Analysis and discussion

The resulting priorities of the alternatives are illustrated in Table 6.5. From the Normalised column, Satellite technology is the most preferred alternative and has the highest score in this AHP model with a priority of 29.21%. Microwave comes next with a priority of 28.59% and then Fibre and Power line technologies with 22.07% and 20.08% respectively. The sum of the priorities in this column is equal to one. This complies with the AHP procedure and demonstrates that its steps are applied properly.

Table 6.5 Synthesized priorities for the alternatives

Priorities

Alternatives _{Normalised Idealised}

G1 Fibre optic cable 0.2207 0.7556

G2 Power line communication 0.2008 0.6874

G3 Microwave links 0.2859 0.9788

G4 Satellite communication 0.2921 1.0000

The idealised column uses a normalisation by dividing the score of each alternative in the normalised column by the highest alternative score (0.2921). Hence, Satellite has a priority of 100%, the other priorities are in the same proportion as in the Normalised and so Microwave, Fibre optic cable and Power line are 97.88%, 75.56% and 68.74% respectively. Moreover, from Table 6.3, one can observe that the comparison of the criteria with respect to the goal yields that the technical criterion has the highest priority of 29.31%, expressing a certain advantage among others, which indicates more importance of the technical aspects in comparison to other economic, infrastructure, etc. factors.

The lowest priorities are for social, environmental and regulatory aspects with 6.90%, 5.84% and 5.63%, respectively. In Table 6.6, the subcriteria are arranged for ranking in a descending order of their global priorities. The table shows that the most important subcriterion among all is C1 ‘Operating cost’ with a priority of 11.47% followed by B1 ‘Coverage’ with 9.20% and C2 ‘Funding’ with 6.37%. The top ten factors comprised a variety of subcriteria that belong to all criteria except the regulatory criterion. The least important subcriteria among others considered in the model with priorities of less than 1% are F1 ‘Climatic conditions’, A9 ‘Latency’, B3 ‘Proposed usage’, E1 ‘Spectrum’ and D1 ‘Demand’ with 0.93, 0.70, 0.62, 0.40, and 0.28, respectively.

From the above analysis, one can observe that the AHP is capable of structuring the problem and providing a systematic approach to decision making. It allowed for diverse qualitative factors to be examined in a mathematical model, which can help to reduce the time needed to evaluate the alternatives. By using the traditional selection process in such

problems, the decision may take months to be reached. As the criteria are clearly defined and the problem is structured systematically, the AHP allows the decision makers to visualise the strengths and weaknesses of each technology alternative by comparing their scores against each factor.

Table 6.6 The ranking of factors according to their global priorities

Rank Subcriteria Global priorities (%)

1 C1 Operating cost 11.47 2 B1 Coverage 9.20 3 C2 Funding 6.37 4 A1 Reliability 6.14 5 A2 Ease of maintenance 5.81 6 A7 Bandwidth 5.42 7 F1 Terrain topography 4.91 8 B6 Remoteness of area 4.26 9 D4 Community of interest 4.06 10 C3 Capital cost 3.99 11 E3 Rights of way 3.92

12 A3 Remote network management 3.58

13 A5 Ease of installation 3.49

14 C4 Return on investments 3.48

15 B8 Parallel infrastructure 3.34

16 B2 Security of physical infrastructure 2.29

17 C5 Economic development of area 2.26

18 B5 Access to existing telecoms infrastructure 2.06

19 A4 Compatibility 1.93

20 B4 Availability of skilled technicians 1.58

21 D3 Population density 1.47 22 B7 Rollout time 1.38 23 E2 Licensing 1.31 24 A6 Scalability 1.15 25 A8 Flexibility 1.09 26 D2 Affordability 1.09 27 F2 Climatic conditions 0.93 28 A9 Latency 0.70 29 B3 Proposed usage 0.62 30 E1 Spectrum 0.40 31 D1 Demand 0.28

The obtained final scores (weights) provide information (to contemplate explicit or implicit knowledge) about the alternatives and the way they are used to satisfy the selected factors, as well as the importance of these factors in order to reach the goal of the model. Taking this into consideration, a result where one can affirm which alternative is more preferable from telecoms experts’ point of view is reached. However, the priority scores of the four technologies are actually quite close to each other and although Satellite technology achieved the highest score, it is only above Microwave’s score by less than 1%. Also, Fibre optic’s score is just less than 2% higher than that of Power lines. Hence, it becomes questionable for the decision makers to arrive at a consensus decision to select either of the

alternatives. This outcome implies that while the AHP method has been studied extensively and used in numerous MCDM applications. However, the simplicity of the hierarchical structure and linear unidirectional hierarchical relationship among criteria and subcriteria in the AHP method hide important issues, such as interdependence among qualitative factors and interaction among decision making levels and so oversimplified the problem.

In AHP, a hierarchy considers the distribution of a property (goal) amongst the element being compared and judges which element has a greater influence on that property. The author believes that there is recognition that better ways of defining interactions are needed; the AHP is limited, as most interactions are currently identified by reducing them to pairwise sets between factors at the same level of the hierarchy. Whereas in actuality, in general, and specifically in rural telecommunications decision problems, functionality or purpose emerges from multiple interactions that thread their way through the system. Thus, there is a need for a holistic approach in which all the criteria and alternatives involved are connected in a network system that accepts various dependencies and interactions. Thus, such problems which are of complex nature, with some aspects compounded by the presence of intangible criteria cannot be structured hierarchically because they involve many interactions and dependencies requiring a MCDM method that can holistically deal with qualitative and quantitative data. Hence, the ANP model presented in the next section can overcome the shortcomings of the strict hierarchical structure inherited in AHP. The ANP can deal with problems having complex relationships among criteria (dependency and feedback) and so it is considered more pragmatic approach to decision making which gives better predictions. The existence of feedbacks in any structure prevents the problem from being modelled hierarchically due to the difficulty in deciding which cluster is higher/lower than the other. Moreover, because of inner dependence, the relationships between the criteria of the same level are not represented hierarchically. Accordingly, based on the above and due to its holistic approach; the ANP is chosen as a dominant methodology for this study.