Skip to content
BY 4.0 license Open Access Published by De Gruyter Open Access March 28, 2023

RFID supply chain data deconstruction method based on artificial intelligence technology

  • Huiying Zhang EMAIL logo and Ze Li
From the journal Open Computer Science


Radio frequency identification (RFID) is a broad rapidly evolving skill in the past few years. It is characterized by non-contact identification, fast read and write speed, small label size, large data storage capacity, and other technical advantages. RFID technology for goods movement has completely changed the traditional supply chain management, greatly improved the operational efficiency of enterprises, and has become an important method for the development of supply chain logistics. This work mainly studies and analyzes the RFID supply chain, introduces the development and application of RFID supply chain sector technology, and discusses the operation of the supply chain in detail. Then, according to the existing RFID supply chain, a RFID supply chain artificial intelligence (AI) based approach to technology is proposed, and the data analysis of RFID supply chain is introduced in detail. In this work, through the research experiment of AI technology RFID supply chain data analysis, the experimental data show that there are several time-consuming links in the supply chain system. The time consumed in the AI RFID system is 9.9, 3.4, 3.5, and 29.9 min, respectively, while each link in the original system takes 13.4, 4.9, 4.9, and 34.9 min. It can be seen from the above data that the amount of time in each system link of the AI RFID supply chain system is less than that of the original supply chain system, which shortens the entire product passing cycle and greatly improves work efficiency.

1 Introduction

Global economic integration and demand diversification have accelerated changes in the competitive environment of companies. Companies are no longer able to meet consumer demands efficiently and quickly. Upstream and downstream enterprises have become an effective form of organizing competition. Supply chain organizations have higher requirements for process optimization and rapid response, especially for perishable products. Due to their relatively short life cycle and sensitivity to time, companies involved in the supply chain must collectively navigate the changing competitive environment. In order to improve service levels, meet customer needs, and improve supply chain efficiency, it is necessary to introduce advanced information technology and inventory management technology, strengthen information exchanges between enterprises, and improve the ability of enterprises to respond to market conditions. The use of Radio frequency identification (RFID) is being actively pushed by numerous players skill in the service industry, but while RFID brings benefits to the supply chain, it also increases transaction costs in the supply chain. Since RFID tags can be reused, suppliers need to coordinate costs and profits for a win–win situation.

Along with the rapid economic standardization and evolution of computer systems, competition among enterprises is gradually shifting from enterprises to supply chains. The level and efficiency of the supply chain is an important element in deciding the future survival of a corporation, which provides a broad development space for the supply chain field [1]. Over the last century, new problems have arisen in this field, the most important of which is the real-time synchronization of supply chain information. On the one hand, food supply information in the supply chain must be delivered in a prompt and exact manner. However, due to delays in data collection methods (manual data entry or barcode scanning), inefficiencies in data entry, unavoidable human errors, and delays in data collection, access to accurate and timely information is often hindered. On the other hand, the physical movement information in the supply chain must be shared among enterprises (including retailers, wholesalers, distribution centers, manufacturers, etc.). In fact, although there are flexible information exchange methods between upstream and downstream enterprises, they cannot cooperate with other enterprises in the supply chain, nor can they eliminate the drawbacks of side effects in principle. The reason for doing this is because traditional methods of handling messages are not favorable for sharing messages in a timely manner, so the physical flow of the supply chain cannot be fully grasped by all companies. The effect of seamless and accurate transmission of supply chain information on supply chain performance is an important topic in supply chain research.

This study uses artificial intelligence (AI) technology to conduct an analytical study on RFID supply chain data. Its data show that the efficiency of identifying cargo units and processing goods in the supply chain system is still the efficiency of the AI RFID system in processing business volume. The processing efficiency of the RFID system is 12% higher than that of the original system. The recognition rate of the RFID system is 100%, while the original system is worse, only 96.9%. There may be accidental factors in the actual operation process, but in general, the recognition rate of electronic tags is quite high. It can be seen from the above analysis that the AI technology research experiment is of major relevance to the growth of the current RFID supply chain data analysis.

2 Related work

This work studies some technologies of RFID supply chain, which can be fully applied to the research in this field. Gautam et al. analyzed the impact of RFID tags on traceability by taking the kiwifruit supply chain as an example [2]. Through Pichoff’s research and practical application, it is clear that the inventory management process requires item-level RFID to improve efficiency, otherwise retailers and brands will face the risk of future obsolescence [3]. Herzl mainly studied RFID utilization, by using electromagnetic fields, the technology enables producers to identify and track foods such as meat, fruit, and dairy products in real time through labels attached to packaging [4]. Biswal et al. studied the effects of RFID implementation in the supply network of not-for-profit industries, looking at the effect of order availability as well as drawdown recall on total costs at depot floor level [5]. Podduturi et al. incorporated a peak-to-peer (P2P) based company system for positioning single item in networks of supply chain managers, allowing extensible targeting of large scale indexes of items in larger spread-out workable forests [6]. Hoek aimed to explore how an RFID implementation framework can inform blockchain considerations in supply chains [7]. Those approaches have given rise to some references for studies, but did not gain any public acceptance due to the short period of time and tiny specimen size of the related research.

