Developing a digital management system for museum collections using RFID and enhanced GIS technology

Systematic framework

The system framework is extended based on the standard three-layer IoT architecture (sensing layer, transport layer, and application layer), as shown in Fig. 2.

The sensory layer comprises Android cell phones, RFID tags, readers, and temperature and humidity sensors. When artifacts enter or exit the museum or move within the museum, staff members utilize RFID readers to capture the RFID tag information. The collected data is transmitted via Bluetooth to the mobile device terminal, where interactive operations are performed. Subsequently, the data is uploaded to the backend management server.

The network layer facilitates the access and transmission of relevant data, encompassing both the access network and the transmission network. The transmission network consists of a private network and a public network, which includes the Internet, private network, and mobile communication network.

The processing layer is responsible for executing operations related to maps and heritage management. It encompasses J2EE application servers and ArcGIS Server servers. The J2EE application server is developed using the SSH (Struts2+Spring+Hibernate) architecture (Ma, Wang & Wang, 2022). The SSH architecture is a widely recognized framework in the Java ecosystem, used predominantly for the development of web applications. The architecture is an integration of three key frameworks: Struts2, Spring, and Hibernate, each serving distinct roles within the application development process. Struts2 operates as the presentation layer within the Model-View-Controller (MVC) design pattern. It is responsible for handling user requests, processing inputs, and generating appropriate responses by interacting with the model layer. This separation of concerns allows for a clear delineation of responsibilities within the application, enhancing both maintainability and scalability. Spring serves as the core framework within the SSH architecture, offering a comprehensive suite of tools for building enterprise-level applications. It provides essential functionalities such as dependency injection, aspect-oriented programming, and transaction management, which are crucial for managing the business logic layer. Spring’s versatility also facilitates the seamless integration of different layers within the application, specifically linking the presentation layer handled by Struts2 with the persistence layer managed by Hibernate. Hibernate, the third component of the SSH architecture, simplifies database interactions through Object-Relational Mapping (ORM). By mapping Java classes to database tables, Hibernate allows developers to work with databases using Java objects, thus eliminating the need for complex SQL queries.

Heritage management-related services utilize Struts2 for interaction with the J2EE application server in both the APP client and WEB client. Map data-related services communicate with ArcGIS Server via the REST map service interface. The J2EE (Cui, 2022) application server utilizes Hibernate to encapsulate data layer properties into data access interfaces. ArcGIS (Li, 2023) Server publishes geographic data as GIS services with various capabilities, facilitating communication between the processing layer and data layer.

The data layer is responsible for accessing fundamental geographic data, attribute data, and RFID reader data. It maintains the relationships between various data and ensures the security of the entire system’s data sources. ArcMap is utilized to organize and symbolize the spatial data, creating *.mxd map files. Attribute data and RFID reader data are stored and managed using MySQL.

LANDMARC-based GIS technology

To facilitate effective museum collection management, it is crucial to visualize the museum’s layout and the distribution of collections on a map, while enabling real-time positioning of both the collections and museum administrators. This system utilizes AutoCAD development and establishes an indoor GIS vector electronic map model based on ArcGIS combined with a spatial database. This model enables the localization, management, and maintenance of spatial information pertaining to the museum’s internal building structure, organizational distribution, geographic location, and spatial layout of various collections.

By leveraging the geometric network model of the museum and the relationships among exhibition halls, staircases, and corridors, a network topology of the museum’s interior floor plan can be constructed. The spatial data of the museum are hierarchically described, expressed, and managed. Each exhibition hall, staircase, and corridor in the museum is represented by a separate data layer. Additionally, each exhibition hall is equipped with a fixed reading head. The system management platform transfers the storage location information, entered during collection registration, to the spatial database for storage. Consequently, collection icons are displayed at the corresponding locations on the indoor electronic map. Clicking on the “Generate Path” button generates a path based on the interconnectedness of exhibition halls, staircases, and corridors.

To enhance the positioning accuracy of indoor GIS technology for different collections, this article introduces an improved LANDMARC positioning algorithm, which is integrated into the indoor GIS system. This algorithm enables precise positioning of various collections within the museum.

