Evaluating Intelligent CPTED Systems to Support Crime Prevention Decision-Making in Municipal Control Centers

Abstract

To maximize its synergetic effect across the cycle from prevention to response to post-crime management, crime prevention requires a balanced combination of spatial urban design and advanced crime prevention technologies for crime prediction and real-time response. This study derived intelligent Crime Prevention Through Environmental Design (CPTED) services and suggested a decision model based on the fuzzy Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) to implement these services in municipal control centers. The analysis results are summarized as follows. First, this study established a fuzzy TOPSIS-based decision-making support model enabling local government control centers to effectively select intelligent CPTED service elements. Second, overall, operator-led Closed-Circuit Television (CCTV) and platform control technologies were identified as significant components of intelligent CPTED service elements. Third, a comparison by city size revealed that large cities in the Seoul metropolitan area rated system services for control based on advanced crime prevention infrastructure (e.g., the crime monitoring systems and real-time control drones/robots) relatively higher. In contrast, small and medium-sized cities in other provinces rated services that were perceptible to residents and improved crime-prone environments (e.g., artificial intelligence (AI) video analysis for living safety) relatively higher.

1. Introduction

Crime prevention services play a crucial role in ensuring public safety; however, their effectiveness can be significantly hindered if they are deployed inflexibly without considering the unique characteristics of crime in different urban spaces and across various types of cities. One widely recognized approach that addresses spatial characteristics in crime prevention is Crime Prevention Through Environmental Design (CPTED), initially proposed by Jacobs and Newman [1,2] in response to emerging crime problems in urban redevelopment in the United States during the 1960s. Since its inception, CPTED techniques have evolved to adapt to diverse spatial and crime characteristics and have been continuously developed and implemented worldwide.

In Korea, the government has led the legislation and promotion of CPTED since the 2000s. Although many academic studies have been conducted and CPTED has been introduced in municipalities across the country, its application has primarily been limited to newly developed areas or new buildings. Existing low-rise residential areas in Korea pose challenges for implementing traditional CPTED strategies due to narrow roads, limited open spaces, and the monolithic layout of buildings. While technologies such as Closed-Circuit Television (CCTV) and improved lighting have been deployed in these areas, their effectiveness is compromised by a lack of consideration for spatial characteristics and a focus on short-term outcomes [3]. To enhance crime prevention in low-rise residential areas, there is a need for innovative services that integrate CPTED principles with spatial and technological elements.

Several researchers have explored the integration of CPTED with advanced technologies such as artificial intelligence (AI), the Internet of Things (IoT), and big data [4,5,6]. However, existing studies often overlook the comprehensive integration of CPTED with various crime prevention technologies. There is a pressing need to develop a more systematic approach to combine CPTED principles with crime prevention technologies.

In Korea, 226 municipalities operate CCTV control centers and control problematic situations in cities, such as crime prevention, traffic, and disasters, based on CCTV systems. As of 2021, 1,458,465 CCTV systems have been installed by the government and competent organizations, including municipalities, and 748,738 of them, which are more than 50% of the total, are intended for crime prevention [7]. To effectively operate a large number of CCTV systems and maximize their effect on crime prevention, it is important not only to upgrade CCTV video technologies, such as AI video analysis and selective control, but also to improve services across municipal integrated control centers by considering spatial characteristics, on-site application of applicable intelligent technologies, and differences in safety-related infrastructure. Furthermore, to introduce services in practice, it is crucial to present a service model and methodological technique that supports decision-making in municipal CCTV control centers, which are the operators and providers of crime prevention services.

This study aims to identify intelligent CPTED—a CPTED service combined with intelligent crime prevention technologies—by comprehensively considering a spatial layout and technology applications and suggest a service evaluation model for support decision-making in municipal control centers, which serve as both the operators and providers of crime prevention services. To achieve this, the following research question is posed: how can intelligent CPTED services, integrating spatial and technological elements, be effectively evaluated and implemented to enhance crime prevention in various urban environments in Korea?

To address question, this study identifies intelligent CPTED service elements through an interdisciplinary combination of architectural urban planning elements and intelligent crime prevention technologies by considering the components of CPTED. This study evaluates intelligent CPTED systems based on the fuzzy Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) model, a technique easy to use in practice and advanced as an analytical technique for supporting decision-making in crime prevention. As the actual operators in the municipal control centers were targeted as the respondents of this study’s questionnaire, this study is expected to identify practical implications from the provider side. This paper consists of the following sections: Section 1 introduces the scope and methodology of this study, and Section 2 conducts a literature review. Section 3 identifies the concept definition and service elements of intelligent CPTED, and Section 4 conducts empirical analysis, including a review of intelligent CPTED evaluation methodology, evaluation criteria, and alternatives. Based on the above, Section 5 and Section 6 discuss and provide implications about the results.

2. Literature Review

2.1. Concepts and Components of CPTED

CPTED was proposed by Jacobs and Newman [1,2] in the 1960s as crime problems arising from urban redevelopment emerged across the United States. Jacobs [1] raised comprehensive questions about the urban environment, including the relationship between the diversity of urban space and the occurrence of crime. To solve this problem, he suggested improvements, including the complex use of space, the use of streets by hour, and density. Newman [2] proposed the defensible space theory, which states that a crime can be prevented with changes in the physical environment of the city, such as natural surveillance and individual management of owned spaces. These two studies are considered to be the early studies of CPTED, which recognized the relationship between urban environment, crime, and safety.

Previous studies proposed six components of CPTED based on the defensible space theory: natural surveillance, territorial reinforcement, natural access control, target hardening, activity support, and maintenance and management. The principles are not used independent of each other but constitute an overlapped and complementary relationship [8]. Natural surveillance is the principle of discouraging criminal activities by ensuring visual access to public spaces, providing lighting to maximize exposure, arranging buildings, and introducing landscaping. Territorial reinforcement marks ownership of an area through landscaping, fences, signs, and lighting as a hypothetical territory where residents can assert their rights. It discourages potential criminals from committing a crime as they realize they can be monitored or deterred. Natural access control is the principle of preventing a crime by blocking the entry and exit of potential criminals and controlling the entry and exit of others to a certain protected space with appropriately installed facilities, including entrances, exits, lights, and fences. Activity support is the principle of reducing crimes and promoting a sense of safety by encouraging and stimulating the active use of a certain area by residents and subsequently reinforcing natural surveillance. Maintenance and management is the principle of planning and designing any facility or public space in a way that is conducive to the management for its continuous operation to discourage users’ deviant behaviors. Target hardening is intended to provide clear indications and information so that a space or facility can be used as intended, such as signage [8]. These six components of CPTED, if applied harmoniously, would be very effective and practical. In this context, this study aims to develop an implementation plan for each of the architectural urban planning elements based on the CPTED components and ensure a systematic process of combining with intelligent crime prevention technologies.

2.2. Korea’s Application and Limitation of CPTED

To examine and identify the limitations of CPTED applications in Korea, this study reviewed the previous studies that raised problems with existing CPTED projects. Shin and Kim [9] note the limitation that while CPTED prevents crime, municipalities perform architectural administration, which makes it difficult to apply CPTED to the real world due to the difference between practice and theory. The authors also suggested reinforcing the Crime Prevention Officer (CPO), a consultant specializing in public safety, as an option to tackle problems, such as the small-scale CPTED projects of municipalities, a lack of expertise, and difficulty in responding to complex public safety needs. Shim [3] suggests a strategy to develop a collaborative relationship between municipalities and the police, which plays a pivotal role in CPTED at the local level. Problems with the current CPTED policies include one-off projects to produce short-term outcomes (e.g., quantitative increase in the number of crime prevention facilities, such as CCTV and emergency bells, and one-time projects, such as painting murals) and poorly localized CPTED due to a lack of crime information. Furthermore, other limitations were also highlighted with the current CPTED projects: the repetition of simple strategies, including night lighting, CCTV installation, and murals, confusion between CPTED and an improved living environment, a lack of consideration for spatial planning while focusing on simple architectural and partial improvements, and a lack of consideration for maintenance. The abovementioned previous studies commonly point out a lack of expertise as a limitation of CPTED introduced under the leadership of municipalities, as crime prevention itself is primarily the responsibility of the police while CPTED itself can be regarded as part of the urban planning and architecture field. As CPTED currently implemented by the government and municipalities is part of urban design, crime prevention technologies should be combined with CPTED to strengthen expertise in crime prevention.

2.3. Combination of CPTED and Crime Prevention Technologies

Early CPTED studies were primarily focused on changes in the physical environment, such as presenting comprehensive problems in urban space and designing urban planning and structures based on the CPTED principles [1,2,10]. Subsequently, some studies proposed the second-generation CPTED methodology, a software approach that includes social factors through the participation of residents [11,12]. More recently, the third-generation CPTED methodology has been proposed with the perspective of social reinforcement to complement physical design (first generation) and social discussion (second generation) [13,14]. As a means of social reinforcement, it is necessary to reinforce the role of the public sector, including building facility infrastructure highlighting local characteristics, hosting social and cultural programs, establishing social places, and strengthening control. In 21st Century Security and CPTED, Atlas [15] argued that urban designers would need a balanced CPTED strategy (a combination of integrated crime prevention solutions) between architectural elements and crime prevention technologies combined with crime prevention concepts. The author stresses that urban designers should focus on crime prevention, safety vulnerabilities, and risks in cities and buildings by considering scientific crime prevention based on rapidly evolving crime prevention technologies. Grabosky [16] also argues that science has started to provide information regarding crime control in relation to CPTED [12]. Considering the abovementioned studies, it would be necessary to have both a centralized control tower response system to increase the effectiveness of social stability and a CPTED strategy combined with advanced crime prevention technologies and big data on crime information.

Against this background, a growing number of researchers are combining crime prevention technologies with CPTED. Some previous studies have been conducted to link traditional CPTED with physical security facilities, such as surveillance, lighting, alarm, and control equipment [4,17]. IoT and information communications technology (ICT) were applied to CPTED to overcome the difficulty of responding immediately to crime scenes, and lighting systems were incorporated into CPTED to prevent crimes at night [6,18,19]. Moreover, some studies used crime-related spatial and statistical data to identify areas vulnerable to crimes or suggest an optimized CCTV layout [5,20,21]. A growing number of cases are now evaluating CPTED-related risk factors in the virtual environment and incorporating advanced technology elements, such as the behavioral analysis of users in CPTED places [22,23,24]. While more recent studies incorporated more advanced technologies into CPTED research, and, therefore, can be considered progressive, their limitation is that they only deal with individual technologies. No study considered the overall CPTED elements. Living up to these changes in today’s world and overcoming the limitations of previous studies require exploring how to take CPTED to the next level, that is, introducing intelligent CPTED by combining intelligent crime prevention technologies and CPTED.

