Prioritizing the Potential Smartiﬁcation Measures by Using an Integrated Decision Support System with Sustainable Development Goals (a Case Study in Southern Italy)

: With the increasing population of cities, expanding roads as one of the essential urban infrastructures is a necessary task; therefore, adverse effects such as increased fuel consumption, pollution, noise, and road accidents are inevitable. One of the most efﬁcient ways to mitigate congestion-related adverse effects is to introduce effective intelligent transportation systems (ITS), using advanced technologies and mobile communication protocols to make roads smarter and reduce negative impacts such as improvement in fuel consumption and pollution, and reduction of road accidents, which leads to improving quality of life. Smart roads might play a growing role in the improved safety of road transportation networks. This study aims to evaluate and rank the potential smartiﬁcation measures for the road network in Calabria, in southern Italy, with sustainable development goals. For this purpose, some potential smartiﬁcation measures were selected. Experts in the ﬁeld were consulted using an advanced procedure: four criteria were considered for evaluating these smartiﬁcation measures. The Integrated fuzzy decision support system (FDSS), namely the fuzzy Delphi analytic hierarchy process (FDAHP) with the fuzzy technique for order performance by similarity to ideal solution (FTOPSIS) were used for evaluating and ranking the potential smartiﬁcation measures. The results demonstrated that the repetition of signals in the vehicle has the highest rank, and photovoltaic systems spread along the road axis has the lowest rank to use as smartiﬁcation measures in the roads of the case study.


Introduction
Roads always play a very vital role in human life as urban and rural arteries. Unfortunately, first of all, in terms of road safety. To persuade yourself that this is an important issue, just think that road traffic deaths, and traffic injuries represent the leading cause of death for people aged 5-29 years. Together with vehicles and driver behavior, infrastructure is one of the five pillars identified by the World Health Organization (WHO) to be adequately managed to ensure high levels of safety in road traffic. On the other hand, roads are essential elements for economic growth, increasing public welfare, and creating environmental issues such as air pollution. Roads are considered one of the main players in planning for sustainable development. Therefore, more attention to this essential infrastructure can significantly impact safety and sustainable development indicators. With the advancement of technology and various sciences such as artificial intelligence, wireless and mobile communications, and remote sensing, smart roads have received much attention. By smarting roads, not only will passenger safety and well-being increase, but smart roads will also enable us to deal with issues such as maintenance costs, congestion-related time-waste, based on triangular or trapezoidal membership functions. In this study, triangular fuzzy numbers (TFNs) ∼ a ij are used, and the membership function is introduced according to Equations (1) to (4) and Figure 1 [40,41]. ∼ a ij = (α ij , δ ij , γ ij ) (1) α ij = Min(β ijk ), k = 1, . . . , n

Survey of Experts and Calculation of Fuzzy Numbers
In the first step, relevant experts conducted a survey on a specific topic. Then the fuzzy numbers are generated directly from the experts' surveys. Fuzzy numbers can be calculated based on triangular or trapezoidal membership functions. In this study, triangular fuzzy numbers (TFNs) ij a  are used, and the membership function is introduced according to Equations (1) to (4) and Figure 1 [40,41].   β ijk presents the relative importance of the ith parameter over the jth parameter from kth expert's point of view. α ij and γ ij represent the upper and lower bound of experts' opinions and δ ij is the geometric median of experts' opinions.

Determining the Fuzzy Pairwise Comparison Matrix
In this step, according to the fuzzy numbers, the fuzzy pairwise comparison matrix is determined based on Equation (5) [40,41].

Calculating the Relative Fuzzy Weight of Parameters
After determining the fuzzy pairwise comparison matrix, the relative fuzzy weight of each parameter ( ∼ W i ) is calculated based on Equations (6) and (7) [42,43].
where the symbols ⊗ and ⊕ are the multiplication of fuzzy numbers and the addition of fuzzy numbers, respectively. ∼ W i represents a row vector consisting of the fuzzy weight of the ith factor ( ∼ W i = (w 1 , w 2 , . . . , w n ), (i = 1, 2, . . . , n)). In the final step, Equation (8) is used for defuzzing the weight of parameters and determining a nonfuzzy number for the weight of each parameter. The defuzzification is calculated based on the geometric mean technique [44].

