Paper Title (use style: paper title)

1

Efficient Management of Perishable Inventory by Utilizing IoT An Updated Review of the Literature Maha Riad Faculty of Informatics and Computer Science The British University in Egypt Cairo, Egypt [email protected]

Amal Elgammal Faculty of Computers and Information Cairo University Cairo, Egypt [email protected]

Abstract—Inventory Management is always a challenging and opportunistic area for practitioners and researchers in diverse domains, as the core of any business success is in its inventory management. Therefore, reaching an efficient management system that is capable of optimizing costs is crucial, especially in perishable inventory management. Due to the different nature of perishable inventory, it needs regular control and monitoring. Perishable Inventory deteriorates over time and its quality is affected by the storage and transportation conditions. The advent of revolutionary technologies including Cyber Physical Systems (CPS), Cloud Computing and Internet of Things (IoT) have opened innovative opportunities with pertaining challenges. This article contributes by an analytical study that reviews recent research and development efforts in the utilization of IoT for the management of perishable inventory. This analysis reveals the opportunities that are not yet fully realized, and the challenges that need to be efficiently solved to enable their realization. Keywords—Internet of Things; Inventory Management; Perishables; Inventory Control; Environmental Monitoring

I.

INTRODUCTION

Inventory management is one of the crucial aspects of any product-based business. It was reported in [1] that the US expenditures reaches $1.1 trillion, which are tied up in inventories and that the number of inventories reserved is larger than the real amount of sales by approximately 43%. Although this reveals the importance of having efficient system for inventory management, only 54% of SMEs (Small and Medium Enterprises) track inventory and the rest merely employ ordinary methods or do not track their inventory at all [1]. Inventory management is essential to ensure that the optimum amount of inventory is stored for the optimum time. This would eliminate unnecessary holding costs, outdating or deterioration costs, shortages costs and working capital issues. It is commonly agreed that inventory is classified into three main categories [2]: (i) perishable inventory: which refers to inventory that deteriorates over time, either in fixed rate, unfixed rate or deteriorates on a fixed date, such as processed food, dairy products, blood, medicine, etc. These products are not only sensitive to time increase, but they are also sensitive to storage environment conditions (ii) non-

978-1-5386-1469-3/18/$31.00 ©2018 IEEE

Doaa Elzanfaly Faculty of Informatics and Computer Science The British University in Egypt Helwan University Cairo, Egypt [email protected]

perishable inventory: which refers to inventory that does not deteriorate and is solid, such as home appliances, clothes, etc., and (iii) service inventory: which focus on intangible resources like hotel rooms, flight tickets, etc. The main focus of this paper is on perishable inventory management, due to its different and challenging nature. In this aspect, the introduction of Internet of Things (IoT) [3], and its integration with many business areas open a new gate to facilitate inventory monitoring and its environmental control, leading to full inventory visibility. In 2014, Gartner [4] announced that “A third-fold increase in internet-connected physical devices by 2020 will significantly alter how the supply chain operates” [5]; as it was forecasted that in 2020 the IoT will reach 26 billion units that are installed affecting the information gathered from the supply chains, opening new insights for managers and facilitating new strategies for supply chain operations and management. In addition, Gartner announced that IoT is already used in fleets monitoring and started emerging in manufacturing processes monitoring. By the end of 2016 it was reported [6] that GT Nexus and Capgemini figured out that 70% of companies working in manufacturing and retail began applying digital transformation plan on their supply chain. The contribution of this paper is twofold:  First, it provides a literature review on perishable inventory management by focusing on the utilization of IoT in perishable inventory management throughout the supply chain. Upon which, we highlight the opportunities that IoT opens in this direction, and the challenges that need further study and research from the academic and industrial communities, to be able to take full advantage of this promising paradigm shift. The review methodology followed [7] guidelines.  Second, we draw on the finding of this comparative study by identifying the requirements of a comprehensive framework that integrates IoT and advanced analytical capabilities throughout the supply chain. Such a framework represents a pre-requisite for the efficient usage and utilization of IoT for inventory management. The rest of the paper is structured as follows: Section II presents the background in the area of perishable inventory

2018 IEEE International Conference on Engineering, Technology and Innovation (ICE/ITMC)

2

and supply chain management. Section III presents prominent inventory management approaches that utilize IoT in different aspects. This is followed by section IV that provides a comparative analysis revealing the strengths and weaknesses of the current state-of-the-art approaches. In Section V, we conclude with a list of requirements that entail an integrated framework for inventory management by utilizing IoT. Finally, section VI concludes the paper and highlights future work directions. II.

