Optimizing marine vehicles industry: a hybrid analytical hierarchy process and additive ratio assessment approach for evaluating and selecting IoT-based marine vehicles

Problem statement

Marine vehicles powered by IoT also have prospects for sustainable development. These vehicles provide optimized energy use, fewer emissions, and effective resource utilization through data-driven understandings. They support the adoption of environmentally beneficial practices such as route optimization, pollution monitoring, and preventative care, which is in line with the growing need for environmentally friendly and sustainable practices in the automotive sector. Successful assessment and selection techniques are essential for maximizing the capability of IoT-based maritime vehicles. The rapid development of the IoT has led to the introduction of IoT-enabled marine vehicles, which have significant effectiveness in the auto industry. However, reliable evaluation and decision-making processes are required to identify and rank the most feasible IoT-powered marine vehicles. This study addresses this difficulty by introducing a hybrid technique combining the AHP and ARAS to analyze and choose marine IoT-based boats according to their integrity and performance-centric criteria. The objective is to serve decision-makers in the vehicle sector with an accurate resource for the assessment and determination of maritime vehicles that are IoT-based while considering critical performance and authenticity related factors and their mutual dependence.

Contribution based on existing literature

This study attempts to close a substantial gap in the literature by integrating the ARAS and AHP to analyze secure authentication methods in IoT applications within the marine sector. Although ARAS and AHP are well-known methodologies, our contribution is the way we use and integrate them in a particular yet unexplored area: the maritime sector. As far as we have learned from the existing studies, there aren’t any appropriate selection procedures specifically designed for the challenge of choosing secure and authentic IoT solutions in the marine industry at this moment. An in-depth review of several topics, such as IoT strategy and authentication system preference, reveals that there aren’t many evaluation mechanisms made especially for this industry. However, the research does more than simply encourage the utilization of tried-and-true methods. Consequently, the proposed approach is thoroughly analyzed and contrasted in this paper with different appraisal mechanisms used in various industries, such as entropy, analytic network process (ANP), Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS), and Decision-Making Trial and Evaluation Laboratory (DEMATEL). Prioritization should be given to describing the relative advantages that our approach provides and identifying topics for enhancement that are unique to the marine industry. As a result, the proposed study improves upon the already adopted assessment methodologies and offers a feature-focused analysis relevant to IoT-based marine systems. Thus, our research’s comprehensive analysis seeks to close existing gaps and provide a solid theoretical and numerical framework for both scholars and practitioners. The published studies cited from numerous libraries have been depicted in Fig. 1.

Features engineering

The formulation of comparative parameters for selecting dependable IoT techniques to address security and autonomous concerns in marine industrial contexts forms the basis of the recommended design. characteristics focusing on identity and access control concerns are defined to guarantee that only permitted gadgets may access the IoT network. The fact that protection measures encompass every aspect of IoT, and autonomous technologies is the primary rationale in favor of adopting them. The attributes of IoT technologies are constructed from a variety of sources. A thorough and precise investigation of the literature is performed to uncover and gather the necessary attributes for the comparison of IoT devices integrated into marine vehicles. These attributes are quite ubiquitous since numerous different authentication systems employ them. After the initial identification of sixty-seven attributes from diverse sources, there are now just seven attributes left, allowing for examination and selection among many other options. In Fig. 2, all the attributes from the literature study are described.

Identifying and selecting features

The suggested evaluation approach’s initial level lays the greatest emphasis on selecting qualities and alternatives. The information from an IoT expertise panel is contrasted with seven viable solutions, including V1-V7. The efficacy of maritime vehicle technologies will be assessed in an IoT environment. Only the option that earned the highest overall rating out of all feasible options considering the established criteria would be regarded as the best choice in terms of their chosen criteria. This approach will work or be regarded as authentic depending on the attributes of each of the seven features-based criteria. These seven criteria consist of safety features (F1), navigation accuracy (F2), environmental impact (F3), transmission throughput (F4), authenticity (F5), energy efficiency (F6), and operational cost (F7), as illustrated in Fig. 3. The effectiveness of the criterion depends on how all of them interact with one another.