Fuzzy Technique for Order Performance by Similarity to Ideal Solution (FTOPSIS)
The fuzzy multi-criteria decision-making methods are reliable systems for dealing with uncertainty when making decisions in engineering applications [45][46][47][48]. The fuzzy technique for order performance by similarity to ideal solution (FTOPSIS) has been used as one of the most effective fuzzy multi-criteria decision-making methods in a wide range of studies such as risk assessment of construction projects, logistics services, mining problems, and the quality of airline services [49][50][51][52]. In the FTOPSIS approach, a set of linguistic variables and their corresponding fuzzy numbers are introduced, and the process of evaluation is done by assigning them to the decision-making matrix and solving mathematical equations of FTOPSIS. In general form, eight steps were introduced for TOPSIS by Chen and Hwang (1992) as follows [53]:

Formation of Decision Matrix
The decision matrix is formed as follows: It is worth mentioning that ∼ x ij = (s, l, r) as the fuzzy triangular numbers is used, that ∼ x ij is the performance of ith (i = 1, 2, 3, . . . , N) alternative about jth (j = 1, 2, 3, . . . , n) criterion. This study indicates linguistic variables and their corresponding triangular fuzzy numbers for ranking alternatives and assessing criteria in Table 1 [40]. Table 1. Linguistic variables and corresponding fuzzy triangular numbers.

Determining the Weight Matrix of Criteria
In this step, each criterion's weight of importance coefficient is calculated, and the weight matrix criteria are determined based on Equation (9) [53]. where ∼ w j is the weight of importance coefficient of each criterion and each component of ∼ w j is used as ∼ w j = (w 1 , w 2 , w 3 ).

Normalization of the Fuzzy Decision Matrix
The normalized values are also fuzzy when fuzzy numbers are used in the decision matrix. According to Equations (10) and (11), normalization of the fuzzy triangular number is calculated for positive and negative criteria, respectively [40].
Then, the normalized fuzzy decision matrix R is molded based on Equation (12) with m criterion and n alternative [40].

Determining of the Weighted Normalized Fuzzy Decision Matrix
In this step, at first, the weight of each criterion is multiplied by the normalized fuzzy decision matrix based on Equation (13) [40].
where w j represents the weight of each criterion. Then, the weighted normalized fuzzy decision matrix is formed according to Equation (14) [40].  (15) and (16), fuzzy positive ideal solution and fuzzy negative ideal solution are determined [40].
i represent the best and worst values of ith criterion from among all alternatives, respectively. The alternatives that are placed in A * and A , represent ultimately better and ultimately worse alternatives, respectively. In this step, the distance of each alternative is gained from the fuzzy positive ideal solution and fuzzy negative ideal solution based on Equations (17) and (18), respectively [40].
where d represents the distance between two fuzzy triangular numbers, which is calculated based on Equation (19) [40].

Determining of the Closeness Coefficient (CC)
According to values of distance from a fuzzy positive ideal solution and a fuzzy negative ideal solution, the closeness coefficient of each alternative is calculated based on Equation (20).

Ranking of Alternatives
In the last step, the rank of each alternative is determined based on the calculated closeness coefficient for each alternative.

Case Study
The main road infrastructures of Calabria can be classified into two separate groups: longitudinal and transversal roads. In the first one, there are the road infrastructures that cross the whole region from north to south, in particular: • A2, Mediterranean Highway, which is the only highway realized in Calabria; • SS 106, which is the main road along the Ionic coast; • SS 18 represents the most significant link between the Tyrrhenian coast's internal areas and coastal settlements.
The second group is instead composed of all those roads that cross Calabria from west to east: As reported in Figure 2, these are the region's main roads for a total length of about 1.400 km.
Without considering A2, SS 280, SS 534, and roads with two separate carriageways, all the other roads mentioned above are characterized by several safety deficiencies. These are strictly related to the frequent transition from rural to urban sections and the relative changing of operative speeds [54]. Therefore, the road test network was chosen to analyze road infrastructure features concerning safety issues accurately. As reported in Figure 2, these are the region's main roads for a total length of about 1.400 km.