Based on AI technology, the following relevant materials were reviewed to optimize the RFID supply chain research. Hassabis et al. surveyed the historical interaction among the areas of AI and neuroscience, and emphasized that the progress made recently in AI is inspired by research on human and otherwise zoological neural counts [8]. Li et al. further introduced the basic concepts of AI and addressed the connection of AI to alternative candidate services in 5G mobile systems [9]. Liu et al. discussed the advantages, limitations, practical implications, and some new research trends of different AI algorithms [10]. Thrall et al. argued that AI surveillance routines can help prioritize work orders for doctors in charge of imaging and recognize doubtful or negative reports for possible early scrutiny [11]. Caviglione et al. designed to find software concealed in switched data using two AI tools such as neural web and policy tree based approaches for inspection [12]. These methods provide sufficient literature basis for people to study AI technology for RFID supply chain data analysis.

3 Overview of AI-enabled RFID supply chain

At present, the practical application of RFID technology in enterprises is not widespread, and the application of RFID technology in the entire supply chain is also less. Compared with barcode technology, RFID technology has a higher cost, which makes many mini and mid-sized businesses discouraged. Therefore, this work studies the RFID supply chain through AI technology to help enterprises obtain higher benefits.

3.1 Overview of AI

With the development of computer graphics technology, the amount of calculations that computers can complete far exceeds people’s imagination, and the artificial intelligence technology that emerged from this has also developed rapidly in recent years [13]. The research on AI is rich in tradition. As early as the 1950s, people have begun to explore the field of AI. The research in this period mainly includes the learning of chess game strategy, and the learning of logic and language. However, limited by the technical level at that time, especially computer software and hardware technology, AI developed slowly in the next period of time, and the field of AI was also ushered into a cold window period. After the 1980s, AI was successfully applied to commercial systems, and AI gradually attracted people’s attention. Since then, AI has been used in logistics, data mining, medicine, and other fields. At the same time, with the rapid development of computer technology, the hardware conditions of AI are getting better and better. Many problems that were previously constrained by equipment are being solved, and AI has once again been developed by leaps and bounds. Today, AI is used in a wide range of applications [14].

AI is an emerging field that studies how to use computers to simulate the human brain's ability to think, map, recognize, understand, plan, reason and solve problems. Some professors define AI as “a branch of computer science that involves the intelligence of computers and represents the future direction of computer technology.”

From an engineer’s perspective, AI is using human exploitation to give computers the capabilities related to human ingenuity. It is the response of a machine to human mind and wisdom. From a geographical point of reference, AI is the science of developing intelligent machines or machines to simulate, refine, and supplement human intelligence. From the perspective of disciplinary status and development level, AI is the forefront of modern science and technology. It is developed on the basis of computer science, information processing, control processing, psychology, mathematics, biology, philosophy, and other fields [15]. It is an ever-evolving discipline with new ideas, theories, and techniques, very broad.

AI is a simulation of the human thought process. While it is not a human brain, it is capable of thinking like a human, maybe even more. AI is a complex science that requires knowledge of computers, psychology, and even philosophy to accomplish its tasks. In general, AI skills mean the smart conduct of an artificial human being, typically concerned with sensing, learning, reasoning, exchanging, and behavior in relation to complex situations. It can also be said that the purpose of AI is to enable machines to perform complex tasks that can only be done by humans [16].

There are many different types of AI algorithms, including deep learning, machine learning, artificial neural networks, natural language processing, statistical learning, etc., but the basic idea is to perceive the environment and then make decisions that maximize the target outcome. Many AI can train and learn algorism to improve system capabilities from data. Popular types of theories include Markov processes, Bayesian networks, and decision trees. Through training, study, and remembering, AI algorithms can eventually mimic humans and make decisions about problems based on human needs. Like humans, they gain experience by constantly learning and memorizing the external environment in order to respond well to different situations.

The Markov (MCL) algorithm is a simple, fast, and easily scalable model-based algorithm. It is an algorithm proposed in 2000 and used to cluster figure-structured data, which has played an important role in the field of bioinformatics. As the algorithm is refined, its applications are not limited to biology, but have been developed in different fields such as social networks and transportation networks.