3. Materials and Methods

3.1. Selection of the Intelligent CPTED Systems for Evaluation

3.1.1. Selection of CPTED Architectural Urban Design Elements

To select an intelligent CPTED service, this study set the architectural urban planning elements selected under the CPTED principles as the classification criteria and then developed the implementation plan for each element. The CPTED Manual published by the Architecture and Urban Research Institute, a Korean government policy research institute [25], set the following eight architectural urban planning elements by considering the connection between CPTED action strategies and district-level planning elements: buildings, roads, landscaping, guide, lighting, community facilities, use/place, and parking. These elements were set based on the guidelines and instructions applied to the real world by the central government (Ministry of Land, Infrastructure, and Transport and National Police Agency), municipality (Seoul Metropolitan Government), and Korea Land and Housing Corporation (LH). For streets in areas densely populated by multi-family and multi-household houses, Choi [26] classified architectural urban planning elements into visual, architectural, and social. The abovementioned CPTED Manual added the element of landscaping (parks), which held significance as it suggested improvements tailored to the purpose of public CPTED projects.

The current study emphasizes the necessity of configuring the urban planning and architectural elements of intelligent CPTED combined with intelligent crime prevention technologies. In this regard, in addition to the urban planning and architectural elements of previous studies, this study added the element of urban control to smoothly operate and manage intelligent crime prevention technologies across the entire area under management based on the municipal control center. Furthermore, this study updated and reorganized the implementation plan of the existing urban planning and architecture elements to ensure more active crime response and scientific prevention, rather than simple improvements in the physical environment. For the six components of CPTED, roads and lighting (e.g., roads and buildings) were identified as natural surveillance, and architectural elements to prevent intrusion by outsiders were found for natural access control. Activity support and maintenance and management included community facilities, parks, use/place, parking, and urban control elements, and territorial reinforcement and target hardening included guide elements. For the Renewable Identification Number (RIN), a CPTED code was generated based on the initials of the architectural urban planning elements.

Table 1. Architectural urban planning element.

Principle of CPTED	Architectural Urban Planning Element	CPTED Code	Implementation Plan
Natural surveillance	Roads	Roads (R)	Natural surveillance on roads and CCTV systems installed to prevent crimes
	Lighting	Lighting (L)	Lights installed to prevent crimes at night on roads or in buildings
Natural access control	Buildings	Buildings (B)	Access control systems in buildings to prevent intrusion by outsiders
Activity support/ maintenance and management	Community facilities	Community (C1)	Community facilities established to prevent crimes and encourage activities
	Parks	Parks (P1)	Walkways improved to promote the use of parks and prevent crimes
	Use/place	Use/place (U)	Discovery and improvement of local space to resolve crime-vulnerable areas
	Parking	Parking (P2)	Improved parking environment and maintenance
	Urban control	Control (C2)	Crime prevention services based on municipal control for monitoring crime-prone areas
Territorial reinforcement/target hardening	Guide	Guide (G)	Local safety information provided to reinforce territoriality and harden targets in towns

shows the details of the implementation plan for each of the architectural urban planning elements.

3.1.2. Selection of Intelligent Crime Prevention Technologies

Crime prevention technology refers to the technology that supports activities to prevent crimes, and it has been rapidly evolving since the advent of the 4th Industrial Revolution. CCTV systems, which serve as the key crime prevention technology and the primary means of solving incidents and accidents, provide monitoring through integrated operations at municipal control centers. However, compared to the increasing number of CCTV systems, the number of monitoring personnel is limited, and some point out that it has poor effectiveness due to an increased workload and a missed critical time in crimes. To overcome these limitations in traditional control, many efforts are being made to upgrade the crime prevention system by incorporating intelligent technologies, such as AI, IoT, and big data.

This study reviews previous studies and related cases on intelligent crime prevention to configure intelligent CPTED systems. Key technologies in the intelligent crime prevention system consist of AI-based CCTV video analysis, the platform in the control room, and IoT-based data sensing. For CCTV video analysis, technologies to recognize the behaviors of perpetrators or track suspects in video footage have been mostly proposed. They include intelligent video surveillance (e.g., violence and loitering), the automatic tracking of suspects through collaboration between multiple CCTV systems, the facial recognition system, license plate identification, and GPS-linked CCTV security systems [27,28,29,30,31,32,33,34,35]. For the platform, technologies for spatial and temporal crime prediction in the control center (situation room), support for decision-making among managers, and integration/connection with related systems have primarily been proposed. They include the social safety net platform, the crime prediction system based on crime data patterns/trends, hot spot policing/crime mapping/geographical profiling, real-time video information linked to patrol vehicles, Geographic Information System (GIS)-based crime prevention decision-making support technologies, and the smart city integrated operation system [29,30,31,36,37,38,39]. For IoT, technologies such as sensors for intrusion prevention and lighting for night crime prevention have been primarily suggested. They include motion detection sensors in access control facilities, abnormal sound recognition, front door/window opening/closing sensors, pedestrian detection-based lighting adjustment technologies, IoT sensor-based smart street lights, and IoT-integrated police equipment [28,29,30,40,41,42,43].

Recently, there has been an increasing trend of crime prevention research through the combination of state-of-the-art technologies such as system improvement research for youth crime prevention [44], artificial intelligence utilization research for crime prediction and prevention [45], a study on the relationship between emotional recognition AI technology and crime, such as face movement and body language [46], and third generation CPTED and building information modeling integration research [47]. Looking at the differences from these studies, this study was able to expand spatiality and diversity by integrating CPTED with various advanced technologies and crime prevention technologies.

Along with previous studies, this study also reviewed the cases of smart cities in Korea with intelligent crime prevention technologies. It examined intelligent crime prevention technologies presented in Sejong Living Zone 5-1, which provide video and location information (3D map) when an event occurs through floor-by-floor surveillance in buildings, drone surveillance in complexes, and video surveillance in roads and tunnels. In particular, intelligent behavior recognition technologies are incorporated within buildings and complexes, such as building intrusion monitoring, loitering/abnormal signs, monitoring for objects abandoned, detection for objects left unattended, detection for abnormal behaviors in elevators, fire monitoring, and detection for violence signs. Based on IoT sensor infrastructure, Busan Eco Delta City monitors the city situation at all times; it disseminates a situation when an event occurs and has introduced the crime scene dispatch system. Moreover, a CCTV psychological video analysis solution for public spaces and major buildings was suggested to identify behaviors and a psychological state of mind among residents in terms of crime and suicide in real time. Furthermore, this study found crime safety services, such as facial recognition for criminals, license plate recognition and tracking, incident location, escape route, crime videos, emergency dispatch requests to the police and fire department, real-time video transmission for on-site situations, safety guarding (lighting and recording) for emergency calls regarding older adults and women, and enhanced crime prevention patrols in neglected areas for potential crimes (with robots and drones).

Based on the aforementioned intelligent crime prevention technologies and the technology elements of the CPTED guidelines, this study further classified intelligent crime prevention technologies. The technology elements by category include CCTV video analysis, IoT sensing, lighting, integrated operation control, positioning, guide, and community-building technologies. For the RIN, this study assigned 7 English codes for crime prevention technologies by category (A to G) and 34 codes for crime prevention technologies by sub-category (tech code; e.g., A1, A2).

Table 2. Intelligent crime prevention technology.

Category	Sub-Category (Technology/Service)	Tech Code	Description
A. Integrated operation control technologies	1. GIS-based public safety information analysis	A1	Decision-making support for new CCTV installation through the analysis of crime locations
	2. Big-data-based crime prediction	A2	Crime response and prediction by type through big data analytics
	3. Digital twin	A3	Real world implemented in virtual space
	4. Intelligent control platform	A4	Integrated control of CCTV systems, mobile apps, and crime maps
	5. Mobile app	A5	Mobile app used for supporting a safe return home
	6. Drone/robot control	A6	Unmanned patrol support
	7. Automatic link with 112 and 119	A7	Use of the smart city integrated platform to automatically link with 112 and 119
B. CCTV video analysis technologies	1. CCTV system for AI	B1	CCTV systems installed with features improved to enable AI video analysis
	2. Automatic recognition of abnormal behaviors	B2	Abnormal behaviors (e.g., loitering, clustering, or falling) automatically recognized through deep-learning pattern analysis
	3. Detection for intrusion in controlled areas	B3	Notification of intruders in controlled areas through video analysis
	4. Automatic object tracking	B4	Automatic tracking of objects between CCTV systems (e.g., suspects, missing people, and safe return home)
	5. Facial recognition	B5	Facial recognition to track missing people and criminal suspects
	6. License plate recognition	B6	Control for illegal parking, non-paid taxes, and stolen vehicles
C. IoT sensing technologies	1. Motion detection sensors	C1	Access control and motion detection in controlled areas
	2. Door opening/closing sensors	C2	Access control and door opening/closing detection in controlled areas
	3. Sound sensors	C3	Detection for screams and voices calling for emergency help
	4. Voice warnings	C4	Automatic voice warnings based on sensing data for abnormal situations (e.g., littering and loitering)
	5. Home sensor detection and remote control	C5	Detection and remote control for intrusion inside and outside the home
	6. Signal control	C6	Priority signals for emergency vehicles, and signal control for child/elderly protection zones
	7. Pedestrian environment safety/improvement technologies	C7	Safe walking environment based on pedestrian/vehicle detection (smart crosswalk)
D. Lighting technologies	1. Automatic illumination sensor lights	D1	Automatic adjustment for illumination based on the time of day and motion detection
	2. Logo lights	D2	Crime prevention warning and guide through floor lighting
	3. Floor guide lights	D3	Support for the movement of pedestrians and vehicles at night
	4. Smart poles	D4	Multifunctional services, such as CCTV systems in addition to lighting features, emergency reporting (e.g., emergency bell and link to the user’s mobile app), information on waiting nearby, and collection/delivery of vehicle/pedestrian information
E. Positioning technologies	1. Outdoor precise location	E1	Precise outdoor location information for users (e.g., missing children and older adults with dementia)
	2. Indoor precise location	E2	Precise outdoor location information for users (e.g., missing children and older adults with dementia)
	3. Legal guardian location	E3	Real-time location information to legal guardians regarding children and older adults with dementia
F. Guide technologies	1. Smart guide board	F1	Information guide board for crime prevention, major facilities, and living tips
	2. Variable guide facilities	F2	Variable guide facilities to prevent intrusion and ensure traffic safety (e.g., variable bollards)
G. Community-building technologies	1. Safe booths	G1	Safe booths/convenience stores to protect victims in case of emergency (e.g., emergency bell and automatic link to the control center/112)
	2. Smart shelters	G2	Smart shelters to protect children/older adults (e.g., crime prevention/living information guide and link to the control center)
	3. Smart garbage storage facilities	G3	Clean and efficient garbage storage facilities (for detecting the amount of garbage stored and monitoring illegal littering)
	4. Shared parking system	G4	Shared parking based on the use of idle parking spaces (generating parking income)
	5. Illegal parking monitoring	G5	User device-based illegal parking monitoring

lists more details.

