Sensitivity Analysis

A sensitivity analysis of the TCO involving the five proposed backhaul architectures was conducted to validate the results. In this analysis, the three basic backhaul technologies were divided into three types: copper, fiber and MW. The most significant cost parameter (i.e., the one having the largest impact on the TCO) was determined for each category. This parameter ranges between -30% and +30% with regard to the original market value (as shown in Table 3.1) so that the impact on the TCO can be evaluated.

With regard to the copper based backhaul, the most relevant cost derives from the CAPEX and this is the DSL modem cost (categorized as Equipment cost in Fig.

3.3). The cost of the DSL modem ranges from 70% to 130% of its market value (as illustrated in Table 3.1) and no significant change was detected in the total TCO of the

77 backhaul architectures. As a result, it can be concluded that varying the DSL modem cost does not affect the conclusions drawn in the previous sections.

With regard to the backhaul based on fiber, the most relevant cost parameter depends on whether the operator is deploying its own fiber infrastructure or is leasing the dark fiber. In the case of the former, the main cost involves the CAPEX and comes from the fiber trenching cost (i.e., Infrastructure cost with reference to Fig. 3.3). Fig.

3.14 shows the results obtained by varying the fiber trenching cost from 70% to 130%

of the market value as shown in Table 3.1. It is clear that when the fiber trenching cost for Architecture 3 is reduced, it becomes the most attractive solution in financial terms (Architecture 3 becomes more cost-efficient than Architecture 1 when the fiber trenching cost is reduced to 91% of its original value). However, Architectures 4 and 5 always remain more expensive. Moreover, increasing the fiber trenching cost does not affect the conclusions drawn in the previous section.

If the operator leases the dark fiber, the most relevant cost involves the OPEX and is the fiber leasing cost (i.e., Spectrum and fiber leasing cost with reference to Fig.

3.3).

Figure 3.14: Sensitivity analysis of variations in the trenching cost in a range of 70% to 130% of the market price.

78 Fig. 3.15 shows the results obtained by varying the fiber leasing cost from 70%

to 130% of the market value (as shown in Table 3.1). It can also be noted that reducing the fiber leasing cost does not significantly affect the conclusions drawn in the previous section. In fact, Architecture 3 remains the most attractive cost solution. The only observable change is that Architecture 5 is found to be more cost-efficient than Architecture 1 if the cost for fiber leasing is reduced to 72% of its original value (but only in the case where there is a maximum distance of 300m between the femto cells and local exchanges).

Figure 3.15: Sensitivity analysis of variations in the dark fiber cost within a range of 70% to 130% of the market price.

With regard to the MW based backhaul, there are two main items of expenditure for the TCO. The first is the cost of the antenna and this is related to the CAPEX (and categorized as an Equipment cost in Fig. 3.3). Accordingly, the cost of the antenna ranged from 70% to 130% of its market value and the results for the TCO of the backhaul architectures are shown in Fig. 3.16. In specific terms, Fig. 3.16(a) shows the results if the operator trenches the fiber, while Fig. 3.16(b) shows the results if the operator leases the dark fiber. Fig. 3.16(a) shows that decreasing the antenna cost does not change the conclusions that have been drawn and that Architecture 1 remains the most cost-efficient solution. However, increasing the antenna cost to 126% of its

79 original value makes Architecture 3 the most attractive solution from an economic standpoint. Fig. 3.16(b) shows similar trends. In particular, reducing the cost of the antenna does not affect the results and Architecture 3 remains the most cost efficient solution. However, increasing the antenna cost to over 103% of its original value makes Architecture 1 more expensive than Architecture 5 (when there is a distance of 300m from the femto cells to the local exchange). In addition, increasing the antenna cost over 124% of its original value also makes Architecture 4 more cost-efficient than Architecture 1.

The other expenditure which has a considerable impact on the MW based backhaul is the spectrum cost, and this is related to the OPEX (i.e., Spectrum and fiber leasing cost with reference to Fig. 3.3). Fig. 3.17 shows the results obtained by varying the spectrum cost from 70% to 130% of its original market value. Only the case with fiber leasing was taken into account because in the case with fiber trenching the conclusions drawn in the previous sections remained unchanged. Fig. 3.17 shows that reducing the spectrum cost does not affect the conclusions, i.e., Architecture 3 remains the most cost-efficient. On the other hand, increasing the spectrum cost by over 28% of its original value makes Architecture 5 more cost-effective than Architecture 1 (a case with a distance of 300m between the femto cells and local exchanges).

80 (a)

(b)

Figure 3.16: Sensitivity analysis of the antenna cost when it varies from 70% to 130%

of the market price. (a) The impact when the operator trenches and owns the fiber infrastructure. (b) The impact when dark fiber is leased.

81 Figure 3.17: Sensitivity analysis of the MW spectrum cost when it varies from 70% to

130% of the market price.