The core of the MCL algorithm is the process of iterative random walks in complex networks. Random walks are essentially the typical performance of Markov chains. Its expression is as follows:

(1) P [ M t + 1 = m t + 1 | M 0 ] = P [ M t + 1 = m t + 1 | M 0 = m 0 M 1 = m 1 M t = m t ] ,

(2) P [ M t + 1 = m t + 1 | M 0 ] = P [ M t + 1 = m t + 1 | M t = m t ] .

Then, the Markov chain is said to be homogeneous, that is, the probability at time t + 1 is only related to the state at time t, and is independent of the parameter t.

The MCL algorithm defines three matrix operations, namely, expansion, dilation and pruning. By iterating the above three operations on the random matrix, the figure-structured dataset is divided into relatively dense multiple subgraphs, and the final random matrix is converged. The algorithm implementation process is as follows:

Expansion process: It is the process of matrix multiplication, which is used to simulate the expansion of random walks, so that the random matrix is homogeneous [17].

(3) X ( i , j ) = E ( i , j ) + I ( i , j ) , k = 1 y E ( k , j ) .

Extended operation: matrix multiplication, that is, multiplying two matrices, which is expressed as follows:

(4) E Exp = Z k , E ( v i , v j ) = X k .

Dilation process: It is a process of enhancing the intra-class flow probability, weakening the inter-class flow probability, and shrinking the random walk.

Dilation operation: matrix dot multiplication and normalization, that is, multiply the corresponding position elements of two matrices and perform matrix column normalization, which is expressed as follows:

(5) T r , X ( v i , v j ) = ( X ( i , j ) ) r k = 1 y ( ( X ( k , j ) ) r ) r = 2 ,

(6) E In f = T r , X .

Pruning process: It is the process of pruning the connection of community nodes. Precise trim:

(7) E Pru = X .

Threshold trimming:

(8) E Pru ( i , j ) = X ( i , j ) , X ( i , j ) > d , 0 , X ( i , j ) d .

Iterate the above operations until

(9) E Pru ( i , j ) = ( E Pru ( i , j ) ) 2 .

Conditional probability: Assuming M and N are two random events, P(N) > 0, the conditional probability of event M occurring when event N exists is defined as follows:

(10) P ( M | N ) = P ( M N ) P ( N ) .

From the above formula, it can be seen that

(11) P ( M N ) = P ( M | N ) P ( N ) .

This is called the multiplicative law of probability and can also be written as follows:

(12) P ( M N ) = P ( M ) P ( M | N ) .

The basic formula of Bayesian theory is as follows:

(13) p ( C | D ) = p ( C , D ) p ( D ) ,

(14) p ( C | D ) = p ( D | C ) p ( C ) p ( D ) .

Among them, p(C|D) is the posterior probability of event C, p(D|C) is the likelihood function or conditional probability, p(C) is the prior probability of event C, p(D) is called the marginal probability of event D, because the independent variable of p(C1D) is C independent of p(D). Therefore, it can be written as an expression in the form of the posterior distribution kernel as follows:

(15) p ( C | D ) = p ( D | C ) p ( C ) p ( D ) p ( D | C ) p ( C ) .

The likelihood function is as follows:

(16) p ( m | θ ) = i = 1 y p ( m i | θ ) .

Through the above formula, the posterior probability density function of the random variable θ is as follows:

(17) p ( θ | m ) = p ( m | θ ) π ( θ ) p ( m ) p ( m | θ ) π ( θ ) .

The whole process of inference is realized according to Bayesian theory.

Artificial neural network is an important core technology in the field of AI. It was devised by scientists according to the principles of how biologic neural networks work. A high-performance computing model formed by abstraction has unique advantages in the field of processing big data, and is increasingly used in important fields such as medicine, imaging, aviation, and military, and has achieved remarkable results. Artificial neural networks were originally designed to teach computers to solve problems like the human brain, but over time they have turned more and more towards solving specific problems. The artificial neural network can complete the learning task according to the sample data, and the system has no prior concept of the learning content before learning. Through learning, the system can grasp the essential core of the learning content, and use the learned content to solve the corresponding problems, just like people have mastered some knowledge through learning [18]. A schematic figure of a single-layer neural network is shown in Figure 1.

Figure 1 
                  Schematic figure of a single-layer neural network.
Figure 1

Schematic figure of a single-layer neural network.

The input set will input each node, and the link between the input and the node has a corresponding weight. In addition, the node will also have a paranoid function, which generates the corresponding output through the weight and paranoid function. By observing the structure figure of a single-layer neural network, we can know that a single-layer neural network cannot cope with complex problems calmly. Therefore, people have proposed a multilayer neural network. The schematic figure of the multilayer neural network is shown in Figure 2.