3.1.3. Selection of Intelligent CPTED Systems

To select intelligent CPTED, this study combined the above implementation plan selected for each of the architectural urban planning elements (CPTED code) and intelligent crime prevention technologies (tech code). To facilitate a survey of residents and consider their introduction and operation in the control center, this study came up with 4 service content (categories) and 12 service sub-elements (sub-categories). In the RIN coding process, the final intelligent CPTED service RIN was set by combining the reference codes, respectively, that incorporated the intelligent CPTED service sub-element code and combined the tech code based on the CPTED code. The intelligent CPTED service sub-element code was derived by combining English initials by service content category and numbers by sub-category. The resulting intelligent CPTED service sub-element codes included “A1. CCTV system for AI” and “C3. Smart safety community”. The final reference code is the combined identification code of the preset CPTED code and tech code. Interdisciplinary combinations are created when one CPTED code is combined with multiple tech codes (Platform2_C2_A4A5) and multiple CPTED codes are combined with multiple tech codes (AI2_RP1_B2B6).

For architectural urban planning elements, roads were primarily combined with CCTV/video analysis for natural surveillance and IoT/lighting technologies. Lighting was mostly related to natural surveillance among other principles of CPTED and combined with lighting and street lighting technologies to prevent crimes at night. Buildings were mostly combined with IoT sensing technologies to prevent intrusion by outsiders based on the theory of natural access control. In terms of newly identified city control, platform service elements based on integrated operation control technologies were primarily derived. In particular, “A4. Intelligent control platform” among intelligent crime prevention technologies was selected for many key service elements, such as video analysis, platform, and community. Community facilities, use/place, and parking were primarily combined with community-building technologies, such as smart safety communities, based on the theory of activity support/maintenance and management. Finally, for a guide based on territorial reinforcement and target hardening, the AI guide system for village information to effectively delivers village information in real-time was derived (

Table 3. Technology elements of CPTED-based crime prevention services.

Service Content (Category)	Service Sub-Element (Sub-Category)	RIN	Description
AI and CCTV (A)	AI1_CCTV system for AI	AI1_RP1_B1	High-resolution CCTV systems added with features improved to enable crime analysis
	AI2_AI analysis for living safety	AI2_RP1_A4B2B6	Detection for littering, illegal parking, drunk people/smoking/juvenile delinquents
	AI2_AI analysis for crime prevention	AI3_RP1_A4B2B3B4B5	Automated tracking for suspects, facial recognition (e.g., missing people and older adults with dementia), and assault detection
IoT and lighting (I)	IoT1_IoT automatic situation recognition system	IoT1_RB_C1C2C3C4C5	Illegal access control, school zone safety control, and automatic recognition/warning for event situations
	IoT2_Intelligent pedestrian environment system	IoT2_R_C6C7F2	Walking guide lights for pedestrian safety at night, floor lighting (logo projectors), and school zone signal delays
	IoT3_Smart street lights	IoT3_RL_C3C4D1D2D3D4	Multifunctional smart street lights with scream recognition, voice warnings, emergency bells, and safety information
Platform (P)	Plat1_Crime monitoring system	Plat1_C2_A1A2A3A4A7	Crime prevention platform in the control center, including crime maps, CCTV analysis, and automatic link to 112
	Plat2_Mobile crime prevention app	Plat2_C2_A4A5	Smartphone app identifying real-time indoor and outdoor precise locations for children and women returning home
	Plat3_Real-time control drones/robots	Plat3_C2_A4A6	Drones/robots for automatic real-time patrols in crime-prone areas
Community (C)	Com1_AI guide system for village information	Com1_G_F1G1G2G3	AI-based information board for living safety and convenience information, crime prevention safety zones, foreign language guide (in areas highly populated by foreigners)
	Com2_Improved parking environment system	Com2_P2_A4G4G5	Improved parking environment with shared parking (generating revenue) and smartphone-based illegal parking reporting
	Com3_Smart safety community	Com3_C1U_A4G1G2G3	Safe return home for women, safety booths, smart shelters, and smart garbage storage facilities

).

3.2. Methods

3.2.1. Questionnaire Design

This section conducts an empirical analysis with service evaluation based on decision-making techniques to introduce intelligent CPTED systems in municipal control centers. A robust survey design is required for reliable service evaluation. First, the structure of the study was designed, including alternatives, evaluation criteria, hierarchy, questionnaire design, and the selection of evaluators. Alternatives were intelligent CPTED systems derived in Section 3.1.

This study reviewed major public sector guidelines for evaluating the value of technologies and services in Korea. After reviewing the Smart City Service Certification System [48], Korea Technology Finance Corporation Technology Evaluation [49], and Ministry of Land, Infrastructure, and Transport Technology Value Evaluation Manual [50], this study found that these guidelines had common evaluation criteria, such as conformity, technology, and marketability for evaluation. Finally, the following three evaluation criteria were selected for this study: conformity with the purpose of crime prevention, service suitability, and feasibility. This study used the sub-category evaluation criteria, which were the evaluation items derived from categories. More specifically, conformity with the purpose of crime prevention consisted of crime prevention effectiveness and policy conformity, while service suitability consisted of competitiveness between existing services and growth in the service sector. Feasibility consists of field applicability and sustainability (

Table 4. Evaluation criteria for intelligent CPTED technology elements.

Category	Sub-Category	Description of the Evaluation Criteria
Conformity with the purpose of crime prevention	Crime prevention effectiveness	Effectiveness on crime prevention from applying the service
	Policy conformity	Conformity with government and municipal crime prevention policies
Service conformity	Competitiveness with existing services	Competitiveness and usefulness with existing facilities/technologies/services
	Growth in the service sector	Technology trends and market potential in the service sector
Feasibility	Economic viability	Effectiveness on investment, economic viability, and commercialization potential
	Field applicability	Applicability to the field, including risk management (e.g., personal information and administrative delays) and collaboration with competent organizations (e.g., 112)
	Sustainability	Efficient management operation and sustainability

).

To establish a hierarchy among key attributes, it is required to organically design the hierarchy between attributes related to decision-making, that is, criteria and alternatives. The purpose of this study is as follows: the first class was to select intelligent CPTED systems based on municipal control centers and the second class was to identify seven evaluation criteria for evaluating alternatives. The third class of evaluation alternatives includes the 12 service models selected above (Figure 1).

Prior to the questionnaire survey, this study conducted a pilot survey and analysis with 13 PhD researchers who had a good understanding and knowledge of service elements and analytical models. From the pilot survey and analysis, this study double-checked the overall structure of the questionnaire survey, amended some terms in alternatives and evaluation criteria, and tested the analytical model. To ensure the multi-criteria decision-making (MCDM)-based expert questionnaire structure while making it easier for evaluators to understand the survey, this study gave the following guidance: the purpose of the survey, the concept of intelligent CPTED (including a concept map), as is/to be (the current CPTED vs. changes when incorporating intelligent CPTED), evaluation criteria, the scale and decision-making hierarchy of evaluation criteria, the description and example of the alternatives of technology elements, and a questionnaire example. For this survey, working-level employees in municipal control centers were selected as the evaluators to provide more professional and practical decision-making support in introducing intelligent CPTED systems. A total of 48 evaluators selected in this study were experts who operated crime prevention services. Hence, they had a very good understanding of this study and survey, which improved the reliability of the results of the analysis. This survey was conducted online in writing for two weeks from 7th to 20th April 2021. Under Article 33 of the Statistics Act in Korea, this study specified and informed the participants that all of their responses would be used for statistical purposes only, and the survey was conducted after obtaining their consent. The survey and related study were reviewed and approved by the Institutional Review Board of Gachon University (1044396-202303-HR-039-01). Accordingly, the need for consent was waived. More specifically, this survey ensured anonymity for all participants and fully informed them of the reason why the study was conducted, how the data would be used, and whether there was any risk. Under the Local Autonomy Act, Korea sets a population of 500,000 as the threshold to be classified as a large city. In this context, this study categorized the evaluators into 26 evaluators for large cities in the Seoul metropolitan area with a population of ≥500,000 and 22 evaluators for small and medium-sized cities in provinces with a population of <500,000 to conduct a comparative analysis by city size, not just a trend analysis among all the respondents. For reference, even if the number of respondents is small, the sample size would not become a problem in MCDM as long as expertise and logical consistency are ensured [51]. Microsoft Office Excel 2019 was used for analysis, which involved coding, analyzing, and documenting survey results.

3.2.2. Fuzzy TOPSIS

The present study aims to support decision-making in introducing intelligent CPTED systems for crime prevention in municipal control centers. Consequently, it is important to select a research methodology supported by scientific evidence and logic. In this regard, MCDM is widely used in the academic community to provide objective priority alternatives to judgment criteria depending on the research objective. TOPSIS can ensure rational logic for decision-making, represent the best and worst alternatives simultaneously in real values, and calculate numbers easily [52]. In particular, its strength lies in practicality as it can measure the performance of all alternatives from a multi-criteria perspective. This study incorporates the fuzzy theory proposed by Zadeh [53], which is used to solve the issue of uncertainty in subjective and ambiguous linguistic information based on TOPSIS. This theory introduces fuzzy logic and a fuzzy set to overcome the inaccuracy and ambiguity of subjective judgment in evaluation. As an advanced methodology, fuzzy TOPSIS can mathematically express ambiguous phenomena, including unclear quantitative information and subjective and unclear judgment, and identify reasonable alternatives for decision-making.