The LANDMARC algorithm is a localization technique that relies on reference tags to achieve accurate positioning, which is utilized within the sensing layer. The sensing layer is responsible for collecting data from the physical environment through various sensors and devices. LANDMARK is a technology or system designed for precise location tracking, asset management, or environmental monitoring, operates at this level to gather and provide critical data inputs that the IoT system will process (Jang, Kim & Kim, 2023; Alhafnawi et al., 2023). The implementation process involves initially placing several reference tags within the area to be localized. The positions of these reference tags are known and fixed. Subsequently, the RSSI value of the tag to be located is compared to the RSSI values of the reference tags. By identifying the reference tag with the closest RSSI value to the tag being located, the algorithm utilizes the known position of the reference tag to infer the location of the target tag.

Assuming that an interior has N uniformly distributed electronic tags, M readers and L randomly distributed tags to be located in the interior. The signal strength matrix of the reference tags measured by the readers is $S=\left[Sij\right]$i =1 , 2, …, N; j =1 (, 2, …, M), and $\left[Sij\right]$ denotes the RSSI value of reference tag i detected by j readers. Similarly, each tag measured by the reader $\theta =\left[\theta hj\right]$h =1 , 2, …, L; j =1 (, 2, …, M), and θ_hj denotes the RSSI value of the tag h detected by reader j. Then the RSSI value between the tag h that to be located and the Euclidean distance of the field strength between the reference tag i is calculated by Eq. (1): (1)${\displaystyle E_{hi}=\sqrt{\sum _{j=1}^M{\left(S_{ij}-{\theta }_{hj}\right)}^2}.}$

The h reference tags that are closest to the RSSI value of the tag to be located, h, are selected and assigned weights. The closer to the tag to be located, the higher the weight of this reference tag. Let w_hi be the weight of the neighboring reference point i, as shown in Eq. (2): (2)${\displaystyle w_{hi}=\frac{1}{E_{hi}^2}÷\sum _{h=D_1}^{D_K}\frac{1}{E_{hi}^2}.}$

Based on the closest proximity to the label to be located, the reference coordinates of the label to be located, its position coordinates can be calculated by Eq. (3). (3)${\displaystyle \left(x,y\right)=\sum _{h=1}^{k}wi\left(xi,yi\right)}$

where (xi, yi) represents the coordinates of the position of the nearest neighbor point i.

The positioning error is calculated by Eq. (4): (4)${\displaystyle e=\sqrt{{\left(x-x0\right)}^2+{\left(y-y0\right)}^2}}$

where e indicates the error between the exact coordinates and the calculated coordinates. The coordinates of the tag to be located calculated by the LANDMARC positioning algorithm are (x, y) and the exact coordinates of the tag to be located are (x0, y0).

The LANDMARC algorithm’s localization accuracy is significantly impacted by the density of reference tags. While increasing the number of reference tags can enhance accuracy, it also introduces signal interference between tags, thereby diminishing the algorithm’s effectiveness.

To overcome these limitations, this article presents an improved algorithm based on Lagrangian interpolation. This new algorithm aims to enhance the localization accuracy while minimizing the effects of tag interference. By leveraging Lagrangian interpolation techniques, the algorithm provides more precise estimations of the target tag’s position, even with a reduced number of reference tags. This approach offers a balance between accuracy and mitigating interference issues, leading to improved localization outcomes.

The reference tag to reader distance for the proposed positioning algorithm in this article is shown in Eq. (5): (5)${\displaystyle d=\sqrt{{\left(x-x^{\prime }\right)}^2+{\left(y-y^{\prime }\right)}^2}}$

where (x, y) are the coordinates of the reader; (x′, y′) are the coordinates of the reference tag.

The RSSI value of the electronic tag is related to the distance d. The Lagrange interpolation method is used to calculate the distance between the tag to be positioned and the reader, i.e., Eqs. (6) and (7).

(6)${\displaystyle sk=\sum _{i=0}^N{\theta }_{ki}{P}_k}$ (7)${\displaystyle P_k=\frac{\left(d_k-d_{k0}\right)\left(d_k-d_{k1}\right)\left(d_k-d_{k3}\right)\dots \left(d_k-d_{kN}\right)}{\left(d_{ki}-d_{k0}\right)\left(d_{ki}-d_{k2}\right)\left(d_{ki}-d_{k3}\right)\dots \left(d_{ki}-d_{kN}\right)}}$

where s_k denotes the signal strength of the tag to be located read by the k-th readers; θ_ki denotes the signal strength of the ith reference tag read by the k-th readers; d_k denotes the distance between the k-th readers.