Figure 2 
                  Schematic figure of a multilayer neural network.
Figure 2

Schematic figure of a multilayer neural network.

A multilayer neural network consists of an input layer, several hidden layers, and an output layer. The purpose of the hidden layer is to train the system parameters and solve the current problem. There is a fully connected relationship between hidden layers. The number of hidden layers is related to the performance of the system. The more the layers, the more accurate the parameters obtained by training. However, if the number of hidden layers is increased due to the excessive pursuit of accuracy, the training time of the system will be greatly increased. At the same time, when the number of hidden layers reaches a certain level, the accuracy increases very slowly with the increase in the number of layers. Therefore, when using a neural network to solve problems, it is necessary to reasonably select the number of hidden layers, and to make a compromise between accuracy and time cost. While ensuring accuracy, it is also necessary to save the time required for training as much as possible [19].

The core of AI technology is to use machines (mainly computers) to simulate and realize human intelligent behavior. After decades of development, AI has developed into many fields and has been widely used in daily life and learning. The main application areas are: IntelliSense, including pattern recognition and natural language understanding. Intelligent thinking, including problem solving, reasoning and theorem proving, expert systems, and automatic programming. Intelligent learning, in which learning ability is undoubtedly one of the most important aspects in AI research. Intelligent action is the broadest and most important field of AI. These include intelligent guidance, intelligent search, intelligent planning and control, robotics, distributed AI and agents, data mining and artificial life, machine learning, and knowledge discovery.

3.2 Overview of RFID supply chain

Under the background of social development, information network construction and the production has become one of the most important criteria for the measurement of overall power. An automatic recognition is an integrated piece of technical technology founded on radio and television systems. It is an important program and tool for computer to automatically read and record information, and it plays an important role in the construction of modern supply management system. At present, automatic identification methods such as optical character recognition, machine identification, barcode, RFID, magnetic card, biometric identification technology, and magnetic color identification have been widely used. Common automatic identification technologies are shown in Figure 3.

Figure 3 
                  Common automatic identification techniques.
Figure 3

Common automatic identification techniques.

Among them, the RFID method is a technology that uses high-frequency signals to transmit information through spatial coupling (alternating magnetic or electromagnetic fields) without contact, and realizes identification through the transmitted information. RFID technology can provide non-contact static or moving automatic distance identification. It has the advantages of strong anti-interference ability, high performance, long service life, small size, high precision, and fast running speed. In particular, if the object is only identified in the field, other identification methods are not comparable [20].

RFID technology provides a non-contact method to automatically identify a tagged person or thing. RFID tags have similar functions to traditional barcodes, but RFID tags have their own advantages. RFID tags are unique in the world, avoiding the problem of repeated identification. The communication between the reader and the tag is carried out by radio frequency, which does not need to be like a barcode. The reader is facing the tag to read and write. As long as the RFID tag is within the range of the reader, it can be read. The manual intervention is reduced and the efficiency is improved. The reader can read a large number of tags, which greatly facilitates supply chain management and can effectively alleviate the phenomenon of queuing. Due to all these advantages, RFID technology has gradually entered our daily life and industrial production. At the same time, RFID technology also brings its own security and privacy concerns.

In a narrow sense, RFID systems include reading and writing systems, electronic tags, and backend servers. The backend server communicates with the read-write system through a secure channel. The composition of the RFID system is shown in Figure 4.

Figure 4 
                  RFID system composition.
Figure 4

RFID system composition.

Among them, the tag stores data related to the object and usually consists of a high-frequency chip and an antenna with a globally unique identifier. Readers are responsible for reading and writing information and are responsible for computing and storage capacity. The main function of the backend server is to store drives, tags, and other information, and to provide authentication services between tags and drives. In some application scenarios, the backend server is often ignored, while other articles treat the backend server as a database.

RFID technology is not a new technology, it was first used in military fields, such as radar systems, and the identification system of friend and foe in World War II. RFID technology inherits and develops the concept of radar, forming a new technology that can automatically identify targets [21].

RFID technology is ideal for managing items, and its importance goes beyond replacing barcodes. More importantly, through the automation of data collection and the resulting automated business processing model, the use of RFID technology has changed the traditional supply chain management process, greatly reducing errors, simplifying business processes, and improving the efficiency of supply chain systems and the quality of customer service. RFID technology will also change the supply chain information system, so that information can not only be utilized within a single enterprise, but also shared across the entire supply chain system and a larger scope. However, RFID technology also has limitations. For example, the current high cost of electronic tags limits the application of RFID technology in certain fields. Due to the influence of object materials (such as metals, liquid media) and the influence of electromagnetic interference in the application environment, RFID technology is subject to many limitations in the application of object detection. There is a stagnation point in the working area of the RFID antenna, which affects the reading rate of the tag and so on. With the development of RFID technology, these current limitations will eventually be solved one by one, bringing a new situation and global scale of RFID technology application.