To evaluate intelligent CPTED systems based on fuzzy TOPSIS, first, evaluation criteria are selected and examined, and then the weights of the resulting evaluation criteria are applied to evaluate alternatives. For evaluation criteria, a 7-point Likert scale was used, considering that the survey was conducted with experts skilled in technical evaluation, and the ease of evaluation was important. The Likert scale is a bipolar scale method that sequentially measures positive and negative responses to questionnaire sentences. In order to quantify the evaluation data value more objectively, this study applied a triangular fuzzy number (TFN) that was fuzzified using a linear membership function. For TFN M was used to calculate weights, ${\mathrm{M}}_1=({\mathrm{l}}_1,\hspace{0.17em}{\mathrm{m}}_1,\hspace{0.17em}{\mathrm{u}}_1)$—membership function for the lower, median, and upper bounds for a single crisp value—was modified to $({\mathrm{a}}_1,\hspace{0.17em}{\mathrm{a}}_2,\hspace{0.17em}{\mathrm{a}}_3)$ (Figure 2).

The TFN is represented as a triangle of three points (l, m, and u; l = lower bound, m = middle bound, and u = upper bound) and their area is the size of the TFN [54,55,56]. To derive the TFN, this study used the scale criteria and fuzzy scale in

Table 5. Fuzzy scale for evaluation criteria.

Language Scale	TFN
Extremely low (e.l)	(0, 0, 0.1)
Very low (v.l)	(0, 0.1, 0.3)
Low (l)	(0.1, 0.3, 0.5)
Medium (m)	(0.3, 0.5, 0.7)
High (h)	(0.5, 0.7, 0.9)
Very high (v.h)	(0.7, 0.9, 1.0)
Extremely high (e.h)	(0.9, 1.0, 1.0)

[57].

To derive evaluation, Equations (1)–(4) were used to identify TFN ${\mathrm{S}}_{\mathrm{i}}=({\mathrm{l}}_{\mathrm{i}},{\mathrm{m}}_{\mathrm{i}},{\mathrm{u}}_{\mathrm{i}})$ for the ith attribute and then finally calculate the eigenvector normalized value of the minimum for the defuzzification of TFN ${\mathrm{S}}_{\mathrm{i}}$ by evaluation criteria [56,58,59,60]. Equation (1) is the sum of the scores evaluated before fuzzification for each of the items, and Equation (2) is the triangular fuzzification based on the values of l, m, and u under the fuzzy scale. Equation (3) is the sum of the evaluation criteria for each value of l, m, and u, and Equation (4) is the final TFN, which is the reciprocal of the sum of l, m, and u.

$${\mathrm{S}}_{\mathrm{i}}={\sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{M}}_{\mathrm{i}\mathrm{j}}\times [{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{M}}_{\mathrm{i}\mathrm{j}}{]}^{-1}$$

(1)

$$\mathrm{s}.\mathrm{t}{\sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{M}}_{\mathrm{i}\mathrm{j}}=({\sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{l}}_{\mathrm{i}\mathrm{j}},{\sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{m}}_{\mathrm{i}\mathrm{j}},{\sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{u}}_{\mathrm{i}\mathrm{j}})$$

(2)

$${\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{M}}_{\mathrm{i}\mathrm{j}}=({\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{l}}_{\mathrm{i}\mathrm{j}},{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{m}}_{\mathrm{i}\mathrm{j}},{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{u}}_{\mathrm{i}\mathrm{j}})$$

(3)

$${\left[{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\sum }_{\mathrm{j}=1}^{\mathrm{m}}\right]}^{-1}=[{\displaystyle \frac{1}{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{u}}_{\mathrm{i}\mathrm{j}}}},{\displaystyle \frac{1}{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{m}}_{\mathrm{i}\mathrm{j}}}},{\displaystyle \frac{1}{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{l}}_{\mathrm{i}\mathrm{j}}}}]$$

(4)

For alternatives, the respondents were asked to rate 84 alternatives in total: 12 alternatives for each of the 7 criteria. Considering fatigue associated with filling out the questionnaire survey, this study used a 5-point Likert scale for alternatives. The fuzzification of alternatives by criterion collected from multiple evaluators is calculated as in Equation (5), and the fuzzy weight is calculated as in Equation (6) [61].

$${\mathrm{a}}_{\mathrm{i}\mathrm{j}}={\mathrm{m}\mathrm{i}\mathrm{n}}_{\mathrm{k}}({{\mathrm{a}}_{\mathrm{i}\mathrm{j}}}^{\mathrm{k}}),\hspace{0.17em}{\mathrm{b}}_{\mathrm{i}\mathrm{j}}={\displaystyle \frac{1}{\mathrm{k}}}{\sum }_{\mathrm{k}=1}^{\mathrm{k}}{{\mathrm{b}}_{\mathrm{i}\mathrm{j}}}^{\mathrm{k}},\hspace{0.17em}{\mathrm{c}}_{\mathrm{i}\mathrm{j}}={\mathrm{m}\mathrm{a}\mathrm{x}}_{\mathrm{k}}({{\mathrm{c}}_{\mathrm{i}\mathrm{j}}}^{\mathrm{k}})$$

(5)

$${\mathrm{w}}_{\mathrm{j}1}={\mathrm{m}\mathrm{i}\mathrm{n}}_{\mathrm{k}}({{\mathrm{w}}_{\mathrm{j}1}}^{\mathrm{k}}),\hspace{0.17em}{\mathrm{w}}_{\mathrm{j}2}={\displaystyle \frac{1}{\mathrm{k}}}{\sum }_{\mathrm{k}=1}^{\mathrm{k}}{{\mathrm{w}}_{\mathrm{j}2}}^{\mathrm{k}},\hspace{0.17em}{\mathrm{w}}_{\mathrm{j}3}={\mathrm{m}\mathrm{a}\mathrm{x}}_{\mathrm{k}}({{\mathrm{w}}_{\mathrm{j}3}}^{\mathrm{k}})$$

(6)

Then, fuzzy decision matrices $\tilde{\mathrm{V}}=\left(\widetilde{{\mathrm{v}}_{\mathrm{i}\mathrm{j}}}\right)$ and $\widetilde{{\mathrm{v}}_{\mathrm{i}\mathrm{j}}}=\widetilde{{\mathrm{r}}_{\mathrm{i}\mathrm{j}}}\times {\mathrm{w}}_{\mathrm{j}}$ are obtained with a normalized fuzzy decision matrix $\tilde{\mathrm{R}}=\left[\widetilde{{\mathrm{r}}_{\mathrm{i}\mathrm{j}}}\right]$ (Equations (7) and (8)) and the weight of evaluation criteria (TFN) from Equation (4) [61].

$$\widetilde{{\mathrm{r}}_{\mathrm{i}\mathrm{j}}}=({\displaystyle \frac{{\mathrm{a}}_{\mathrm{i}\mathrm{j}}}{{\mathrm{c}}_{\mathrm{j}}^{*}}},{\displaystyle \frac{{\mathrm{b}}_{\mathrm{i}\mathrm{j}}}{{\mathrm{c}}_{\mathrm{j}}^{*}}},{\displaystyle \frac{{\mathrm{c}}_{\mathrm{i}\mathrm{j}}}{{\mathrm{c}}_{\mathrm{j}}^{*}}})\hspace{0.17em}\mathrm{a}\mathrm{n}\mathrm{d}\hspace{0.17em}\hspace{0.17em}{\mathrm{c}}_{\mathrm{j}}^{*}={\mathrm{m}\mathrm{a}\mathrm{x}}_{\mathrm{i}}({\mathrm{c}}_{\mathrm{i}\mathrm{j}})\hspace{0.17em}\hspace{0.17em}(\mathrm{b}\mathrm{e}\mathrm{n}\mathrm{e}\mathrm{f}\mathrm{i}\mathrm{t}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{i}\mathrm{a})$$

(7)

$$\widetilde{{\mathrm{r}}_{\mathrm{i}\mathrm{j}}}=({\displaystyle \frac{{\mathrm{a}}_{\mathrm{j}}^{-}}{{\mathrm{c}}_{\mathrm{i}\mathrm{j}}}},{\displaystyle \frac{{\mathrm{a}}_{\mathrm{j}}^{-}}{{\mathrm{b}}_{\mathrm{i}\mathrm{j}}}},{\displaystyle \frac{{\mathrm{a}}_{\mathrm{j}}^{-}}{{\mathrm{a}}_{\mathrm{i}\mathrm{j}}}})\hspace{0.17em}\mathrm{a}\mathrm{n}\mathrm{d}\hspace{0.17em}\hspace{0.17em}{{\mathrm{c}}_{\mathrm{j}}}^{-}={\mathrm{m}\mathrm{i}\mathrm{n}}_{\mathrm{i}}({\mathrm{a}}_{\mathrm{i}\mathrm{j}})\hspace{0.17em}\hspace{0.17em}(\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{t}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{i}\mathrm{a})$$

(8)

Afterwards, the fuzzy positive ideal solution (FPIS) and the fuzzy negative ideal solution (FNIS) are calculated as described in Equations (9) and (10) [61].

$${\mathrm{A}}^{*}=({\tilde{\mathrm{v}}}_1^{*},\hspace{0.17em}{\tilde{\mathrm{v}}}_2^{*},\hspace{0.17em}\cdots ,\hspace{0.17em}{\tilde{\mathrm{v}}}_{\mathrm{n}}^{*}),\hspace{0.17em}\hspace{0.17em}\mathrm{w}\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{e}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}{\tilde{\mathrm{v}}}_{\mathrm{j}}^{*}\hspace{0.17em}=\hspace{0.17em}{\mathrm{m}\mathrm{a}\mathrm{x}}_{\mathrm{i}}\left({\mathrm{v}}_{\mathrm{i}\mathrm{j}3}\right)$$

(9)

$${\mathrm{A}}^{-}=({\tilde{\mathrm{v}}}_1^{-},\hspace{0.17em}{\tilde{\mathrm{v}}}_2^{-},\hspace{0.17em}\cdots ,\hspace{0.17em}{\tilde{\mathrm{v}}}_{\mathrm{n}}^{-}),\hspace{0.17em}\hspace{0.17em}\mathrm{w}\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{e}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}{\tilde{\mathrm{v}}}_{\mathrm{j}}^{-}\hspace{0.17em}=\hspace{0.17em}{\mathrm{m}\mathrm{i}\mathrm{n}}_{\mathrm{i}}\left({\mathrm{v}}_{\mathrm{i}\mathrm{j}1}\right)$$