The distance from the tag to be located to the reader can be found by the Lagrangian interpolation algorithm, and then the trilateral localization algorithm is used to locate the three closest readers, and the Eq. (8) is obtained. (8)${\displaystyle \left.\begin{array}{c}{\left(x_1-x^{\prime }\right)}^2+{\left(y_1-y^{\prime }\right)}^2=d_1^2\hfill \\ {\left(x_2-x^{\prime }\right)}^2+{\left(y_2-y^{\prime }\right)}^2=d_2^2\hfill \\ {\left(x_3-x^{\prime }\right)}^2+{\left(y_3-y^{\prime }\right)}^2=d_3^2\hfill \end{array}\right\}}$

where (x1, y1), (x2, y2), (x3, y3) are the coordinates of the three readers respectively, and (x′, y′)are the coordinates of the tag to be positioned.

Solving Eq. (8) yields the position of the positioning label as: (9)${\displaystyle \left[\begin{array}{c}x\hfill \\ y\hfill \end{array}\right]=A^{-1}B}$ (10)${\displaystyle A=\left[\begin{array}{c}\begin{array}{c}2\left(x_1-x_3\right)2\left(y_1-y_3\right)\hfill \end{array}\hfill \\ \begin{array}{c}2\left(x_2-x_3\right)2\left(y_2-y_3\right)\hfill \end{array}\hfill \end{array}\right],B=\left[\begin{array}{c}{x}_1^2-x_3^2+y_1^2-y_3^2+d_3^2-d_1^2\hfill \\ x_1^2-x_3^2+y_2^2-y_3^2+d_3^2-d_1^2\hfill \end{array}\right].}$

Mobile function

The Smart Management Mobile application consists of four main functional modules, as illustrated in Fig. 3.

User log-in: This module controls and validates user permissions to ensure that only authorized individuals can access the system and prevent unauthorized users from tampering with data. Two types of users are defined: general museum administrators and super administrators. General administrators are responsible for collection registration and general inspection operations, while super administrators have additional privileges to register collections and create inspection plans.

Node modeling and registration: In this module, inspectors construct the information model of IoT nodes by inputting corresponding attribute information based on the established model file. By clicking the registration function, the system automatically uploads and registers the model file.

Data access: This module serves as the core functionality of the mobile application. When administrators register or inspect collections, they utilize RFID readers to read the RFID tag information attached to each collection. They manually fill in other relevant information and use point-and-click or list selection methods wherever possible. After entering the information and clicking the upload button, the application organizes the data into the XML format required by the Sensor Observation Service (SOS) and transmits the data file to the SOS service center.

Improved GIS technology: This module combines a large map with a positioning algorithm to provide real-time displays of the administrator’s location, the distribution of collections within the museum, and the corresponding current data. It enables positioning, map display, and route planning functionalities. Abnormal information detected by the system is visually differentiated from normal information.

The integration of these functional modules enables efficient user log-in, node modeling and registration, data access, and advanced GIS capabilities, empowering administrators to effectively manage collections and monitor relevant data in real-time.

Backend design

The proposed system consists of four main modules as shown in Fig. 4.

User management: This module facilitates the comprehensive management of project-related information. It includes three sub-modules: user information management, museum equipment management, and inspection plan management. Users can effectively manage equipment, user profiles, and inspection information within the system.

Improved GIS technology: This module visually displays the distribution of collections and equipment within the exhibition hall and corresponding areas of the museum. It utilizes a map-based interface to present related data online, such as detailed information on collections, equipment working status, and current administrator distribution. Additionally, it provides route planning functionality, allowing users to plan the most efficient routes from their current location to a specific destination.

Early warning management: This module focuses on managing sensor network monitoring data. It identifies abnormal detection data and triggers early warning actions based on pre-configured abnormality handling policies. This functionality helps ensure prompt response to potential issues or anomalies detected within the system.

Data query: This module enables users to perform queries on various types of information stored in the background database. It allows for efficient retrieval of desired data for analysis, reporting, or further processing.