BACKGROUND

Inventory management is a multi-disciplinary domain, which is realized by the interplay and integration of diverse intricate areas. In this section, the background pertaining to the two main-yet not fully integrated- areas considered in this paper is discussed. That are: (i) inventory management general concepts and background are presented in Section I.A, and (ii) Internet of Things (IoT) us presented in Section I.B. This is followed by a discussion of state-of-the-art studies, which is presented in Section III. A. Inventory Management Background When a business is product-based all the business processes inside the organization work for supporting the product in some way or another, so it is very important to manage all the aspects related to inventory and in this context, it becomes essential to efficiently manage the business supply chain. Sanders and Nada [8] defined the supply chain as a network that links all the organizations that contribute in the product production and delivery process, starting from sourcing the raw materials till delivering it to the end-customer. The typical entities involved in the supply chain are: (i) Suppliers: who are responsible for facilitating the raw materials that are used in the production processes. (ii) Manufacturers/ Producers: who are responsible for using the raw materials to manufacture products. (iii) Wholesalers: who sell the products in bulks to retailers. (iv) Retailers: who sell the products to the end-customers with no restriction on the quantities like the wholesaler. And (v) Customers: are the end-customers who actually uses and benefit from the products features, i.e., they do not resell the products they use them directly. According to Rogers et al. [9], it was agreed that there are eight main areas in the supply chain management process. They are briefly discussed in the following sub-section: Customer Relationship Management Customer Relationship Management (CRM) represents the processes that targets the development of a framework to construct and maintain a continuous relationship with customer [9]. This is done through starting with identifying the target customers (the market segments) and then redeveloping the products/services according to the needs of the customers. Customer Service Management As stated by Kilger [10], the customer service is composed of elements that reflect its multi-dimensional nature. These elements are related to the order pretransaction, the order transaction itself and the post-

978-1-5386-1469-3/18/$31.00 ©2018 IEEE

transaction aspects. Pre-transaction elements are concerned with the products and services being offered to and viewed by the customer, in addition to the construction link between the organization and the customers. Transactional elements include order fulfilment, order cycle times, availability of stock, supply on demand, delivery and shipments arrangements and tracking. The post-transactional elements include the products maintenance and customer feedback gathering and analysis. Demand Management Demand management represents the processes of developing a framework to enable planning for production depending on a prediction of the expected customer demand [9]. These processes start by determining the forecast method and time frame that will be used, in addition to determining the data sources which includes the historical data, business sales’ targets and market research. In vendor managed inventory (VMI), collaborative planning, forecasting and replenishment systems the primary source of data is the customer. There should be contingency plans for unexpected demand or supply. A feedback is typically maintained to enhance the quality of forecasting/prediction. Order Fulfilment Order fulfillment represents the processes that aim at developing a framework to coordinate between the company’s marketing, manufacturing and logistics to fill the customers’ orders. After the order is received all the information about the order is documented and then transferred to the manufacturing department to execute the request. This is followed by the shipping details being provided as well as the bills, then after ensuring the delivery, any remaining payments are collected and feedback is taken [9]. Manufacturing Flow Management Manufacturing flow management represents the processes of developing a framework related to the control, data and material flow for all the activities carried out to manufacture the product and reach satisfactory level of manufacturing flexibility. These activities start with identifying the companies’ manufacturing constraints and capabilities. Then determining production schedule based on the previously determined constraints, the primary forecasted demand and the suppliers manufacturing priority list (which is a list that set priority for each supplier based on their supplies and commitment) [9]. Procurement/Supplier Relationship Management The processes involved under this category concentrate on developing a framework for the activities related with dealing with suppliers and managing the relationships with them [9]. These activities primarily start with categorizing the suppliers according to predetermined schema that classifies them based on several aspects that include: the supplier’s position in the market, the supplier’s service level and components quality, the capabilities of the supplier’s and the capacity the supplier can produce in a specific time range. Each key supplier can have a customized Preferred Supplier Agreement (PSA), which is an agreement/contract