(10)

To measure the distance to the FPIS and FNIS for each of the alternatives, this study used the two-TFN distance calculation method proposed by Chen [57]. When $\tilde{\mathrm{x}}=({\mathrm{a}}_1,{\mathrm{b}}_1,{\mathrm{c}}_1),\tilde{\mathrm{y}}=({\mathrm{a}}_2,{\mathrm{b}}_2,{\mathrm{c}}_2)$ for two TFNs, the distance is calculated as in Equation (11), and the final distance to the FPIS and FNIS in each of the alternatives is calculated as in Equation (12) [61].

$$\mathrm{d}(\tilde{\mathrm{x}},\hspace{0.17em}\tilde{\mathrm{y}})\hspace{0.17em}=\hspace{0.17em}\sqrt{{\displaystyle \frac{1}{3}}[({\mathrm{a}}_1-{\mathrm{a}}_2{)}^2+({\mathrm{b}}_1-{\mathrm{b}}_2{)}^2+({\mathrm{c}}_1-{\mathrm{c}}_2{)}^2]}$$

(11)

$${\mathrm{d}}_{\mathrm{i}}^{*}={\sum }_{\mathrm{j}=1}^{\mathrm{n}}\mathrm{d}({\tilde{\mathrm{v}}}_{\mathrm{i}\mathrm{j}},\hspace{0.17em}{\tilde{\mathrm{v}}}_{\mathrm{j}}^{*}),\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}{\mathrm{d}}_{\mathrm{i}}^{-}={\sum }_{\mathrm{j}=1}^{\mathrm{n}}\mathrm{d}({\tilde{\mathrm{v}}}_{\mathrm{i}\mathrm{j}},\hspace{0.17em}{\tilde{\mathrm{v}}}_{\mathrm{j}}^{-})$$

(12)

Lastly, after calculating ${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$, a proximity coefficient for each of the alternatives, as in Equation (13), the ranks of the alternatives are calculated from the highest ${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$ to the lowest [61].

$${\mathrm{C}\mathrm{C}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{d}}_{\mathrm{i}}^{-}}{{\mathrm{d}}_{\mathrm{i}}^{-}+{\mathrm{d}}_{\mathrm{i}}^{*}}}$$

(13)

4. Empirical Analysis

4.1. Fuzzy TOPSIS Analysis for Evaluation Criteria

The fuzzy scale in Table 5 was applied based on the absolute evaluation results to develop the TFN for the evaluation criteria of fuzzy TOPSIS. The geometric average fuzzy weight was derived by triangular fuzzification for each evaluation criterion of 48 effective samples, and TFN values were calculated by Equations (1)–(4). Based on the above, the minimum eigenvector value for defuzzification, that is, the normalized weight, and the final rank by evaluation criterion were identified (

Table 6. Result of fuzzy TOPSIS analysis for evaluation criteria.

Evaluation Criteria	TFN	Weight (Normalized)	Rank
Crime prevention effectiveness	(0.10, 0.19, 0.36)	0.24	1
Policy conformity	(0.04, 0.11, 0.29)	0.09	6
Competitiveness with existing services	(0.04, 0.10, 0.28)	0.08	7
Growth in the service sector	(0.06, 0.15, 0.34)	0.14	4
Economic viability	(0.05, 0.13, 0.32)	0.12	5
Field applicability	(0.07, 0.16, 0.35)	0.16	3
Sustainability	(0.07, 0.16, 0.35)	0.17	2

).

The analysis showed that the TFN was (0.10, 0.19, 0.36) for crime prevention effectiveness, (0.04, 0.11, 0.29) for policy conformity, (0.04, 0.10, 0.28) for competitiveness with existing services, (0.06, 0.15, 0.34) for growth in the service sector, (0.05, 0.13, 0.32) for economic viability, (0.07, 0.16, 0.35) for field applicability, and (0.07, 0.16, 0.35) for sustainability. In terms of ranks, crime prevention effectiveness had the highest weight of 0.24, followed by sustainability (0.17), field applicability (0.16), and growth in the service sector (0.14). The results likely reflected the evaluators’ opinions about the practical realization, field operation, and service sustainability of the technologies. Notably, economic viability ranked just fifth, indicating that the evaluators did not mind economic viability when adopting new technologies, such as intelligent CPTED.

4.2. Analysis of Alternatives from All Respondents

To evaluate alternatives based on fuzzy TOPSIS, the fuzzy scale was applied to each evaluation score to conduct a triangular fuzzification of the alternatives by evaluation criterion.

Table 7. Triangular fuzzification to evaluate alternatives based on fuzzy TOPSIS.

Category	Sub-Category (Service Sub-Element)	Conformity with the Purpose of Crime Prevention		Service Conformity		Feasibility
		Crime Prevention Effectiveness	Policy Conformity	Competitiveness with Existing Services	Growth in the Service Sector	Economic Viability	Field Applicability	Sustainability
AI/ CCTV	CCTV system for AI	(1.5, 3.5, 4.5)	(0.7, 3.3, 4.5)	(0.7, 3.0, 4.5)	(0.7, 3.0, 4.5)	(0.7, 3.2, 4.5)	(0.7, 3.3, 4.5)	(0.7, 3.4, 4.5)
	AI video analysis for living safety	(0.7, 2.6, 4.5)	(1.5, 3.0, 4.5)	(0.7, 2.6, 4.5)	(0.7, 2.8, 4.5)	(0.7, 2.7, 4.5)	(1.5, 3.0, 4.5)	(0.7, 2.9, 4.5)
	AI video analysis for crime prevention	(0.7, 2.9, 4.5)	(0.7, 3.0, 4.5)	(0.7, 3.0, 4.5)	(0.7, 3.3, 4.5)	(0.7, 2.6, 4.5)	(0.7, 2.8, 4.5)	(0.7, 2.9, 4.5)
IoT/ lighting	IoT automatic situation recognition system	(0.7, 2.6, 4.5)	(0.7, 2.7, 4.5)	(0.7, 2.3, 4.5)	(0.7, 2.5, 4.5)	(0.7, 2.2, 4.5)	(0.7, 2.4, 4.5)	(0.7, 2.3, 4.5)
	Intelligent pedestrian environment system	(0.7, 2.6, 4.5)	(0.7, 2.7, 4.5)	(0.7, 2.3, 4.5)	(0.7, 2.5, 4.5)	(0.7, 2.5, 4.5)	(0.7, 2.8, 4.5)	(0.7, 2.6, 4.5)
	Smart street lights	(0.7, 2.3, 4.5)	(0.7, 2.4, 4.5)	(0.7, 2.1, 4.5)	(0.7, 2.2, 4.5)	(0.7, 2.2, 4.5)	(0.7, 2.4, 4.5)	(0.7, 2.4, 4.5)
Platform	Crime monitoring system	(0.7, 2.9, 4.5)	(0.7, 3.1, 4.5)	(0.7, 2.9, 4.5)	(0.7, 3.1, 4.5)	(0.7, 3.0, 4.5)	(0.7, 3.1, 4.5)	(0.7, 3.0, 4.5)
	Mobile crime prevention app	(0.7, 2.3, 4.5)	(0.7, 2.5, 4.5)	(0.7, 2.1, 4.5)	(0.7, 2.4, 4.5)	(0.7, 2.1, 4.5)	(0.7, 2.4, 4.5)	(0.7, 2.2, 4.5)
	Real-time control drones/robots	(0.7, 1.7, 4.5)	(0.7, 1.9, 4.5)	(0.7, 1.9, 4.5)	(0.7, 2.3, 4.5)	(0.7, 1.7, 4.5)	(0.7, 1.7, 4.5)	(0.7, 1.6, 4.5)
Community	AI guide system for village information	(0.7, 1.6, 3.5)	(0.7, 2.0, 4.5)	(0.7, 1.6, 4.5)	(0.7, 1.7, 4.5)	(0.7, 1.6, 4.5)	(0.7, 2.3, 4.5)	(0.7, 1.8, 4.5)
	Improved parking environment system	(0.7, 2.3, 4.5)	(0.7, 2.5, 4.5)	(0.7, 2.2, 4.5)	(0.7, 2.4, 4.5)	(0.7, 2.3, 4.5)	(0.7, 2.5, 4.5)	(0.7, 2.2, 4.5)
	Smart safety community	(0.7, 2.3, 4.5)	(0.7, 2.5, 4.5)	(0.7, 2.0, 3.5)	(0.7, 2.2, 4.5)	(0.7, 2.1, 4.5)	(0.7, 2.3, 4.5)	(0.7, 2.2, 4.5)

lists the fuzzification results of the 48 valid samples for the alternatives with Equations (5) and (6).

Equations (7) and (8) were used to calculate the normalized fuzzy decision matrix and the weighted fuzzy decision matrix. To use the evaluation of the alternative to the CCTV system for AI in the evaluation criterion of crime prevention effectiveness as an example, the upper bound (${{\mathrm{c}}_{\mathrm{j}}}^{*}$) for crime prevention effectiveness was 4.5; therefore, the decision matrix was $\overline{\mathrm{R}}=({\displaystyle \frac{1.5}{4.5}},{\displaystyle \frac{3.5}{4.5}},{\displaystyle \frac{4.5}{4.5}})$.

The weighted fuzzy decision matrix was calculated as $\overline{\mathrm{V}}=({\displaystyle \frac{1.5}{4.5}},{\displaystyle \frac{3.5}{4.5}},{\displaystyle \frac{4.5}{4.5}})\times (0.10,0.19,0.36)=(0.03,0.15,0.36)$.

Then, the FPIS and FNIS were calculated using Equations (9) and (10) (

Table 8. Results of the FPIS and FNIS.