Determining Criteria's Weights Using FDAHP
After determining smartification measures, four decision-making criteria were selected, including environmental sustainability (C1: This criterion refers to the obligation to save natural resources and maintain global ecosystems in order to promote current and future health and welfare), economic sustainability (C2: It refers to policies that promote the nation's long-term economic growth), safety (C3: It refers to the management of known hazards in order to reach a safe level of risk), and benefit-cost ratio (C4: It refers to the monetary or qualitative link between a project's relative costs and benefits). These criteria were selected after consultation with experts and by the contribution of experienced technicians to evaluate smartification measures. Moreover, these criteria were selected to achieve the goals of sustainable development in road transport for Calabria in consultation with experts. The decision-making team consisted of university professors and experienced technicians. They were well familiar with the Calabria roads and Intelligent Transportation Systems (ITS). These criteria also play an influential role along with sustainable development goals. It should be noted that all these criteria have positive aspects. These survey forms were prepared (such as Table 2) and completed in several meetings with experts and after explaining the scoring system. Using Saaty's 1-9 scales, the pairwise comparison is made for each decision-maker (Di). After filling in the questionnaires by the experts' opinion, Equations (1)-(5) are used, and TFNs are formed with a triangular function according to Figure 1. The values of decision-makers' pairwise comparison matrix based on TFNs are indicated in Table 2. Then, using the pairwise comparison matrix obtained in the previous step, the relative fuzzy weights of the criteria are calculated based on Equations (6) and (7).
In the last step, Equation (8) is used for de-fuzzing the weight of the criteria, and the final weights of each criterion are determined. The priorities of weights are indicated for each criterion in Table 3. The results obtained show that safety (C3), from the point of view of experts, plays a vital role in evaluating and ranking the smartification measures of the studied roads. Moreover, benefit-cost ratio (C4), environmental sustainability (C1), and economic sustainability (C2) are next in importance, respectively.

Ranking of Smartification Measures Using FTOPSIS
To evaluate and prioritize the smartification measures, according to the several conditions in the Calabria area, such as roads and technical conditions, and considering the initial investigation of the existing smartification measures, a list of potential smartification measures was assigned. Then, the 27 smartification measures were determined following a series of discussions and consultations, including expert brainstorming. For this purpose, several meetings were held with the participation of all experts. Then, a questionnaire (such as Table 4) was prepared based on the finalized measures. Finally, another meeting with the experts was arranged to fill out the questionnaire (the fuzzy decision matrix). The experts used the linguistic variables for prioritizing the twenty-seven smartification measures based on Table 1 in this step. After the final summation, all the fuzzy decision matrices were integrated, and the combinatorial fuzzy decision matrix was formed according to Table 4.