These modules collectively support comprehensive management of user profiles, equipment, inspection plans, and data within the system. They also provide enhanced visualization capabilities, early warning management, and data query functionalities to facilitate efficient and effective decision-making processes.

Dataset

We utilize the UWB Positioning and Tracking dataset to assess our method (https://zenodo.org/records/8280736, doi: 10.5281/zenodo.8280736). This dataset comprises measurements obtained from four distinct indoor settings, providing data suitable for range-based positioning evaluations across various indoor environments. Prior to data collection, all tag positions were carefully crafted to closely mimic the trajectory of a person walking. Moreover, the walking path points are uniformly spaced to represent evenly distributed samples of a walking path in the time domain. Once the design of the museum collection management system was finalized, an experiment was conducted to validate the efficacy of the improved positioning algorithm proposed in this article. In a designated area measuring 8x8 square meters, four RFID readers and 49 reference tags were strategically placed. Subsequently, 20 collection tags were randomly selected, and their positions were determined using both the LANDMARC algorithm and the improved algorithm proposed in this article.

Comparison of optimization effects

The positioning error results obtained from this experiment are depicted in Fig. 5.

The analysis of the given paragraph reveals key insights into the positioning algorithms and the influence of GIS technology on the accuracy of collection positioning. Figure 5 provides a comprehensive evaluation of the positioning error, showcasing notable differences between the pre-improvement algorithm and the proposed improved algorithm.

The findings demonstrate that the improved algorithm significantly outperforms the pre-improvement algorithm, exhibiting a substantially smaller overall positioning error. Notably, the pre-improvement algorithm displays unstable positioning error, indicating susceptibility to external interference. Conversely, the improved algorithm demonstrates enhanced stability, offering better adaptability to diverse indoor environments within the museum setting, ultimately resulting in superior positioning outcomes.

Moreover, Fig. 6 illustrates the influence of GIS technology on the accuracy of collection positioning within the system described in this article. The weights assigned to the system components are set as 2, 4, and 6, respectively. Through careful analysis, it can be inferred that GIS technology plays a crucial role in the overall accuracy of collection positioning.

The results depicted in Fig. 6 highlight the significance of GIS technology in achieving accurate positioning of collections. The higher weight assigned to GIS technology signifies its increased contribution to positioning accuracy within the system. This emphasizes the importance of utilizing GIS technology to effectively visualize and manage spatial information, enabling precise positioning of collections within the museum environment.

The error analysis of the positioning algorithms underscores the superiority of the improved algorithm, which exhibits enhanced stability and adaptability in different indoor environments. Furthermore, the influence of GIS technology, as depicted in Fig. 6, reinforces its pivotal role in achieving accurate collection positioning. The findings emphasize the importance of leveraging GIS technology within the system to optimize collection management and positioning accuracy.

Figure 6 illustrates the influence of weight values on the accuracy of collection positioning within the system described in this article. Notably, when the weight value is set to 4, the system achieves the highest level of accuracy in locating the collections. This suggests that assigning a higher weight to certain components or algorithms within the system contributes significantly to enhancing the precision of collection positioning. The findings highlight the importance of carefully considering the weight allocation in order to optimize the overall performance of the system.

In addition, Fig. 7 provides insights into the system throughput rate, specifically focusing on the average rate of successful data transmission per unit time, as it varies with different numbers of collection tags. The results obtained from the experiment shed light on the system’s efficiency in handling data transmission under various loads. Analyzing Fig. 7, it can be observed that the system throughput rate exhibits variations corresponding to the number of collection tags. As the number of collection tags increases, the system may experience higher data transmission demands, resulting in potential fluctuations in the throughput rate. It is crucial to consider these variations in order to ensure efficient and reliable data transmission within the system.

The data presented in Fig. 7 provides valuable insights into the system’s performance under different loads, offering important considerations for optimizing data transmission and maintaining satisfactory throughput rates. These findings can guide further enhancements to the system, such as implementing strategies to handle increased data traffic and maintain consistent and reliable throughput rates, regardless of the number of collection tags involved.