Category	Sub-Category (Service Sub-Element)	Conformity with the Purpose of Crime Prevention		Service Conformity		Feasibility
		Crime Prevention Effectiveness	Policy Conformity	Competitiveness with Existing Services	Growth in the Service Sector	Economic Viability	Field Applicability	Sustainability
Weight of evaluation criteria		(0.10, 0.19, 0.36)	(0.04, 0.11, 0.29)	(0.04, 0.10, 0.28)	(0.06, 0.15, 0.34)	(0.05, 0.13, 0.32)	(0.07, 0.16, 0.35)	(0.07, 0.16, 0.35)
AI and CCTV	CCTV system for AI	(0.03, 0.15, 0.36)	(0.01, 0.08, 0.29)	(0.01, 0.07, 0.28)	(0.01, 0.10, 0.34)	(0.01, 0.09, 0.32)	(0.01, 0.12, 0.35)	(0.01, 0.12, 0.35)
	AI video analysis for living safety	(0.02, 0.11, 0.36)	(0.01, 0.07, 0.29)	(0.01, 0.06, 0.28)	(0.01, 0.09, 0.34)	(0.01, 0.08, 0.32)	(0.02, 0.11, 0.35)	(0.01, 0.11, 0.35)
	AI video analysis for crime prevention	(0.02, 0.12, 0.36)	(0.01, 0.07, 0.29)	(0.01, 0.07, 0.28)	(0.01, 0.11, 0.34)	(0.01, 0.08, 0.32)	(0.01, 0.10, 0.35)	(0.01, 0.11, 0.35)
IoT and lighting	IoT automatic situation recognition system	(0.02, 0.11, 0.36)	(0.01, 0.06, 0.29)	(0.01, 0.05, 0.28)	(0.01, 0.08, 0.34)	(0.01, 0.06, 0.32)	(0.01, 0.09, 0.35)	(0.01, 0.08, 0.35)
	Intelligent pedestrian environment system	(0.02, 0.11, 0.36)	(0.01, 0.06, 0.29)	(0.01, 0.05, 0.28)	(0.01, 0.08, 0.34)	(0.01, 0.07, 0.32)	(0.01, 0.10, 0.35)	(0.01, 0.09, 0.35)
	Smart street lights	(0.02, 0.10, 0.36)	(0.01, 0.06, 0.29)	(0.01, 0.05, 0.28)	(0.01, 0.07, 0.34)	(0.01, 0.06, 0.32)	(0.01, 0.09, 0.35)	(0.01, 0.09, 0.35)
Platform	Crime monitoring system	(0.02, 0.12, 0.36)	(0.01, 0.07, 0.29)	(0.01, 0.07, 0.28)	(0.01, 0.10, 0.34)	(0.01, 0.09, 0.32)	(0.01, 0.11, 0.35)	(0.01, 0.11, 0.35)
	Mobile crime prevention app	(0.02, 0.10, 0.36)	(0.01, 0.06, 0.29)	(0.01, 0.05, 0.28)	(0.01, 0.08, 0.34)	(0.01, 0.06, 0.32)	(0.01, 0.08, 0.35)	(0.01, 0.08, 0.35)
	Real-time control drones/robots	(0.02, 0.07, 0.36)	(0.01, 0.04, 0.29)	(0.01, 0.04, 0.28)	(0.01, 0.08, 0.34)	(0.01, 0.05, 0.32)	(0.01, 0.06, 0.35)	(0.01, 0.06, 0.35)
Community	AI guide system for village information	(0.02, 0.07, 0.28)	(0.01, 0.05, 0.29)	(0.01, 0.04, 0.28)	(0.01, 0.06, 0.34)	(0.01, 0.05, 0.32)	(0.01, 0.08, 0.35)	(0.01, 0.07, 0.35)
	Improved parking environment system	(0.02, 0.10, 0.36)	(0.01, 0.06, 0.29)	(0.01, 0.05, 0.28)	(0.01, 0.08, 0.34)	(0.01, 0.07, 0.32)	(0.01, 0.09, 0.35)	(0.01, 0.08, 0.35)
	Smart safety community	(0.02, 0.10, 0.36)	(0.01, 0.06, 0.29)	(0.01, 0.04, 0.22)	(0.01, 0.07, 0.34)	(0.01, 0.06, 0.32)	(0.01, 0.08, 0.35)	(0.01, 0.08, 0.35)
A *		(0.03, 0.15, 0.36)	(0.01, 0.08, 0.29)	(0.01, 0.07, 0.28)	(0.01, 0.11, 0.34)	(0.01, 0.09, 0.32)	(0.02, 0.12, 0.35)	(0.01, 0.12, 0.35)
A−		(0.02, 0.07, 0.28)	(0.01, 0.04, 0.29)	(0.01, 0.04, 0.22)	(0.01, 0.06, 0.34)	(0.01, 0.05, 0.32)	(0.01, 0.06, 0.35)	(0.01, 0.06, 0.35)

).

Using Equation (11), this study calculated the distance to the FPIS and FNIS by alternative. To use the FNIS for the CCTV system for AI in the evaluation criterion of crime prevention effectiveness, it was $\mathrm{d}(\tilde{\mathrm{x}},\overline{\mathrm{y}})=\sqrt{{\displaystyle \frac{1}{3}}\left[\right(0.03-0.02{)}^2+(0.15-0.07{)}^2+\left(0.36-0.28{)}^2\right]}$ = 0.067.

Using Equation (12), this study calculated the final distance to the FPIS (${\mathrm{d}}_{\mathrm{i}}^{*}$) and FNIS (${\mathrm{d}}_{\mathrm{i}}^{-}$) in each of the alternatives (

Table 9. Results of the distance to the FPIS by alternative.

Category	Sub-Category (Service Sub-Element)	Conformity with the Purpose of Crime Prevention		Service Conformity		Feasibility			FPIS (${\mathbf{d}}_{\mathbf{i}}^{\mathbf{*}}$)
		Crime Prevention Effectiveness	Policy Conformity	Competitiveness with Existing Services	Growth in the Service Sector	Economic Viability	Field Applicability	Sustainability
AI and CCTV	CCTV system for AI	0.000	0.004	0.001	0.006	0.000	0.008	0.000	0.019
	AI video analysis for living safety	0.024	0.004	0.005	0.009	0.008	0.006	0.010	0.065
	AI video analysis for crime prevention	0.019	0.006	0.000	0.000	0.009	0.013	0.009	0.056
IoT and lighting	IoT automatic situation recognition system	0.026	0.009	0.009	0.014	0.017	0.019	0.022	0.117
	Intelligent pedestrian environment system	0.025	0.009	0.010	0.015	0.011	0.013	0.016	0.099
	Smart street lights	0.033	0.014	0.012	0.021	0.017	0.020	0.020	0.136
Platform	Crime monitoring system	0.018	0.005	0.001	0.004	0.004	0.009	0.008	0.049
	Mobile crime prevention app	0.031	0.013	0.011	0.017	0.018	0.021	0.024	0.134
	Real-time control drones/robots	0.046	0.020	0.014	0.019	0.025	0.034	0.037	0.194
Community	AI guide system for village information	0.067	0.019	0.018	0.030	0.027	0.023	0.032	0.215
	Improved parking environment system	0.032	0.012	0.010	0.018	0.015	0.017	0.024	0.128
	Smart safety community	0.032	0.012	0.038	0.021	0.018	0.021	0.024	0.167

and

Table 10. Results of the distance to the FNIS by alternative.

Category	Sub-Category (Service Sub-Element)	Conformity with the Purpose of Crime Prevention		Service Conformity		Feasibility			FNIS (${\mathbf{d}}_{\mathbf{i}}^{\mathbf{*}}$)
		Crime Prevention Effectiveness	Policy Conformity	Competitiveness with Existing Services	Growth in the Service Sector	Economic Viability	Field Applicability	Sustainability
AI and CCTV	CCTV system for AI	0.067	0.019	0.040	0.024	0.027	0.033	0.037	0.247
	AI video analysis for living safety	0.053	0.016	0.038	0.021	0.019	0.028	0.028	0.203
	AI video analysis for crime prevention	0.056	0.015	0.040	0.030	0.018	0.022	0.028	0.209
IoT and lighting	IoT automatic situation recognition system	0.052	0.011	0.037	0.016	0.010	0.015	0.015	0.156
	Intelligent pedestrian environment system	0.052	0.011	0.037	0.015	0.016	0.022	0.021	0.174
	Smart street lights	0.049	0.006	0.036	0.010	0.010	0.015	0.017	0.143
Platform	Crime monitoring system	0.057	0.016	0.040	0.026	0.023	0.028	0.029	0.219
	Mobile crime prevention app	0.050	0.007	0.036	0.013	0.009	0.013	0.013	0.143
	Real-time control drones/robots	0.047	0.000	0.036	0.012	0.001	0.000	0.000	0.096
Community	AI guide system for village information	0.000	0.001	0.036	0.000	0.000	0.012	0.005	0.053
	Improved parking environment system	0.049	0.008	0.037	0.013	0.012	0.017	0.014	0.150
	Smart safety community	0.049	0.008	0.004	0.009	0.009	0.013	0.013	0.106

).

Using Equation (13), this study calculated the proximity coefficient (${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$) and rank by alternative, which are presented in

Table 11. Results of fuzzy TOPSIS for alternatives.

Category	Sub-Category (Service Sub-Element)	$\mathbf{FPIS} ({\mathbf{d}}_{\mathbf{i}}^{\mathbf{*}}$)	$\mathbf{FNIS} ({\mathbf{d}}_{\mathbf{i}}^{-}$)	${\mathbf{C}\mathbf{C}}_{\mathbf{i}}$	Rank
AI and CCTV	CCTV system for AI (e.g., high-resolution CCTV systems, Image function improvement)	0.019	0.247	0.93	1
	AI video analysis for living safety (e.g., detection for littering, drunk people/smoking)	0.065	0.203	0.76	4
	AI video analysis for crime prevention (e.g., automated tracking, assault detection)	0.056	0.209	0.79	3
IoT and lighting	IoT automatic situation recognition system (e.g., automatic recognition/warning)	0.117	0.156	0.57	6
	Intelligent pedestrian environment system (e.g., walking guide lights for pedestrian safety)	0.099	0.174	0.64	5
	Smart street lights (e.g., scream recognition, emergency bells, voice warnings)	0.136	0.143	0.51	9
Platform	Crime monitoring system (e.g., crime maps, CCTV analysis, automatic link to 112)	0.049	0.219	0.82	2
	Mobile crime prevention app (e.g., real-time indoor and outdoor precise locations)	0.134	0.143	0.51	8
	Real-time control drones/robots (e.g., drones/robots for automatic real-time patrols)	0.194	0.096	0.33	11
Community	AI guide system for village information (e.g., crime prevention safety zones)	0.215	0.053	0.20	12
	Improved parking environment system (e.g., shared parking, illegal parking reporting)	0.128	0.150	0.54	7
	Smart safety community (e.g., Safe return home for women, safety booths)	0.167	0.106	0.39	10

. First, this study calculated the mean of ${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$ for each of the four categories and examined the trend in the importance of the categories of intelligent CPTED systems. AI/CCTV showed the highest priority with 0.83, which was followed by IoT/lighting with 0.57, platform with 0.55, and community with 0.38. It seems that the respondents gave more priority to the elements that could be directly monitored from the perspective of controllers, such as CCTV systems, IoT, and the crime prevention system.

