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

Figure 7.18 Assumed traffic patterns for voice and data.
1 2 3 4 5 6 7 8 9 10
1 2 3 4 5 6 7 8 9 10
Year Year

2000.00
1800.00
1600.00
1400.00
1200.00
1000.00
800.00
600.00
400.00
200.00
0.00
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Year Year
Figure 7.19 Carrier, E1 link, capacity site and RNC additions that are required to serve all the offered traffic.
ROI for Multi-Technology SON 259

Figure 7.20 Extra revenue per year due to reduced churn.

	Year 1	Year 2	Year 3	Year 4	Year 5	Year 6	Year 7	Year 8	Year 9	Year 10
CAPEX reduction	0.98	2.68	6.97	11.11	17.62	17.24	20.81	19.03	20.60	20.81
Carriers Site facilities RNCs	0.83 0.16 0.00	1.87 0.72 0.10	4.31 2.56 0.10	6.02 4.79 0.30	8.81 8.51 0.30	8.08 8.76 0.40	9.49 10.92 0.40	8.59 10.14 0.30	9.29 10.91 0.40	9.57 10.85 0.40
OPEX reduction	6.55	7.62	9.31	10.95	12.54	13.70	11.57	9.26	8.60	3.70
Fewer E1	6.55	7.62	9.31	10.95	12.54	13.70	11.57	9.26	8.60	3.70
Extra revenue	6.86	8.23	9.87	11.85	14.22	17.06	20.48	24.57	29.48	35.38
Lower churn	6.86	8.23	9.87	11.85	14.22	17.06	20.48	24.57	29.48	35.38
Total (€ MEUR)	14.39	18.53	26.15	33.92	44.37	48.00	52.85	52.87	58.68	59.90

Figure 7.21 Cash flow projection for Self-Optimization.

Figure 7.22 ROI breakdown.
260 Self-Organizing Networks
Table 7.6 NPV of Self-Optimization versus traffic growth

Relative traffic growth pattern (%; years 2–10)	NPV (€ million)
[10, 10, 10, 10, 10, 10, 10, 10, 10]	101.77
[20, 20, 20, 20, 20, 20, 20, 20, 20]	138.36
[30, 30, 30, 30, 30, 30, 30, 30, 30]	175.17

7.5. Case Study: ROI for Self-Healing

The benefits from Self-Healing can be categorized in two main areas:

1. 
   OPEX reduction due to reduced workload through automation of monitoring and t roubleshooting tasks that were also being carried out when the network was managed manually. In this respect, field experience has shown that, by means of smart a utomation, the time to find the root cause for the 50 worst-performing sectors in a 2G/3G network with 2000 sites can be reduced up to 90%.
   
2. 
   Extra revenue due to improved QoS through faster resolution of network incidences, which leads to reduced churn.
   

7.5.1. OPEX Reduction through Automation

The method for computing annual, recurrent OPEX savings (OPEX_Savings) due to reduced workload can be simply summarized by means of the following equation, assuming that troubleshooting activities are normally carried out on a daily basis:
OPEX_Savings = Num_Sites ⋅
NumSitesPerEngineer (7.26)
· I · OPEXPerEngineerAndDay · NumWorkingDaysPerYear · ReductionWorkload
where Num_Sites is number of sites; Num_{SitesPerEngineer} is the number of sites that every engineer is assigned to monitor and troubleshoot on a daily basis; I is the job intensity (100% for a f ull-time job, 50% for a part-time job, etc.); OPEX_{PerEngineerAndDay} is the daily OPEX per engineer; Num_{WorkingDaysPerYear} is the effective number of days in which noncritical maintenance and monitoring activities are carried out; and ^Reduction_Workload is an efficiency factor describing the percentage of workload that can be reduced through automation.

7.5.2. Extra Revenue due to Improved Quality and Reduced Churn

The same model that was introduced in Section 7.4.5 can be used.
ROI for Multi-Technology SON 261
Table 7.7 Assumptions for the ROI assessment of Self-Healing