The concept of supply chain was first developed in the 1980s and became the focus of attention in the late 1990s. Some scholars believe that the supply chain is the strategy of integration, coordination and interaction among the functional departments of the alliance. Other scholars believe that the supply chain is a functional network structure centered on key enterprises. Controlling the flow of information, logistics and capital starting with raw material procurement, and finally, intermediate and final products are delivered to consumers through distribution networks. In this structure, suppliers, retailers, manufacturers, distributors, and end customers are connected as a whole.

From the perspective of theoretical development, the International Supply Chain Association defines supply chain management as the response and coverage to market demand, the entire process of an enterprise’s demand for raw materials, services, and information. The American Association of Logistics believes that the supply chain covers all planning and logistics activities, including procurement, outsourcing, transition, etc. Some scholars believe that the supply chain is a comprehensive method, and in the 21st century, more and more enterprises begin to apply the concept of value chain [22].

At present, many enterprises are involved in the supply chain, because the main advantage of the supply chain is to reduce inventory and manage the supply chain effectively, thereby reducing the duplication of labor among members, and eliminating redundant links in the supply chain for cost-effective and efficient management. It can provide services for decision makers, analyze the uncertainty of supply chain, determine inventory, formulate purchasing policy, and optimize investment evaluation plan to choose the best plan. It can increase overall value and improve business relationships by coordinating and evaluating the impact of different strategies on supply chain inventory and service policies. In the context of manufacturing, supply chain firms change their rivalry to become inter-supply chain rivalry. The focus is on the strategic partnership between the core enterprise and its upstream and downstream partners, and each company uses its own advantages to achieve a win–win situation. In order to improve service quality, internal and external coordination and cooperation can stimulate consumer demand in the supply chain, greatly shorten the product life cycle, and provide consumers with marketable products in a timely manner. In order to ensure the effective integration of supply and demand in the supply chain, manufacturers and suppliers must be closely linked, coordinated, and optimized.

Collectively, supply chains have transformed businesses and their branches into an integrated network. This accelerates the process of making the change from manufacturing to spending, reducing throughput and marketing cycles and allowing businesses to react quickly to evolving product needs, and greatly improves the business competitiveness of enterprises in the supply chain [23].

Since 1985, RFID technology has been widely used in the commercial field. Since RFID tags can be read and written, they are especially suitable for situations where frequent changes to data content are required. Its role is to collect and transmit data to execute system instructions. It is commonly used in supply chain, such as warehouse management, transportation management, production management, product tracking, vehicle and shelf identification, and goods theft prevention in stores, especially supermarkets, etc. The application of RFID in the supply chain is shown in Figure 5.

Figure 5 
                  RFID in the supply chain.
Figure 5

RFID in the supply chain.

As can be seen from Figure 5, the introduction of RFID technology is directly transferred from the current supply chain to the source of the supply chain. Manufacturers attach electronic tags to goods when they arrive at the factory, and companies involved in the subsequent supply chain only use RFID to read and write. Reading electronic product labels and uploading these data to an information server can accurately capture product movement information at different stages of the supply chain. In this way, the amount of information collected at each link of the supply chain will be greatly increased. It is estimated that the amount of information collected through RFID is approximately 30 times that of the barcode identification methods currently used. Intensive information collection will inevitably improve the timeliness and reliability of information, thereby increasing the sensitivity of supply chain response measures.

4 RFID supply chain from different perspectives

4.1 RFID supply chain of AI technology

This section considers producers and retailers in the retail supply chain. The manufacturer defines the wholesale price, the retail price is defined as an external variable, and the retailer only needs to determine the order quantity based on the wholesale price and customer demand. At the same time, aiming at the distortion and loss of inventory in the RFID supply chain, the cost-effectiveness of the RFID supply chain is further analyzed, and the inaccuracy of the inventory is eliminated by using the AI RFID supply chain. This part analyzes the overall revenue of the supply chain through simulation experiments, and uses AI RFID technology to solve the coordination impact of inaccurate inventory. Manufacturers determine the best wholesale price based on the sales share or repurchase price in the retail industry and the best order volume. The corresponding indicators of not using RFID and using AI RFID are shown in Tables 1 and 2.