Evaluation framework based on multicriteria approaches

The AHP and ARAS methodologies operate as assessment instruments and selection options to choose the best target solution for maritime IoT-based vehicles. The fusion of ARAS and AHP within marine optimization presents notable advantages attributable to their synergistic interrelation. ARAS is characterized by its simplicity and adaptability, facilitating prompt evaluations in decision-making processes, whereas AHP excels in organizing intricate decision variables and prioritizing alternatives based on their significance. This methodology proves particularly valuable in the evaluation of multifaceted elements such as operational efficiency, safety, environmental consequences, and cost-effectiveness. The fusion of ARAS and AHP amplifies risk management by streamlining comprehensive risk assessments and prioritizing strategies for risk mitigation. AHP’s validation procedures ensure coherence and diminish biases in decision-making processes, which are essential for the evaluation and mitigation of risks inherent in maritime operations. Furthermore, this integration fosters engagement with stakeholders and sustains transparency in decision-making protocols. Through the incorporation of ARAS’s simplicity and AHP’s structural framework, decision-makers can arrive at well-informed, data-driven decisions that account for diverse variables and perspectives of stakeholders. This integration enhances the quality of decisions across various facets of marine optimization, thereby aligning with objectives such as efficiency, safety, sustainability, and stakeholder contentment. In addition, the envisioned appraisal framework’s three primary stages are as follows: The initial step is integrating the attributes into the testing scheme. In the second computing stage, numerical computations are used to determine the alternatives’ intrinsic attributes. When all empirical evidence comes in, the most efficient choice is picked as the IoT solution using the ARAS method based on the scores given for the criterion. Below is a complete mathematical calculation and processes involved in the AHP and ARAS techniques.

AHP approach

This MCDM technique works well for organizing and analyzing complicated choices. AHP offers a practical method for handling several issues pertaining to various areas while making precise and practical selections. Saaty initially introduced this method in 1980, and by utilizing the presented Saaty scale (Fang et al., 2022), we can choose wisely from the numerous options that are accessible based on a number of distinct factors. The current study applies the AHP approach to the chosen parameters to determine how much every attribute is weighted. Using these weights, ARAS ranks several computer-based health monitoring applications to improve the performance of the healthcare sector. Figure 4 illustrates all the phases required to complete the procedure to determine the weights of the chosen attributes using AHP.

Step 1. AHP hierarchical model

This presents the issue as an ordered framework with three levels. The goal is displayed at level 1, and the criteria are displayed at level 2. The final level is where the choices are addressed, as seen in Fig. 5.

Step 2. Pair-wise comparison matrix representation of features

Diagrammatic notation cannot be utilized for computational analysis but is best suited for visual inspection. Like this, when a framework gets bigger it is more difficult to comprehend the accompanying graphic visually. Therefore, it is essential to provide a description that is easy to comprehend so that computers can successfully store, retrieve, and analyze data. Continuing with the Saaty scale, a matrix is created by assigning particular scores to each factor per the expert’s needs. Finally, a pair-wise assessment matrix of size 7 × 7 is created and assembled for the present situation. The comparative weights of each criterion obtained from the IoT experts using a scale are as displayed below in the matrix. C r i t e r i a F 1 F 2 F 3 F 4 F 5 F 6 F 7 F 1 1 . 00 3 . 00 2 . 00 2 . 00 0 . 33 3 . 00 3 . 00 F 2 0 . 33 1 . 00 2 . 00 0 . 33 0 . 50 3 . 00 2 . 00 F 3 0 . 50 0 . 50 1 . 00 0 . 50 0 . 33 0 . 50 2 . 00 F 4 0 . 50 3 . 00 2 . 00 1 . 00 2 . 00 3 . 00 3 . 00 F 5 3 . 00 2 . 00 3 . 00 0 . 50 1 . 00 3 . 00 2 . 00 F 6 0 . 33 0 . 33 2 . 00 0 . 33 0 . 33 1 . 00 0 . 50 F 7 0 . 33 0 . 50 0 . 50 0 . 33 0 . 50 2 . 00 1 . 00 .

Step 3. Normalizing comparison matrix and criteria weights

The matrix of pairwise comparison is normalized for each criterion using Eq. (1). (1)N i j = X i j S u m o f e v e r y c o l u m n .