3

that states the terms in which the company and the suppliers have agreed upon to ensure achieving the required benefits [11], other than the standard PSA supplied to other suppliers. Product Development and Commercialization Product Development is related to developing products and enhancing them from time to another in order to continuously accomplish competitive advantage. The supply chain management process involves using suppliers and customers’ opinions, needs assessment, feedback and previous transactions analysis and provide it a basic for the process of product development. After gathering the basic information and coming up with new ideas for existing or new products the resources are assessed and allocated on the development process tasks. After which the development marketing and promotion take place, which is followed with the product’s project plan being prepared and established [9]. Returns Management The activities involved under this category should primarily start with the awareness of the laws that are concerned with the used products, packaging issues and recall campaigns. After that the company’s guidelines for returns should be developed, which is followed by settling the return procedures involving the transportation and credit rules. These bases structure the daily returns that are triggered when a customer decides to return a product. After the company receives the product it starts product disposition and deciding whether to refurbish it, or to recycle it or send it back to the supplier [9]. Standardization Initiative Supply-chain operations reference model (SCOR)A [12] is a framework established by the Supply Chain Council (SCC) and it is used worldwide by organizations and researchers as a standard and reference model. It divides the supply chain processes architecture into 5 main categories; i.e., plan, source, make, deliver and return. Plan processes involve determining ahead all the needed activities and resources throughout the whole supply chain. Source processes involve the arrangements of the orders requested to suppliers of raw materials needed in the product production. The make processes do not only involve the tasks required to produce the product, but also processes related to its recycling, conversion and even machines maintenance. The deliver processes involve orders fulfilment, starting order creation, scheduling till delivering the product and billing the customers. Finally, the return processes describe the processes related to the return of products from the customers and the activities involved for handling these issues. B. Internet of Things (IoT) Background Internet of Things (IoT) term emergence comes from the capability of connecting physical objects and virtual components using the internet. Daya et al. [13] identified IoT from supply chains perspective as “The Internet of Things is a network of physical objects that are digitally connected to sense, monitor and interact within a company

978-1-5386-1469-3/18/$31.00 ©2018 IEEE

and between the company and its supply chain enabling agility, visibility, tracking and information sharing to facilitate timely planning, control and coordination of the supply chain processes”. However, as the vision of IoT from a supply chain perspective is more wider and not limited only to the mentioned capabilities, we propose a more generic definition as follows: “Internet of Things is a network that can connect anything inside a supply chain, including people, machines and systems, in which an efficient supply chain management is guaranteed. This is done through visualizing any object/thing inside the supply chain by monitoring and tracking it and giving a third dimension to organizations’ data, that if analysed can enhance all supply chain processes”. IoT architecture is typically classified into four main layers with respect to Li et al., i.e., (from bottom to top) sensing layer, network layer, service layer and interface layer. The sensing layer represents the physical devices used to collect data such as RFIDs and sensors. The network layer represents how the data are transferred, either wireless sensor network is used (most common) or wired network [14]. The service layer represents a middleware [15] in which services can be created, requested and sent, using the concept of Service-Oriented Architecture (SOA) [14]. In other words, the minimum requirements needed by different applications to communicate and exchange data are delivered via the service layer through which four main components may exist [13]: service discovery, service decomposition, trustworthiness management and service APIs. The most outer layer represents the interface layer inhibiting the different applications that meets different businesses and organizations needs and represents a gateway to interact with the IoT technology and use it in an interactive and human understandable manner [16]. III. TOWARDS MARRYING INTERNET OF THINGS TO INVENTORY MANAGEMENT

