Return on Investment (roi) for Multi-Technology son juan Ramiro, Mark Austin and Khalid Hamied

A.5.4. Impact of Using PD Measurements

Previous results are focused only on events reporting at least three different sites. However, two-site events can be geo-located by combining PD and OTD measurements. The use of PD is reliable when it is close enough in time to the call establishment. In this case a time-window of 20 s is considered. The CDF of the geo-location error (a) and the event status summary (b) for the European scenario are depicted in Figure A.6. As expected, only a small percentage of events (less than 40%) can make use of the reported PD, which also shows poorer accuracy than events reporting at least three different sites. This means that the benefit of PD measurements is limited.

A.6. Concluding Remarks

This Appendix evaluates a time-delay based geo-location algorithm providing promising results in both accuracy and percentage of located events in two different real UMTS networks

1. 
   (b)
   

0
0 50 100 150 200 0 25 50 75 100
Error [m] Frequency [%]
Figure A.5 CDF of the geo-location error (a) and event status summary

2. 
   for the European scenario comparing results when enabling and disabling the event filtering. This analysis only includes events reporting at least three different sites.
   

(a) (b)

0
0 50 100 150 200 250 300 0 10 20 30 40 50 60 Error [m] Frequency [%]
Figure A.6 CDF of the geo-location error (a) and event status summary (b) for the European scenario considering only two-site events.
Appendix A: Geo-Location Technology, UMTS 271
[10]. Between 65–80% of the events are geo-located with an error lower than 150 m at the fiftieth percentile. The algorithm performs slightly worse in the selected US scenario than in the European one that was analyzed, which can be explained by the fact that the former includes noticeably less reported sites per event, and such events are also worse conditioned. The use of filtering prior to geo-location is proven to be beneficial since it accurately discards events that are more likely to be erroneously geo-located. It is also shown that there are long periods of time with almost total absence of MRs, which makes proper traffic estimation in those areas more difficult. Finally, the study illustrates that there is a large percentage of twosite events, which cannot be directly geo-located. However, the use of PD measurements has a limited benefit, which leaves room for improvement using, for instance, a combination of the aforementioned techniques with RSL measurements.