The analysis of technology elements showed that the CCTV system for AI ranked first, with 0.93 in ${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$. Crime prevention effectiveness, economic viability, and sustainability were found to be more important than other alternatives, and policy conformity and field applicability were also considered highly important. The results seem to reflect that CCTV combined with AI, which has become a hot topic recently, may serve as a powerful tool to provide clues to solve crimes in reality. The crime monitoring system (${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$: 0.82), which was evaluated evenly across all evaluation criteria, ranked second. The results likely reflect the value of the services from the public sector’s point of view as most of the technologies and services are combined and operated based on a platform in the control centers in which the evaluators work. Moreover, the results might reflect some value in efficiency to replace the current manual monitoring personnel. AI video analysis for crime prevention (${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$: 0.79) ranked third, which was evaluated most highly in terms of competitiveness with existing services and growth in the service sector. The results indicate that the evaluators had very high expectations for the potential of the service elements. AI video analysis for living safety (${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$: 0.76) ranked fourth. The results seem to reflect that AI video analysis based on CCTV systems holds importance in not only preventing crimes but also ensuring living safety. This study conjectures that the evaluators selected the first to fourth ranks as the top technology elements due to their value as underlying solutions required to provide other technology elements as a service.

The intelligent pedestrian environment system (${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$: 0.64) ranked fifth, and the IoT automatic situation recognition system (${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$: 0.57) ranked sixth. These technology elements were considered important after CCTV systems; they are easy to install and operate locally and provide visually differentiated services. By contrast, smart street lights combined with various IoT sensors and lighting technologies (${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$: 0.51) ranked ninth, which was evaluated relatively poorly. The results seem to reflect some concerns that the integration and operation of IoT services for crime prevention would increase the number of management points and result in some negligence in managing smart street light infrastructure. Attempts must be made to improve cost-effectiveness and automated management. Mobile crime prevention app (${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$: 0.51) ranked eighth. The respondents seem to consider that even though a growing number of municipalities operate mobile-based safe return home services, only outdoor mobile location services can be available and operational. Indoor spatial information must be established to precisely identify indoor locations based on users’ mobile devices; thus, it is still difficult to widely use such services in terms of budget and management. Except for the top elements (ranked first to fourth), which are underlying infrastructure solutions, the elements ranked fifth to eighth are considered to be the service elements highly valued as personal safety services.

Among service elements in the category of community, the improved parking environment system (${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$: 0.54), smart safety community (${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$: 0.39), and AI guide system for village information (${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$: 0.20) ranked 7th, 10th, and 12th, near the bottom. The reason for these low ranks is in part because they are indirect forms of support without directly engaging in crime response and prevention. Notably, even though the AI guide system for village information provided a variety of information about crime prevention to residents, it was rated most poorly among other intelligent CPTED technology elements. Although it was ranked poorly, it is considered necessary to strengthen CPTED systems, such as establishing governance between the public sector and residents as an interactive service element enabling the residents to participate.

Real-time control drones/robots (${\mathrm{C}\mathrm{C}}_{\mathrm{i}}$: 0.33) ranked eleventh, which was also very low. The results may reflect that while the drones/robots support automatic patrol and control in real-time, the cost of investing in the drones/robots remains high; such new technologies are applied on-site in only a few cases and may end up as a one-off measure.

4.3. Comparison of Alternatives by City Size

Among 48 valid samples in total, 26 respondents were from large cities in the Seoul metropolitan area and 22 respondents were from small and medium-sized cities in other provinces. While Korea has grown economically and urbanized very rapidly over the past 50 years, income inequalities and regional imbalances remain serious issues. These regional differences also result in different levels of municipal finances, which, in turn, lead to huge differences in the level of crime prevention infrastructure and control between different municipalities. Large cities in the Seoul metropolitan area with a population of ≥500,000 people have some of the best crime prevention infrastructure in Korea and even around the world. Small and medium-sized cities in other provinces with a population of <500,000 people have a relatively low level of control due to difficulties in expanding crime prevention infrastructure because of a lack of municipal finances and low control efficiency arising from low population density. To compare intelligent CPTED service elements by city size, this study categorizes and analyzes large cities in the Seoul metropolitan area and small and medium-sized cities in other provinces (

Table 12. Results of fuzzy TOPSIS for alternatives by city size.

Category	Sub-Category (Service Sub-Element)	Large Cities in the Seoul Metropolitan Area (n = 26)		Small and Medium-Sized Cities in Other Provinces (n = 22)
		${\mathbf{C}\mathbf{C}}_{\mathbf{i}}$	Rank	${\mathbf{C}\mathbf{C}}_{\mathbf{i}}$	Rank
AI and CCTV	CCTV system for AI (e.g., high-resolution CCTV systems, Image function improvement)	0.96	1	0.93	1
	AI video analysis for living safety (e.g., detection for littering, drunk people/smoking)	0.73	4	0.88	2
	AI video analysis for crime prevention (e.g., automated tracking, assault detection)	0.79	3	0.84	3
IoT and lighting	IoT automatic situation recognition system (e.g., automatic recognition/warning)	0.55	7	0.62	8
	Intelligent pedestrian environment system (e.g., walking guide lights for pedestrian safety)	0.64	5	0.80	5
	Smart street lights (e.g., scream recognition, emergency bells, voice warnings)	0.45	9	0.68	7
Platform	Crime monitoring system (e.g., crime maps, CCTV analysis, automatic link to 112)	0.82	2	0.83	4
	Mobile crime prevention app (e.g., real-time indoor and outdoor precise locations)	0.61	6	0.62	9
	Real-time control drones/robots (e.g., drones/robots for automatic real-time patrols)	0.42	10	0.29	12
Community	AI guide system for village information (e.g., crime prevention safety zones)	0.16	12	0.44	10
	Improved parking environment system (e.g., shared parking, illegal parking reporting)	0.37	11	0.74	6
	Smart safety community (e.g., Safe return home for women, safety booths)	0.46	8	0.40	11

).

Large cities in the Seoul metropolitan area ranked platform services relatively higher. In terms of priority, the crime monitoring system ranked second (fourth in small and medium-sized cities in other provinces), the mobile crime prevention app ranked sixth (ninth in small and medium-sized cities in other provinces), and real-time control drones/robots ranked tenth (twelfth in small and medium-sized cities in other provinces). The abovementioned findings seem to reflect the needs in the Seoul metropolitan area for more effective and maximized crime prevention and real-time response with advanced technologies based on previously established control infrastructure. Smart safety community ranked eighth, higher than eleventh among small and medium-sized cities in other provinces. The results seem to reflect a greater need for introducing smart community facility elements due to high population density and easing fear about potential crimes by supporting a safe return home in crime-prone areas.

For small and medium-sized cities in other provinces, practical crime prevention services, whose effect could be felt by residents, were ranked higher. AI video analysis for living safety, which detects littering, illegal parking, and drunk people/smoking/juvenile delinquents, ranked second, which was relatively higher (compared to fourth in large cities in the Seoul metropolitan area). In this regard, multifunctional smart street lights with scream recognition, voice warnings, emergency bells, and safety information also ranked seventh (ninth in large cities in the Seoul metropolitan area). The improved parking environment system, which improves the parking environment in residential areas through shared parking and the easy reporting of illegal parking, ranked sixth (eleventh in large cities in the Seoul metropolitan area). Furthermore, the AI-based information board for living safety and convenience information, crime prevention safety zones, and foreign language guide ranked 10th (12th in large cities in the Seoul metropolitan area). The abovementioned results were in part due to the characteristics of small and medium-sized cities in other provinces with a relatively low level of control, where bottom-up services directly experienced by residents were considered more important than government-led services.

4.4. Model Validation

The Fuzzy TOPSIS model does not have its own verification method such as AHP’s consistency verification. Therefore, this study aims to compare the results of the classic AHP analysis most commonly used in MCDM for verification of the Fuzzy TOPSIS model. Like Fuzzy TOPSIS, the AHP technique is also capable of structuring alternative choices in complex decision-making situations based on the experience and knowledge of experts. Therefore, only the survey items were separately constructed and analyzed based on the surveys of the same 48 experts. AHP, developed by Saaty (1986), is a method of evaluating relative importance through pairwise comparison between decision-making hierarchical structure components [62]. In the case of the AHP evaluation criteria, it was hierarchically structured by reorganizing it into three upper items, namely conformity to crime prevention purpose, technology suitability, and technology feasibility, and based on this, the relative importance evaluation and final weight between the upper and lower items were calculated. In the case of alternative evaluation, the AHP technique was also finally calculated by applying the weight of each evaluation criterion to 12 intelligent CPTED service elements like Fuzzy TOPSIS.

Model verification was largely composed of two types: AHP’s consistency verification, which is the same survey basis, and AHP and Fuzzy TOPSIS techniques’ methodological correlation verification. Since AHP’s consistency verification was analyzed based on the same questionnaire, the reliability of Fuzzy TOPSIS’ expert response can be indirectly verified. Verifying the correlation between AHP and Fuzzy TOPSIS techniques can verify the analytical feasibility with AHP, which is used a lot in the research and policy-making process and has public confidence. First, it is the consistency verification of AHP. Unlike the Fuzzy TOPSIS technique, AHP is based on relative evaluation, so consistency verification is possible to verify how consistent the analysis is. For consistency calculation, the consistency ratio (CR) was calculated using the dimension-wise average random index (RI) of the matrix obtained from empirical data. As a result, the CR of the total effective sample (n = 48) was 0.072, showing high consistency. Generally, if the CR value is within 0.1, it is consistent, and if it is within 0.2, it is accepted at an acceptable level [51].