Due to the critical role inventory management plays in businesses, industrial and research communities are continuously striving to improve the efficiency of inventory management throughout its supply chain. Accordingly, the literature includes several related surveys studies, the most prominent ones are reported in [17], [13], [18] and [19] that are conducted on integrating IoT and supply chain management. However, our review addresses the integration of IoT to perishables specifically, through more detailed use cases and analyzes the key benefits for businesses and intended objectives in each area. Furthermore, based on this analytical study, we concluded a set of requirements that need to be realized for an integrated comprehensive framework for efficient inventory management by utilizing IoT. In addition, more focused surveys also exist. The most prominent are reported in [20], [21] and [13]. The study in [20] basically focused on inventory management models, however it collects only traditional operational models that do not use IoT. In [21], Ramundo et al. reviewed the different technologies (including IoT) in the food sector generally. However, their review did not specifically target


4

the integration of IoT in the whole supply chain. Furthermore, the use cases considered, mainly focused on the state of practice and did not reveal research challenges. In [13], the study considered the SCOR model (illustrated in details in Section A) to categorize the different research efforts, which investigated the different ways of integrating IoT to supply chain, nonetheless their spectrum targeted the supply chains in a coarse-grained manner and did not emphasize the processes related to perishables and the return on business in each model. In [18], the research concentrated on the timetemperature management facility only and did not investigated any other areas that would benefit from the IoT integration. The work in [19] is a fit far from our concern as it analyzed the current technologies used in smart supply chain applications and concluded that IoT is one of the crucial elements from a smart supply chain perspective. However, the study did not go deeper to reach the use cases that IoT integration can be used in. The analytical study presented in this paper can be distinguished by the comprehensive investigation and analysis of the potential of integrating IoT and inventory management considering complementary and diverse perspective. Integrating IOT into inventory management can be considered in several areas; that’s: inventory management takes place starting from managing the raw materials, needed to produce the product until managing the returns. As a key of managing perishable inventory is the ability to monitor and track it in real-time transportation and storage, in the following sub-sections we discuss prominent state-ofthe-art efforts, mainly categorized into five classes representing potential areas for IoT integration. A. Inventory Location and Quantity Tracking Inventory is transferred several times throughout the supply chain. Typically, an organization needs to track where exactly it is and estimate the remaining time till its arrival. In addition, organizations which have tracking systems can enhance their customer service management (which is illustrated in section I.A), through the right estimation of the delivery time, congestion pointes and through offering customer’s tracking functionality. Verdouw et al. [22] highlighted the advantages of adding RFID readers to the inventory or the trolleys transporting them. They stated that inventory tracking can help to rely less on physical inspection and manual activities, which will decrease errors. More generally, a lot of the management aspects related to logistics can be enhanced; as unnecessary inspections and corrections are no longer needed. Paul et al. [23] proposed a model, which used IoT infrastructure as a base for order fulfillment in a collaborative warehouse. They used the RFID readers for updating the ERP information of the inventory getting into and out of the warehouses. Furthermore, they used sensors for recognizing specific inventory storage area. Moreover, they introduced dynamic scheduling for handling shipment in case of unexpected delay. According to Hahn and Packowski [24], CAMELOT used historical and real-time transportation information to enhance the delivery services. This was done through using