The comparative matrix is normalized to eliminate subjectivity. Then, the relative importance of every criterion is figured out by averaging the normalized comparative matrix for every row in accordance with Eq. (2). (2)C r i e r i a W e i g h t s = ∑ N i j n .

The determined normalized matrix and criterion weights are provided in the below matrix.

C r i t e r i a F 1 F 2 F 3 F 4 F 5 F 6 F 7 F 1 0 . 17 0 . 29 0 . 16 0 . 40 0 . 07 0 . 19 0 . 22 F 2 0 . 06 0 . 10 0 . 16 0 . 07 0 . 10 0 . 19 0 . 15 F 3 0 . 08 0 . 05 0 . 08 0 . 10 0 . 07 0 . 03 0 . 15 F 4 0 . 08 0 . 29 0 . 16 0 . 20 0 . 40 0 . 19 0 . 22 F 5 0 . 50 0 . 19 0 . 24 0 . 10 0 . 20 0 . 19 0 . 15 F 6 0 . 06 0 . 03 0 . 16 0 . 07 0 . 07 0 . 06 0 . 04 F 7 0 . 06 0 . 05 0 . 04 0 . 07 0 . 10 0 . 13 0 . 07 S u m 1 1 1 1 1 1 1 C r i t e r i a W e i g h t s 0 . 21 0 . 12 0 . 08 0 . 22 0 . 23 0 . 07 0 . 07.

Step 4. Multiplication of pairwise comparison matrix and criterion weights

The pair-wise assessment matrix and the derived criterion weights are multiplied to assess the precision of the estimated value. The required process is as illustrated below in matrix form.

C r i t e r i a F 1 F 2 F 3 F 4 F 5 F 6 F 7 F 1 1 . 00 3 . 00 2 . 00 2 . 00 0 . 33 3 . 00 3 . 00 F 2 0 . 33 1 . 00 2 . 00 0 . 33 0 . 50 3 . 00 2 . 00 F 3 0 . 50 0 . 50 1 . 00 0 . 50 0 . 33 0 . 50 2 . 00 F 4 0 . 50 3 . 00 2 . 00 1 . 00 2 . 00 3 . 00 3 . 00 F 5 3 . 00 2 . 00 3 . 00 0 . 50 1 . 00 3 . 00 2 . 00 F 6 0 . 33 0 . 33 2 . 00 0 . 33 0 . 33 1 . 00 0 . 50 F 7 0 . 33 0 . 50 0 . 50 0 . 33 0 . 50 2 . 00 1 . 00 × C r i t e r i a W e i g h t s 0 . 21 0 . 12 0 . 08 0 . 22 0 . 23 0 . 07 0 . 07.

Step 5. Results of the above product matrix and calculation of the weighted sum value

The required outcomes figured out by the above multiplication matrix are as provided. Furthermore, the total of every row in the generated matrix is computed to get the weighted sum score. The matrix having the multiplied values and the calculated weighted sum scores from it are as illustrated below. C r i t e r i a F 1 F 2 F 3 F 4 F 5 F 6 F 7 F 1 0 . 21 0 . 35 0 . 16 0 . 44 0 . 08 0 . 21 0 . 22 F 2 0 . 07 0 . 12 0 . 16 0 . 07 0 . 11 0 . 21 0 . 15 F 3 0 . 11 0 . 06 0 . 08 0 . 11 0 . 08 0 . 03 0 . 15 F 4 0 . 11 0 . 35 0 . 16 0 . 22 0 . 45 0 . 21 0 . 22 F 5 0 . 64 0 . 23 0 . 24 0 . 11 0 . 23 0 . 21 0 . 15 F 6 0 . 07 0 . 04 0 . 16 0 . 07 0 . 08 0 . 07 0 . 04 F 7 0 . 07 0 . 06 0 . 04 0 . 07 0 . 11 0 . 14 0 . 07 W e i g h t e d S u m V a l u e 1 . 67 0 . 89 0 . 61 1 . 72 1 . 81 0 . 52 0 . 57 .

Step 6. Identification of the ratio of weighted sum value and attributes weights

To calculate the ratio of the weighted sum score and criteria weights, the weighted sum scores were divided by the criterion weights. In Table 1, the estimated ratio between the weighted aggregate score and the criterion weights is as displayed.