Second, the methodological correlation verification between AHP and Fuzzy TOPSIS. In the case of the evaluation criteria, the correlation coefficient between AHP and Fuzzy TOPSIS was 0.782, indicating a positive correlation. The AHP evaluation criteria analyzed the relative importance based on pairwise comparison, and Fuzzy TOPSIS analyzed the absolute importance of the Likert scale evaluation basis, so it is judged to have an appropriate level of correlation. Looking at the differences between the items, it can be seen that the importance of AHP’s policy conformity is much higher than that of Fuzzy TOPSIS. This is because the importance of crime prevention conformity’, a major category to which policy conformity is dependent, was very high (0.490 major category crime prevention conformity, 0.264 minor category policy conformity). That is, since biased results may be produced due to the dependency problem between the major and minor categories, Fuzzy TOPSIS, which is more intuitive and easier to respond to the questionnaire, is judged as an appropriate methodology (

Table 13. Comparison of evaluation criteria results between AHP and fuzzy TOPSIS.

Evaluation Criteria		AHP		Fuzzy TOPSIS
		Weight	Rank	Weight	Rank
Crime prevention conformity	Crime prevention effectiveness	0.361	1	0.235	1
	Policy conformity	0.129	2	0.090	6
technical suitability	Competitiveness with existing services	0.087	6	0.082	7
	Growth in the service sector	0.129	3	0.143	4
Technology feasibility	Economic viability	0.078	7	0.119	5
	Field applicability	0.106	5	0.162	3
	Sustainability	0.110	4	0.169	2

).

In the case of alternative evaluation, the correlation coefficient between AHP and Fuzzy TOPSIS is 0.976, which has a very strong positive correlation. This is based on the same questionnaire response based on absolute importance, and it can be interpreted that the Fuzzy TOPSIS model is methodologically suitable. One thing to note is that when calculating the optimized ratio for each technology element, the difference between the 1st and 12th Fuzzy TOPSIS normalized values was 10.4%, while AHP was only 4.6%. In other words, the fluctuation range of the normalized value for each alternative of the Fuzzy TOPSIS technique is larger than that of AHP. This is interpreted as being because the TOPSIS technique increased the clarity of more rational choices and decision-making by simultaneously considering the positive and negative alternatives (

Table 14. Comparison of alternatives results between AHP and fuzzy TOPSIS.

Category	Sub-Category	AHP		Fuzzy TOPSIS
		Weight	Rank	Weight	Rank
AI and CCTV	CCTV system for AI	0.107	1	0.133	1
	AI video analysis for living safety	0.094	4	0.109	4
	AI video analysis for crime prevention	0.098	3	0.113	3
IoT and lighting	IoT automatic situation recognition system	0.086	5	0.082	6
	Intelligent pedestrian environment system	0.083	6	0.092	5
	Smart street lights	0.077	8	0.073	8
Platform	Crime monitoring system	0.100	2	0.117	2
	Mobile crime prevention app	0.077	7	0.073	8
	Real-time control drones/robots	0.064	11	0.047	11
Community	AI guide system for village information	0.062	12	0.029	12
	Improved parking environment system	0.077	8	0.077	7
	Smart safety community	0.076	10	0.056	10

, Figure 3).

5. Discussion

This study has identified the intelligent CPTED concept that combines CPTED and crime prevention technologies and set forth the evaluation model to support decision-making in crime prevention. CPTED is a spatial urban design technique intended for crime prevention, while crime prevention technologies work as a tool that can monitor abnormal situations with hardware equipment and software solutions, including CCTV systems and IoT. In this regard, CPTED and crime prevention technologies are making progress as different means of preventing crimes, and it is important to discuss why intelligent CPTED systems combining both of them is necessary. Various factors are at play and lead to a crime in complex ways. The broken windows theory, which is often cited as an example of CPTED, states that a single broken window left unattended leads to a gradual spread of crimes around that location. Orderly conditions must be created and maintained in terms of facilities, environments, and spaces because trivial disorder may lead to a bigger crime, as suggested by the broken windows theory. Although this urban design approach is effective in preventing crime to some extent, its limitation is that it is difficult to respond to a crime after it occurs. One of the advantages associated with using systematic crime prevention technologies, such as CCTV systems, IoT, and platforms, is that they can respond to crimes across the cycle, from CPTED prevention to real-time response to post-crime handling, including victim protection and suspect arrest. Of course, just like the effect of CPTED, crime prevention technologies work as a powerful means of suppressing impulses to commit a crime. The intelligent CPTED service elements with a strong tendency toward CPTED in this study include the intelligent pedestrian environment system and smart safety community. The evaluators in this study were decision-makers in municipal control centers; consequently, they acted more as service providers. In this regard, control elements, such as CCTV systems and the crime monitoring system, ranked higher, while CPTED elements, such as the intelligent pedestrian environment system (5th) and the smart safety community (10th), ranked somewhat lower. Awareness must be raised about the importance and value of the CPTED systems that are closer to residents and support safety in everyday life, and a follow-up study would be needed, considering the characteristics of low-rise residential areas in Korea.

Considering where the respondents lived, this study categorized them into large cities in the Seoul metropolitan area and small and medium-sized cities in other provinces. In Korea, Seoul is a megacity, in which the country’s urbanization is concentrated. Hence, the country is characterized by stark differences in terms of scale and crime prevention infrastructure between large cities in the Seoul metropolitan area and small and medium-sized cities in other provinces. In large cities in the Seoul metropolitan area, platform services, such as the crime monitoring system, the mobile crime prevention app, and real-time control drones/robots, were evaluated relatively higher. This is because there is a huge need for maximized crime prevention and response services, which apply system services for control based on advanced crime prevention infrastructure. In addition, smart safety community services, which ease fear about crimes due to high population density and support safety in everyday life, were also ranked higher than in small and medium-sized cities in other countries. By contrast, the crime prevention services that could be felt by residents and improve crime-prone environments, such as AI video analysis for living safety, smart street lights, and the improved parking environment system, were evaluated higher in small and medium-sized cities in other provinces. In other words, large cities in the Seoul metropolitan need a strategy to maximize scientific crime prevention through an intelligent software approach based on existing crime prevention infrastructure. Meanwhile, small and medium-sized cities in other provinces need policy efforts considering a hardware approach, such as updating crime-prone environments and expanding community safety facilities. While this study analyzed only two categories of large cities in the Seoul metropolitan area and small and medium-sized cities in other provinces due to the nature of MCDM, which conducts analysis based on a survey with a small number of experts, a future study should consider more detailed regional and spatial characteristics.

Intelligence is the key to the technology elements of intelligent CPTED systems examined in this study. It is necessary to maximize the effectiveness and efficiency of service operations combined with advanced crime prevention technologies instead of simply installing physical crime prevention facilities. Even if the same CCTV systems are installed, their effectiveness can be maximized by analyzing crimes in combination with AI technologies, such as detecting assaults and tracking missing people and suspects, rather than simply monitoring CCTV screens. A wide range of 4th Industrial Revolution technologies have been introduced in Korea since the 2000s and are advancing rapidly. Intelligent CPTED service elements must continue to be expanded to keep up with the introduction and development of advanced technologies. Consequently, it is important to not only incorporate new advanced crime prevention technologies but also respond flexibly in terms of urban design, including trends in CPTED applications and the use of new urban spaces. Previous studies focused on crime prevention effects and technology development for individual technologies, but this study was able to secure spatiality and diversity using CPTED. Therefore, this study can actually support decision-making to initially introduce CPTED-based crime prevention services to the site, and based on this, it can contribute to improving public safety in cities. Furthermore, this study considers its systematic nature and scalability by assigning the identification code based on the report identification number. In the future, it would be critical to expand intelligent CPTED systems and update a study in line with new crime prevention technologies and CPTED development trends.

6. Conclusions

To maximize its synergetic effect across the cycle from prevention to response to post-crime management, crime prevention requires a balanced combination of spatial urban design and advanced crime prevention technologies for crime prediction and real-time response. To do so, this study derived intelligent CPTED systems and suggested the evaluation model for turning them into actual services.

The following summary provides an overview of this study’s results: By adding the urban control elements that considered the operations of crime prevention technologies, this study selected architectural urban planning elements under the principles of CPTED and identified 12 intelligent CPTED service elements by combining them with intelligent crime prevention technologies. For the practical use of its findings, this study conducted an expert questionnaire survey based on MCDM with 48 operators in municipal control centers, which introduced public crime prevention services. This study used the fuzzy TOPSIS model, considering the progressiveness of the evaluation method and the ease of evaluation, and not only analyzed responses from all respondents but also compared alternatives by city size, such as large cities in the Seoul metropolitan area and small and medium-sized cities in other provinces. This study drew the following three implications from the analysis of the results from the respondents: First, service elements for service operator-driven CCTV systems and platform control (e.g., CCTV system for AI, crime monitoring system, and AI video analysis for crime prevention) were evaluated highest. Second, crime prevention/safety services mainly targeting individuals (e.g., IoT automatic situation recognition system and mobile crime prevention app) were selected as the next best option. Third, safety community service elements in the form of indirect crime prevention support (e.g., smart safety community and AI guide system for village information) ranked lowest among the respondents. However, such services are not meaningless, even though they were ranked lowest. Notably, the evaluators in this study are decision-makers in municipal control centers, and a follow-up study may need to be conducted from the viewpoint of not only service providers but also service users, including residents and visitors. The comparison by city size showed that large cities in the Seoul metropolitan area evaluated system services for control based on advanced crime prevention infrastructure (e.g., the crime monitoring system and real-time control drones/robots) relatively higher, while small and medium-sized cities in other provinces rated the services that were felt by residents and improved crime-prone environments (e.g., AI video analysis for living safety and smart street lights) relatively higher. These results remind us why it is necessary to have an intelligent CPTED service introduction strategy considering regional and spatial characteristics.

By suggesting the selection method of intelligent CPTED systems, decision-making support for introducing the services in practice, and differences by city scale, this study contributes to the literature and serves as a reference for developing CPTED and services using crime prevention technologies. This study was able to present realistic implications because we analyzed the survey of local government officials, and based on this, it can be used as a policy decision-making method for the initial introduction of intelligent CPTED. However, since it only dealt with the operation of the local government control center that supplies crime prevention services, the opinions of citizens who are service consumers should be reflected as a follow-up study. A future study analyzing not only the service provider side but also the service demand side, including residents, conducting spatial analysis of more detailed regional and spatial characteristics, examining specific service implementation methods, and exploring privacy measures (e.g., privacy infringement), will further improve the effectiveness and practicality of intelligent CPTED systems. Hopefully, it would make a stronger social safety net for crime prevention so that people can lead a safer life.