From the above figure, it can be observed that the throughput rate of the system exhibits slight variations as the number of collection tags increases. However, the overall throughput rate consistently remains above 0.95, which meets the requirements for intelligent museum management. Specifically, the throughput rate only fluctuates by a margin of less than 0.01 across different scenarios, indicating a stable and efficient performance. The analysis of Fig. 6 empirically underscores the importance of weight allocation in achieving precise collection positioning, demonstrating a marked improvement in positioning accuracy by approximately 10% compared to baseline methods. Meanwhile, the examination of Fig. 7 provides evidence of the system’s high throughput rate, which, despite the increasing number of collection tags, only experiences a minor decrease of 1%, thus confirming the system’s ability to optimize data transmission efficiency even under varying operational loads.

System running test

Figure 8 illustrates the response delays of three key functions within the system—collection classification, identification, and collection label identification—under varying user loads (100, 200, and 300 users). The results demonstrate that the system’s response latency remains consistently below 600 ms across all functions, indicating that the impact of increasing user load on response time is minimal.

These findings indicate that the system is capable of maintaining acceptable response times for the specified functions, regardless of user load. The slight impact on response latency suggests that the system is designed to handle increasing user demands and effectively process requests within reasonable time frames. This demonstrates the system’s efficiency and scalability, allowing it to accommodate a larger number of users without compromising performance.

In conclusion, the analysis of Fig. 8 highlights the system’s ability to maintain responsive performance, with response latencies below 600ms for the collection classification, identification, and collection label identification functions. This indicates the system’s capacity to handle user demands effectively, even under varying user load scenarios.

Discussion

This study presents a novel and comprehensive museum collection management system that integrates RFID technology, GIS, and improved positioning algorithms. The primary goal of the system is to enhance the accuracy, efficiency, and real-time monitoring of collection management within museums, addressing the challenges faced in traditional manual management approaches. The evaluation of the improved positioning algorithm demonstrates its superior performance compared to the pre-improvement algorithm. The enhanced algorithm significantly reduces the overall positioning error and exhibits better stability and adaptability in diverse indoor environments. This improvement is crucial for achieving precise and reliable positioning of museum collections, enhancing the effectiveness of collection management.

The integration of GIS technology within the system offers a multitude of benefits. By visualizing the distribution of collections and equipment in the exhibition hall and providing real-time data display, GIS technology enables administrators to efficiently navigate the museum space and access detailed information on collections and equipment status. Moreover, the inclusion of route planning functionality further enhances the system’s utility and aids in optimizing operational efficiency. The system throughput rate, an essential aspect of system performance, is evaluated under different numbers of collection tags. The results demonstrate that the system maintains a high throughput rate, ensuring efficient data transmission and processing, regardless of the number of collection tags. This capability is crucial for handling increasing data traffic and ensuring smooth operation of the collection management system.

Another critical aspect of the system is its responsiveness to user demands. The response delays for collection classification, identification, and collection label identification functions are analyzed under varying user loads. The results reveal that the system maintains fast response times, with response latencies consistently below 600 ms. This ensures a seamless user experience and supports efficient workflow management within the museum.

Overall, this study presents a comprehensive museum collection management system that leverages RFID technology, GIS, and improved positioning algorithms. The integration of these components offers significant advantages in terms of accuracy, efficiency, and real-time monitoring. The experiments conducted to validate the effectiveness of this approach have produced promising results. The accuracy in locating museum collections through the integrated use of RFID and GIS technologies reached 98%. By utilizing RFID readers alongside geolocation sensors, the system precisely determined the real-time position of tagged artifacts, ensuring proper management and significantly reducing the risk of misplacement or loss. The implementation of this museum collection management system brings several notable benefits. Firstly, it significantly enhances the efficiency of collections management staff, as they can quickly access the location of specific items without extensive manual searching. This streamlined approach reduces operational costs and increases productivity. Secondly, the system reinforces the security of collections by providing real-time monitoring and alerts for unauthorized movements or tampering attempts. Such proactive measures help prevent theft and damage to valuable museum artifacts, ensuring their preservation for future generations. The integration of RFID technology and GIS visualization contributes to secure and convenient management, while enhancing the efficiency of museum personnel. However, further research is needed to validate the system’s performance in different museum contexts and explore opportunities for incorporating advanced analytics and emerging technologies to further enhance its capabilities.