Table 1

Analysis table of indicators without RFID

Variable Q W ψ b πR πM π
Focus 2,068 6,250
Decentralized supply chain Wholesale 1,078 4.4 1,032 2,338 3,371
Revenue sharing 2,068 0.44 0.1 929 5,320 6,250
2,068 0.95 0.4 3,013 3,239 6,250
2,068 1.46 0.7 5,091 1,158 6,250
Buyback 2,068 2.83 1 5,025 1,124 6,250
2,068 4.17 3 3,608 2,641 6,250
2,068 5.5 5 2,211 4,038 6,250
Table 2

Analysis table of RFID indicators using AI

Variable Q w ψ b πR πM π
Focus 1,376 6,789
Decentralized supply chain Wholesale 717 6.4 1,178 2,908 4,087
Revenue sharing 1,376 0.4 0.1 1,012 5,776 6,789
1,376 0.95 0.4 3,319 3,469 6,789
1,376 1.49 0.7 5,641 1,147 6,789
Buyback 1,376 2.72 1 6,324 464 6,789
1,376 4.45 3 4,609 2,179 6,789
1,376 6.18 5 2,894 3,894 6,789

In the table, Q is the order quantity, w is the wholesale price, b is the repurchase price, πR is the income of the store, πM is the revenue from producers, and π is the expected revenue. It can be seen from Tables 1 and 2 that the expected revenue of the supply chain under the wholesale price situation is significantly lower than a focused service line, and the wholesale expected revenue is 3,371 and 4,087, respectively. While the centralized expected returns are 6,250 and 6,789, respectively, and concentrated power lines is 2,879 and 2,702 higher than the wholesale supply chain, respectively. Among them, the double marginal effect is the reason for the obvious decrease in the benefits of the decentralized supply chain. In order to maximize the expected benefits of a decentralized supply chain, coordination and optimization must be carried out.

By adopting AI RFID technology, the inventory inaccuracy rate can be effectively reduced, and it can be seen that the benefits of both the decentralized and concentrated supply systems have improved. After adopting AI RFID, the revenue of centralized and decentralized supply chain without RFID increased by 2,879 and 2,702, respectively. The expected revenue of revenue sharing before and after the adoption of AI RFID is shown in Figure 6.

Figure 6 
                  Expected revenue of revenue sharing before and after the adoption of AI RFID.
Figure 6

Expected revenue of revenue sharing before and after the adoption of AI RFID.

As can be seen from Figure 6, after considering revenue sharing, the expected income decentralized supply chain in both cases before and after the adoption of AI RFID technology reaches 3,600, which is the revenue level of the centralized supply chain. At the same time, with the increase in the revenue sharing ratio, the expected revenue of the retailer increases, and the expected revenue of the manufacturer decreases. Among them, there is the sharing ratio that is the easiest to achieve revenue sharing, so that retailers and manufacturers can obtain equal expected benefits. At the same time, the optimal wholesale price is lower than the manufacturer’s production cost, which well reflects the coordination effect of revenue sharing. Figure 7 shows the expected return of the repurchase situation before and after the adoption of AI RFID.

Figure 7 
                  Expected return of repurchase before and after the adoption of AI RFID.
Figure 7

Expected return of repurchase before and after the adoption of AI RFID.

As can be seen from Figure 7, after the use of AI RFID for repurchase, the expected income of the decentralized supply chain also reaches the level of the centralized supply chain. At the same time, with the increase in the repurchase price, the income level of the manufacturer increases and gradually exceeds the income level of the retailer. This shows that with the increase in the repurchase price, the retailer’s order quantity increases, the manufacturer’s production capacity is more satisfied, and the wholesale price is significantly higher than the production cost.

4.2 Application cases of RFID supply chain

This part is based on the actual situation, combined with AI RFID technology, to carry out further examination of the technology’s use in supply chain based systems.

By taking the cycle of a product from production to destination as the time unit, the AI RFID system and the original process system are compared through data records, and their parameters such as efficiency, resource utilization, and accuracy are compared. The time table of the main links of the system is shown in Table 3.

Table 3

Time table of main links of the system

Project Working time (min)
RFID system Original system
Total laytime 9.9 13.4
Warehousing 3.4 4.9
Ex warehouse 3.5 4.9
Total transport time 29.9 34.9

As can be seen from Table 3, the time-consuming links in the supply chain system are 9.9, 3.4, 3.5, and 29.9 min, respectively in the AI RFID system, while the time consumed for each link in the original system is 13.4, 4.9, 4.9, and 34.9 min. It can be seen from the above data that the amount of time in each system link of the AI RFID system is less than that of the original system, which shortens the entire product passing cycle and greatly improves the work efficiency. The comparison of the efficiency of the system handling goods is shown in Figure 8.

Figure 8 
                  Comparison of system handling cargo efficiency.
Figure 8

Comparison of system handling cargo efficiency.

As can be seen from Figure 8, the efficiency of identifying cargo units and processing goods in the system is still the highest in the AI RFID system in processing business volume, and the processing efficiency of the RFID system is 12% higher than that of the original system. Therefore, the AI RFID system has high efficiency, reduces labor intensity, and saves costs. The system unidentified tag rate is shown in Table 4.