Step 7. Calculation of λ_max, CI, and CR

(3)λ m a x = 7 . 80 + 7 . 58 + 7 . 67 + 7 . 76 + 8 . 03 + 7 . 61 + 7 . 73 7 .

To calculate the value of the Consistency Index (C.I.), we apply the following Eq. (4) (4)C . I = λ m a x − n n − 1 ,

while n stands for the number of attributes/criteria.

C . I = 7 . 74 − 7 7 − 1 C . I = 0 . 12 .

To figure out the Consistency Ratio (C.R), we apply Eq. (5) (5)C . R = C . I R . I .

While RI stands for the Random Index, the value of RI is 1.32 at the seven positions. C R = 0 . 12 1 . 32 .

CR = 0.09, Hence the value of CR is acceptable because it is less than 0.1.

Figure 6 visually stands for the criterion weights that were identified by the AHP method.

ARAS approach

The Additive Ratio Assessment (ARAS) technique is used by many industrial sectors, to help with making decisions. The ARAS technique chooses “the finest” choice by streamlining complicated decision-making situations. The relative signal could draw attention to distinctions between the suggested course of action and the available choices. This is carried out by taking different measuring units’ effects out of the equation (Alqurashi et al., 2022). The incorporation of ARAS is directed towards addressing the constraints encountered in the utilization of AHP for decision-making purposes. Although AHP demonstrates proficiency in structuring hierarchies and conducting pairwise evaluations, it encounters challenges when handling extensive decision matrices and aiming for simplicity. ARAS, recognized for its flexibility, is deemed appropriate for prompt assessments and the management of diverse criteria without unnecessarily complicating the procedure. The suggested methodology strives to leverage the strengths of both approaches, simplifying the process and presenting a clear-cut evaluation framework. ARAS’s additive characteristic fosters transparency and the participation of stakeholders, thereby enriching decision-making quality, effectiveness, and alignment with the objectives of maritime operations. The ARAS data analysis technique includes weights and criterion additions during the computation phase to aid the IoT experts and industrialists in making an optimal decision practical. The ARAS approach is used as an analytical instrument and decision-making process to choose the best intended alternative for a secure and effective IoT-based marine transportation system. IoT experts were enlisted, and as stated before, AHP was used to classify performance and authentication-related features. The highest-ranking alternative is then selected as the target authentication solution based on the specified attribute values. An illustration of the steps involved in this method used to evaluate and rank the intended strategy for IoT-based marine transportation systems is as shown in Fig. 7. This section provides information on the steps and details for ARAS.

Step 1. Decision matrix representation of features and alternatives

The graphic depiction is most suited for visual screening; nevertheless, computational analysis cannot be done with it. The accompanying graph also grows complex for big systems, which therefore affects how easily it can be understood visually. Therefore, it is crucial to provide a model that is simple to understand so that computers can efficiently gather, access, and analyze data. The entire set of chosen attributes is beneficial in this process. The performance attributes and data gathered from IoT and marine vehicle specialists are provided in a decision matrix with a dimension of 7 by 7. Seven choices are chosen for selection based on the defined attributes in this matrix, which includes seven performance factors listed in columns. The optimum value (OV) in each column has also been found. The viewpoints gathered from the IoT specialists are separated into seven choices as shown in the decision matrix below based on the weights provided to each attribute. C r i t e r i a A l t e r n a t i v e s F 1 F 2 F 3 F 4 F 5 F 6 F 7 V 1 2 . 00 5 . 00 3 . 00 7 . 00 6 . 00 4 . 00 8 . 00 V 2 4 . 00 7 . 00 2 . 00 5 . 00 3 . 00 6 . 00 2 . 00 V 3 6 . 00 3 . 00 5 . 00 2 . 00 4 . 00 7 . 00 9 . 00 V 4 3 . 00 2 . 00 6 . 00 4 . 00 8 . 00 5 . 00 6 . 00 V 5 5 . 00 3 . 00 4 . 00 3 . 00 9 . 00 2 . 00 7 . 00 V 6 8 . 00 6 . 00 2 . 00 7 . 00 5 . 00 3 . 00 4 . 00 V 7 7 . 00 4 . 00 6 . 00 8 . 00 2 . 00 5 . 00 3 . 00 O V 8 7 6 8 9 7 9

Step 2. Normalizing a decision matrix

The selection matrix has been normalized for both beneficial and non-beneficial attributes. Greater scores are preferred for the selection matrix’s attributes since all of them are beneficial. The formulas listed below are used to figure out factors that are both beneficial and not.

