and give no indication of the reproducibility of a problem.

Suppose the drive test shows an area with very low coverage: is this problem significant in terms of traffic affected, or do only one or two customers encounter this problem? If a quality problem has been detected while drive-testing a cell, is it because the wrong cell was selected due to particular conditions during the drive-test (temporary congestion, for example) or do all customers experience similar problems in the cell?

Moreover, drive-test data only provides detailed information on the downlink. Uplink data is limited to some information as sent by the mobile, but no information is available to indicate how the base station receives this information.

The solution It would not be a wise engineering practice to start adjusting cell and HO parameters based on a specific drive test. It is vital that a specific problem be reproducible and statistically representative before any network action is taken. Adjusting network parameters on a sporadic problem, while bringing a possible local solution, may degrade the overall quality of the network. Only a more global approach can help the engineer validate a problem encountered during a drive-test and identify its cause.

Abis is the ideal data source for level, quality and interference analysis. It provides a global view of the cells under investigation, while providing detailed reports on all mobiles (calls) handled by these cells during the recording period. Moreover, Abis traces provide synchronized information both on the uplink and downlink.

By carefully selecting a recording time—including the network-critical periods—engineers can have at their disposal a set of a data from which statistically representative problems can be extracted and analyzed.

Quality Distribution Counts of uplink and downlink quality, and graphs of downlink quality vs. downlink level and uplink quality vs. uplink level.

The quality distribution histogram gives an overview of the cell behavior and helps validate frequency planning. A ‘sane’ cell should show a distribution of quality samples smoothly decreasing while quality decreases.

The bubble charts showing quality against receive level can help identify interference problems. Typically, interference will appear as a peak of bad quality with a relatively high level (above –90 or –85 dBm). A ‘sane’ cell would normally show good quality until the level becomes quite low (depending on the environment and network design, but usually –100 dBm), then rapidly yet regularly degrading quality at lower levels.

Poor Quality Contributors Poor qualities for uplink and downlink plotted against corresponding level.

The bubble charts in this report compare the level and timing advance distribution of poor quality samples against the distribution for all measurement samples. This indicates at what distance and level poor quality is occurring, and the proportion this represents of the traffic at that point.

Level and Interference Interference vs. downlink Rx level followed by interference definition chart..

This chart shows the distribution of level measurements. It compares overall RxLev distribution with the distribution of interfered measurements. Interference is implied by a combination of RxQuality and RxLevel, defined below the chart.

The Rxlev distribution plot can also be used to validate power control settings. The aim of power control is to maintain the RxLev of the link in a relatively small window where quality criteria can be easily met. Its advantage is to save battery life on the mobile side, since only the minimum necessary power will be used. To a lesser extend, it can help reduce interference, although rendering it more ‘bursty’.

If power control is working properly, the RxLev distribution will show a clear statistical concentration of data samples between the trigger values defined for the power control window. Moreover, if the power control parameters have been chosen correctly, the corresponding quality in the power control level window should be very good.

Timing Advance and Interference Interference vs. timing advance followed by interference definition chart.

This chart shows the distribution of timing measurements. It compares overall timing advance distribution with the distribution of interfered measurements.

Interference is implied by a combination of RxQuality and RxLevel, defined below the chart. It is useful for identifying the location of island coverage or coverage holes, and for assessing the quality of service in those locations.

A space distribution plot of RxLev samples (RxLev vs. TA) is usually a safe source of information when adjusting antenna tilts. Actually, excessive tilt typically shows up on such a plot as a peak of higher RxLev clearly separated from the main set of data. This is due to the secondary lobe of the antenna radiating below the horizon after the antenna has been tilted.

When adjusting antenna tilts, it is a good idea to record a trace of the cell before the adjustment to serve as a reference. All islands are not necessarily related to secondary lobe problems. Other elements, such as repeaters or even terrain configuration, can lead to similar results. But indications of terrain- or repeater-related coverage islands remain fixed, whereas the secondary lobe problems significantly change location when the tilt is adjusted.