Safety (C3) Benefit-Cost Ratio (C4)
Traffic Speed limit notification (A17) (3,5.5,9) (3,7,10) (0,6,10) (7,9.5,10) Traffic information and recommended itineraries (A18) ( Green islands for charging electric vehicles (A27) (9,10,10) (7,9.75,10) (0,0.75,3) (5,9,10) After the fuzzy decision matrix was formed, since the decision criteria have a positive aspect, the normalized fuzzy decision matrix is formed based on Equations (10) to (12). Then, by multiplying the weights of each criterion that were calculated from the FDAHP method (  (13) and (14). Table 5 indicates the normalized weighting fuzzy decision matrix (NWFDM) for 27 smartification actions.  In the next step, the fuzzy positive ideal solution and fuzzy negative ideal solution are calculated based on Equations (15) and (16) Table 6 and Figure 3. For instance, the distance from the fuzzy positive ideal solution, fuzzy negative ideal, and the CC for the first smartification action (A1) are calculated as follows:  As shown in Table 6 and Figure 3, there have been dramatic cuts in the importance of two categories, including Internet services and solar and ecological roads with other categories. Hence, photovoltaic systems spread along the road axis (A26), Green islands for charging electric vehicles (A27), and insurance and financial services (A24) have the lowest ranks, correspondingly. The first smartification action was the repetition of signals in the vehicle (A21) with a closeness coefficient equal to 0.428, which belongs to the traffic category. Then, the five smartification actions from three categories including risk of accident alert (A11), crowdsourced data: dangerous site (A12), traffic information and recommended itineraries (A18), assisted navigation (A20), and automatic management of parking and accesses (A23) achieved the second rank with the closeness coefficient equal to 0.426 among the twenty-seven measures. Finally, these calculations, the ranking of smartification measures of the Calabria's roads with FDAHP-FTOPSIS is (A21 > A23 = A20 = A18 = A12 = A11 > A17 > A7 = A10 = A22 > A82 = A8 = A13 = A14 = A15> A16 > A5> A9 > A6> A1 > A19 > A3 > A25> A4 > A24> A27 > A26). According to this ranking based on experienced experts' opinion, the repetition of signals in the vehicle (A21) was considered as the most significant action for smarting roads of the case study. Hence, wireless communications, such as vehicle-to-everything (V2X) communications, can be useful for smarting roads, and it is suggested that prioritizing investment and research in future smart plans be considered for V2I (vehicle-to-infrastructure), V2N (vehicle-tonetwork), and V2D (vehicle-to-device). Moreover, these three vehicular communication systems can be applied as effective tools for the five smartification actions, which had the second rank in terms of importance and priority. For example, in safety messages, the message of a crash report along the route can increase drivers' awareness of their route by using A21 and A11. Using A18 and 20 allows users to arrange an itinerary, check for amenities along the road, bookmark important sights, and so on. Furthermore, in-car warnings can use data collection and identification of dangerous road sections to slow vehicles in risky zones by using A12. A23 which can help to alleviate traffic congestion and properly manage parking demand in the Calabria road network, also received a positive evaluation.  As shown in Table 6 and Figure 3, there have been dramatic cuts in the importance of two categories, including Internet services and solar and ecological roads with other categories. Hence, photovoltaic systems spread along the road axis (A26), Green islands for charging electric vehicles (A27), and insurance and financial services (A24) have the lowest ranks, correspondingly. The first smartification action was the repetition of signals in the vehicle (A21) with a closeness coefficient equal to 0.428, which belongs to the traffic category. Then, the five smartification actions from three categories including risk of accident alert (A11), crowdsourced data: dangerous site (A12), traffic information and recommended itineraries (A18), assisted navigation (A20), and automatic management of parking and accesses (A23) achieved the second rank with the closeness coefficient equal to 0.426 among the twenty-seven measures. Finally, these calculations, the ranking of smartification measures of the Calabria's roads with FDAHP-FTOPSIS is (A21 > A23 = A20 = A18 = A12 = A11 > A17 > A7 = A10 = A22 > A82 = A8 = A13 = A14 = A15> A16 > A5> A9 > A6> A1 > A19 > A3 > A25> A4 > A24> A27 > A26). According to this ranking based on The car population in Calabria is much older than in other parts of Europe. This is an explanation for the low ranking of A27. The same study in a different geographical area with more new vehicles would have brought different results regarding this specific measure dedicated to electric vehicles.
Many intersections in Calabria are dangerous due to poor design, signalization system, and other local factors, and this might have affected the poor evaluation of an innovative measure, such as A3. In other words, implementing new smart systems might not get a good evaluation when traditional road design and improvement measures are not correctly applied.
Conversely, other specific measures that have been considered essential for Calabria might not be helpful in other geographical areas.
It should be emphasized that the derived values for each smartification measure and their ranking are specific to the roads in Calabria and cannot be applied to other projects. Furthermore, the most significant constraint of this study was assembling the right team of local experts with the required expertise and capacity to diagnose and comprehend local issues and smart road measure effectiveness.

Conclusions
In this study, using expert opinions and combining smart road concepts to prioritize smartification actions, twenty-seven smartification measures in five categories are considered. Since smartening is an approach for achieving sustainable development, the four decision-making criteria, including environmental sustainability (C1), economic sustainability (C2), safety (C3), and benefit-cost ratio (C4), are also selected to achieve the goals of safe and sustainable development. Evaluating and priority of smartification measures are determined by the integrated FDSS, namely the FDAHP-FTOPSIS. The present study of the Calabria road network indicates that safety (C3) was identified as the most important contributing decision-making criteria for achieving the goals of safe and sustainable development, with a global weight equal to 0.486. The obtained results from the FDAHP-FTOPSIS technique indicated that the repetition of signals in the vehicle (A21) gained the highest rank with a closeness coefficient equal to 0.428 in the category of traffic. Moreover, the smartification measures of the last category (solar and ecological roads) achieved the lowest ranks among the twenty-seven smartification actions by closeness coefficients equal to 0.342 (A26) and 0.343 (A27). Whereas in this study, there were many options and uncertainty in the issue of decision, the FDAHP-FTOPSIS, which uses the language values and knowledge of experts, can be considered an alternative tool for initial assessments. Therefore, it is suggested to use other types of FDSS to achieve the effectiveness of the FDSS in investigating and prioritizing other case studies in future works.