For beneficial criteria, (6)X ¯ i j = X i j ∑ i = 0 m X i j

For non-beneficial criteria,

(7)X i j = 1 X i j ; X ¯ i j = X i j ∑ i = 0 m X i j .

The selection matrix has been normalized using Eq. (6) to eliminate variability because all the chosen attributes are beneficial in nature. Then, the criterion weights figured out by AHP will be multiplied with it to get the weighted normalized matrix in the next step. The normalized selection matrix’s specifics are mentioned below in the matrix form.

C r i t e r i a A l t e r n a t i v e s F 1 F 2 F 3 F 4 F 5 F 6 F 7 V 1 0 . 05 0 . 14 0 . 09 0 . 16 0 . 13 0 . 10 0 . 17 V 2 0 . 09 0 . 19 0 . 06 0 . 11 0 . 07 0 . 15 0 . 04 V 3 0 . 14 0 . 08 0 . 15 0 . 05 0 . 09 0 . 18 0 . 19 V 4 0 . 07 0 . 05 0 . 18 0 . 09 0 . 17 0 . 13 0 . 13 V 5 0 . 12 0 . 08 0 . 12 0 . 07 0 . 20 0 . 05 0 . 15 V 6 0 . 19 0 . 16 0 . 06 0 . 16 0 . 11 0 . 08 0 . 08 V 7 0 . 16 0 . 11 0 . 18 0 . 18 0 . 04 0 . 13 0 . 06 O V 0 . 19 0 . 19 0 . 18 0 . 18 0 . 20 0 . 18 0 . 19 * C r i t e r i a W e i g h t s 0 . 21 0 . 12 0 . 08 0 . 22 0 . 23 0 . 07 0 . 07

Step 3. Weighted normalized decision matrix

The factor values acquired through the application of AHP are multiplied with every factor of the normalized selection matrix using Eq. (8) to derive the weighted normalized selection matrix. The results obtained from this calculation are provided in matrix form as given below. (8)X ˆ i j = X ¯ i j ∗ W j C r i t e r i a A l t e r n a t i v e s F 1 F 2 F 3 F 4 F 5 F 6 F 7 V 1 0 . 010 0 . 016 0 . 007 0 . 035 0 . 029 0 . 007 0 . 012 V 2 0 . 020 0 . 022 0 . 005 0 . 025 0 . 015 0 . 011 0 . 003 V 3 0 . 030 0 . 010 0 . 012 0 . 010 0 . 020 0 . 012 0 . 014 V 4 0 . 015 0 . 006 0 . 014 0 . 020 0 . 039 0 . 009 0 . 009 V 5 0 . 025 0 . 010 0 . 009 0 . 015 0 . 044 0 . 004 0 . 011 V 6 0 . 040 0 . 019 0 . 005 0 . 035 0 . 024 0 . 005 0 . 006 V 7 0 . 035 0 . 013 0 . 014 0 . 040 0 . 010 0 . 009 0 . 005 O V 0 . 040 0 . 022 0 . 014 0 . 040 0 . 044 0 . 012 0 . 014

Step 4. P_i-Optimality function for ith alternative and calculation of utility degree

Equation (9) has been used to get the values of the optimality function (Pi) for every option as the aggregate of each row of the weighted normalized matrix was determined. (9)P i = ∑ J = 1 n X ˆ i j ; i = 0 ¯ , m .

For the determination of the utility degree (K_i)

The comparison of the alternative with Eq. (10), yields the K_i for different alternatives. (10)K i = P i P 0

while the value of P_o is 0.187 as well.

The results obtained by these equations are as shown in Table 2.

Step 5. Finalize the ranking of alternatives

The options are sorted in an ascending sequence of K_i. The most desirable option is the one with an elevated K_i value, whereas the most undesirable option is the one with the smallest K_i score. The calculated ranking of each alternative based on the ARAS method is as shown in Fig. 8.