Table 4

Comparison table of system unidentified label rate

Project Electronic label (RFID) Bar code
Unrecognized number 0 18
Total 570 570
Unrecognized rate 0% 3.1%

From Table 4, it can be seen that the application of AI RFID in the supply chain system is very promising. Compared with the original system, it has higher accuracy. From the table data, the recognition rate of the RFID system is 100%, while the original system is worse, only 96.9%. There may be accidental factors in the actual operation process, but in general, the recognition rate of electronic tags is quite high. It has the advantages of strong penetrating power, fast scanning of objects, one-to-many scanning, etc. These characteristics are very important in the identification process, and the table corresponds to this point. So, the application of AI RFID in the supply chain should be vigorously developed.

5 Conclusion

In today’s society with global economic integration and fast evolution of messages, scientific production, advanced technology, and the degree of informatization are the common goals pursued by every manufacturing enterprise. A variety of reasons, such as the influence of diversified customer needs, changes in the competitive environment, and cost pressure, it is difficult for any enterprise to achieve success in all its business fields. It must unite the upstream and downstream enterprises in its supply chain to establish a close business partnership through strategic cooperation between enterprises, learn from each other’s strengths and complement each other’s weaknesses, and give full play to their respective advantages to achieve a win–win situation. The competition among enterprises in the information age is not only the competition of the business environment, but also the competition between the information resources and the supply chain. How to use new technologies to improve production efficiency, reduce costs, optimize workflow, and shorten lead times is the basis of competition. Only by giving full play to the role of the supply chain can we take the lead in the market competition. Based on this, the application of RFID technology has attracted the attention of many manufacturers, experts, and scholars. The application of RFID technology in suppliers also provides many new methods and ideas for supply chain management. Through the research on AI technology, this study puts forward practical suggestions for the development of RFID supply chain, which has important theoretical and practical significance.

  1. Funding information: The author(s) received no financial support for the research, authorship, and/or publication of this article.

  2. Conflict of interest: The authors declare that there are no conflict of interest regarding the publication of this article.

  3. Data availability statement: The data that support the findings of this study are available from the corresponding author upon reasonable request.


[1] A. F. Abd El-gawad, S. Zaki, and E. Kamal, “A survey on machine learning techniques for supply chain management,” Am. J. Bus. Oper. Res., vol. 2, no. 1, pp. 24–38, 2021.10.54216/AJBOR.020103Search in Google Scholar

[2] R. Gautam, A. Singh, and K. Karthik, “Traceability using RFID and its formulation for a Kiwifruit Supply Chain,” Comput. Ind. Eng., vol. 103, no. JAN, pp. 46–58, 2017.10.1016/j.cie.2016.09.007Search in Google Scholar

[3] S. Pichoff, “How to achieve 100% order accuracy: RFID optimizes supply chain shipping and receiving,” Supply Chain Brain, vol. 23, no. 1, pp. 45–45, 2019.Search in Google Scholar

[4] M. Herzl, “RFID and sensor network automation in the food industry: Ensuring quality and safety through supply chain visibility,” Qual. Prog., vol. 50, no. 1, pp. 60–61, 2017.Search in Google Scholar

[5] A. K. Biswal, M. Jenamani, and S. K. Kumar, “Warehouse efficiency improvement using RFID in a humanitarian supply chain: Implications for Indian food security system,” Transp. Res. Part. E Logist. Transp. Rev., vol. 109, no. jan, pp. 205–224, 2018.10.1016/j.tre.2017.11.010Search in Google Scholar

[6] P. R. Podduturi, T. Maco, and P. Ahmadi, “RFID implementation in supply chain management,” Int. J. Interdiscip. Telecommun. Netw., vol. 12, no. 2, pp. 34–45, 2020.10.4018/IJITN.2020040103Search in Google Scholar

[7] R. V. Hoek, “Exploring blockchain implementation in the supply chain: Learning from pioneers and RFID research,” Int. J. Oper. & Prod. Manag., vol. 39, no. 6/8, pp. 829–859, 2019.10.1108/IJOPM-01-2019-0022Search in Google Scholar

[8] D. Hassabis, D. Kumaran, and C. Summerfield, “Neuroscience-inspired artificial intelligence,” Neuron, vol. 95, no. 2, pp. 245–258, 2017.10.1016/j.neuron.2017.06.011Search in Google Scholar PubMed

[9] R. Li, Z. Zhao, X. Zhou, G. Ding, Y. Chen, Z. Wang, et al., “Intelligent 5G: When cellular networks meet artificial intelligence,” IEEE Wirel. Commun., vol. 24, no. 5, pp. 175–183, 2017.10.1109/MWC.2017.1600304WCSearch in Google Scholar