Network size Number of sites								10000
Cost structure for OPEX Number of sites per engineer								100
Job intensity (100% for full time job; 50% for part-time job)								50%
OPEX per engineer and day								€300
Number of working days per year								52 × 5 = 260
Self-Healing impact Workload reduction factor (in line with field observations)								90%
DCRManual								0.7%
DCRSON								0.5%
Subscribers, ARPU and churn Number of subscribers								20 million
Monthly ARPU								€10
Churn rate								10%
Percentage of churners due to bad network quality								10% (out of the 10% above, i.e. 1%)
		Year 1	Year 2	Year 3	Year 4	Year 5	Year 6		Year 7	Year 8	Year 9	Year 10
	OPEX reduction	3.51	3.51	3.51	3.51	3.51	3.51		3.51	3.51	3.51	3.51
	Automation	3.51	3.51	3.51	3.51	3.51	3.51		3.51	3.51	3.51	3.51
	Extra revenue	6.86	6.86	6.86	6.86	6.86	6.86		6.86	6.86	6.86	6.86
	Lower churn	6.86	6.86	6.86	6.86	6.86	6.86		6.86	6.86	6.86	6.86
	Total (€ million)	10.37	10.37	10.37	10.37	10.37	10.37		10.37	10.37	10.37	10.37

Figure 7.23 Cash flow projection for Self-Healing.

7.5.3. Sample Scenario and ROI

The assumptions for this sample calculation are presented in Table 7.7. Based on the models presented in Section 7.5.1 and Section 7.5.2, the annual OPEX savings due to Self-Healing is €3.51 million and the annual extra revenue is €6.86 million. Adopting a time horizon of 10 years, and applying the same methodology that was used for Self-Planning and SelfOptimization, the cash flow projection in Figure 7.23 can be built. Assuming a discount rate of 20%, the resulting NPV (described in Equation 7.1) is €43.48 million.

7.6. References

1. 
   Ferreira, J., Vellasco, M., Pacheco, M. and Barbosa, C. (2004) Data Mining Techniques on the Evaluation of Wireless Churn, European Symposium on Artificial Neural Networks Bruges, Belgium, 28–30 April 2004.
   
2. 
   Pendharkar, P. C. (2009) Genetic algorithm based neural network approaches for predicting churn in cellular wireless network services, Expert Systems with Applications: An International Journal, 36(3), pp. 6714–6720.
   
3. 
   J.D. Power and Associates Reports (2007) Call Quality Plays an Increasingly Important Role in Customer Satisfaction With the Wireless Phone Experience, 19 April 2007, http://www.prnewswire.com/news-releases/ jd-power-and-associates-reports-call-quality-plays-an-increasingly-important-role-in-customer-satisfactionwith-the-wireless-phone-experience-58607222.html (accessed 3 June 2011).
   
4. 
   Lawson, C. L. and Hanson, R. J. (1974) Solving Least Squares Problems, Prentice-Hall, New Jersey.
   

Appendix A

Geo-Location Technology for UMTS
Carlos Úbeda

A.1. Introduction

The use of geo-located Measurement Reports (MRs) has been proven to be useful in applications like traffic map generation or accurate radio propagation modeling. A first approach is to use Received Signal Level (RSL) measurements, which are widely and easily available. However, distance-dependency is measured with very high intrinsic uncertainty leading to poor geo-location accuracy, and therefore current trends support the use of timedelay m easurements [1, 2] or a combination of both [3]. Some studies [4, 5] also claim that tracking techniques instead of geo-location of single events provide better accuracy.
Geo-location algorithms based on time-delay measurements make use of the so-called Observed Time Differences (OTDs) [6] reported every time an MR is triggered. OTDs are subject to a certain number of constraints that will affect the accuracy when geo-locating the User Equipment (UE), such as multipath propagation, limited number of measured Base Transceiver Stations (BTSs), nonperfect BTS synchronization recovery, measurement errors and MRs not being sent in a continuous way but in an event-driven fashion.
Although there is wide literature [1-5] on geo-location and tracking algorithms for mobile networks, those studies mostly analyze theoretical scenarios using unrealistic tracking paths and too optimistic sets of MRs. This Appendix quantifies the effect of the aforementioned limitations in real networks, proposes several techniques to mitigate them, and finally a nalyzes in detail their impact on the geo-location accuracy.
This Appendix is organized as follows: Section A.2 briefly describes the use of OTDs applied to geo-location. Section A.3 provides a detailed description of the geo-location algorithm. Section A.4 states the assumptions used to assess the algorithm performance.
Self-Organizing Networks: Self-Planning, Self-Optimization and Self-Healing for GSM, UMTS and LTE, First Edition. Edited by Juan Ramiro and Khalid Hamied.
© 2012 John Wiley & Sons, Ltd. Published 2012 by John Wiley & Sons, Ltd.
Section A.5 evaluates the performance of the proposed algorithm under different scenarios.
Finally, A.6 summarizes the main conclusions from this Appendix.