[10] R. Liu, B. Yang, and E. Zio, “Artificial intelligence for fault diagnosis of rotating machinery: A review,” Mech. Syst. & Signal. Process., vol. 108, no. AUG, pp. 33–47, 2018.10.1016/j.ymssp.2018.02.016Search in Google Scholar

[11] J. H. Thrall, X. Li, and Q. Li, “Artificial intelligence and machine learning in radiology: Opportunities, challenges, pitfalls, and criteria for success,” J. Am. Coll. Radiol., vol. 15, no. 3, pp. 504–508, 2018.10.1016/j.jacr.2017.12.026Search in Google Scholar PubMed

[12] L. Caviglione, M. Gaggero, and J. F. Lalande, “Seeing the unseen: Revealing mobile malware hidden communications via energy consumption and artificial intelligence,” IEEE Trans. Inf. Forensics Secur., vol. 11, no. 4, pp. 799–810, 2017.10.1109/TIFS.2015.2510825Search in Google Scholar

[13] L. Qiao, Y. Li, D. Chen, S. Serikawa, M. Guizani, and Z. Lv, “A survey on 5G/6G, AI, and Robotics,” Comput. Electr. Eng., vol. 95, p. 107372, 2021.10.1016/j.compeleceng.2021.107372Search in Google Scholar

[14] T. Yang, A. A. Asanjan, and E. Welles, “Developing reservoir monthly inflow forecasts using artificial intelligence and climate phenomenon information,” Water Resour. Res., vol. 53, no. 4, pp. 2786–2812, 2017.10.1002/2017WR020482Search in Google Scholar

[15] F. Meng, Y. Zheng, S. Bao, J. Wang, and S. Yang, “Formulaic language identification model based on GCN fusing associated information,” PeerJ Comput. Sci., vol. 8, p. e984, 2022.10.7717/peerj-cs.984Search in Google Scholar PubMed PubMed Central

[16] A. F. Chen, A. C. Zoga, and A. R. Vaccaro, “Point/counterpoint: Artificial intelligence in healthcare,” Healthc. Transform., vol. 2, no. 2, pp. 84–92, 2017.10.1089/heat.2017.29042.pcpSearch in Google Scholar

[17] C. Cath, S. Wachter, and B. Mittelstadt, “Artificial Intelligence and the ‘Good Society’: the US, EU, and UK approach,” Sci. Eng. Ethics, vol. 24, no. 7625, pp. 1–24, 2017.10.1007/s11948-017-9901-7Search in Google Scholar PubMed

[18] M. Hutson, “Artificial intelligence faces reproducibility crisis,” Science, vol. 359, no. 6377, pp. 725–726, 2018.10.1126/science.359.6377.725Search in Google Scholar PubMed

[19] D. T. Bui, Q. T. Bui, and Q. P. Nguyen, “A hybrid artificial intelligence approach using GIS-based neural-fuzzy inference system and particle swarm optimization for forest fire susceptibility modeling at a tropical area,” Agric. For. Meteorol., vol. 233, no. Complete, pp. 32–44, 2017.10.1016/j.agrformet.2016.11.002Search in Google Scholar

[20] J. Lemley, S. Bazrafkan, and P. Corcoran, “Deep learning for consumer devices and services: Pushing the limits for machine learning, artificial intelligence, and computer vision,” IEEE Consum. Electron. Mag., vol. 6, no. 2, pp. 48–56, 2017.10.1109/MCE.2016.2640698Search in Google Scholar

[21] R. Chatila, K. Firth-Butterflied, and J. C. Havens, “The IEEE global initiative for ethical considerations in artificial intelligence and autonomous systems [Standards],” IEEE Robot. & Autom. Mag., vol. 24, no. 1, pp. 110–110, 2017.10.1109/MRA.2017.2670225Search in Google Scholar

[22] S. Price and P. A. Flach, “Computational support for academic peer review: a perspective from artificial intelligence,” Commun. Acm, vol. 60, no. 3, pp. 70–79, 2017.10.1145/2979672Search in Google Scholar

[23] E. Burton, J. Goldsmith, and S. Koenig, “Ethical considerations in artificial intelligence courses,” Ai Mag., vol. 38, no. 2, pp. 22–34, 2017.10.1609/aimag.v38i2.2731Search in Google Scholar

Received: 2022-08-23
Revised: 2022-12-06
Accepted: 2023-01-17
Published Online: 2023-03-28

© 2023 the author(s), published by De Gruyter

This work is licensed under the Creative Commons Attribution 4.0 International License.

Downloaded on 1.6.2023 from
Scroll to top button