978-1-5386-1469-3/18/$31.00 ©2018 IEEE

historical data that included the time taken to deliver products to a specific destination while keeping track of congestions and time slots. As a result, better estimation of the delivery timing can be available. Moreover, their approach also estimates the fastest route for each time slot throughout the day. Using another perspective, Sizakele et al. [25] investigated the effect of both the IoT and Web 2.0 [26] tools on inventory management in general. They made a software prototype for developing countries to identify low stock levels and wrong inventory places and then notify managers. There is a consensus that tracking location and quantity would enhance the awareness of the inventory quantity and location. Further, this will facilitate faster and more reliable decisions of inventory ordering, trying to reach almost just-in-time ordering strategy if the expected time for delivery, production, transportation and holding is known exactly. In other means the supplier relationship management and manufacturing flow management, illustrated in Section I.A as two key elements in supply chain management, can be boosted. B. Inventory Storage, Environment Monitoring and Control Due to the different nature of perishable inventory, its storage and transportation environment conditions, need to be regularly monitored. As illustrated in Section II.B, the IoT sensing layer, has facilitated the monitoring and control process by introducing the various sensors types that can gather the required information. In addition, through actuators, IoT has enabled the environment control after a change in the environment is detected or when the threshold is reached [27]. This would ensure the quality of the products delivered to the customers or received from the suppliers, supporting the procurement/supplier relationship management processes stated in section II.A. According to Verdouw et al. [22], Smart Agri-Food project used the capability of inventory visualization provided by the IoT to monitor the quality of inventory. In which, the quality monitoring system use as a reference the treatment guidelines as well as standardized pictures to determine quality. Moreover, Smart Agri-Food has the capability to predict the deterioration rate of the transported inventory based on tests undertaken in laboratories on the perishable product in advance. Agro Highway is an ongoing European project, which started in 2015 [28]. The project aims to reach an innovative transportation model for liquids such as milk, by means of controlling temperature and reaching self-dependent cargo system that ensures flexible maritime transportation. Furthermore, it aims to reduce the environmental effect on the inventory to ensure its quality. It is inevitable that storage and transportation environment monitoring and control has the great potential to enable minimizing the costs of outdating products before its expiration. In other words, environmental control ensures that outdating costs are based mainly on products that deteriorated due to time and not due to incorrect storage, which boosts returns management (section I.A) through minimizing returns.


5

C.

Production Machinery Damages and Errors Prediction This area is concerned with the monitoring and control processes related to the production machinery, which is related to any manufacturing process and not limited to processed perishables only. In which, sensors are added to machines to monitor any unexpected event, predict complex events and possible future failures and recommend proactive actions. In addition, these recommendations might include expected maintenance time frames or totally machinery replacement. This would facilitate the manufacturing flow management (referred to in section I.A); as the primary scheduling and awareness of the machines maintenance would ensure having more reliable production plans and schedules. Liu et al[29] proposed IIASMP, which is a concept that integrates IoT technology into the assembly system of mechanical products manufacturing. One of the main characteristics which is identified in IIASMP, is selfadaption. Self-adaption refers to the capability of handling assembly process’ disturbances, through adding RFID tags to the workstations and keeping track of their tolerance. Accordingly, non-conforming products can be disregarded primarily based on the historical capabilities of the station. Another important characteristic identified in the IIASMP concept is the self-maintenance, where machines are capable of detecting failures and autonomously undergo reconfiguration. Similarly, Zinnikus et al. [30] proposed a framework that monitors plants/factories using sensors and accordingly predicts abnormal behavior and deviations. In addition, after failures prediction, maintenance recommendation takes place. Jay et al. [31] carried out a case study to evaluate and track a diesel engine’s health. The proposed system used IoT technology for monitoring the engine by means of measuring the temperature, pressure, engine rotation speed and fuel flow rate. Furthermore, they used the available information gathered from sensors in addition to historical data and predefined thresholds determined by manufacturer, who experienced similar problems with the engine before, for determining the causes of failures. Also, this approach enabled determining degradation at an early stage and the root-causes of such degradation. As revealed from the previous discussion, working proactively towards machine failure, enables the companies to produce more reliable plans that consider the possible maintenance dates and make informed decisions about the production lines and machinery that would produce the target quantity. This would guarantee higher productivity and reduce production wastes costs. D. Shelf-Life Prediction and Remaining Life-Time Estimation Shelf-Life Prediction is concerned with predicting how long will a product stay on shelf before outdating. Shelf-life prediction and/or remaining life-time estimation is crucial for perishables because it is essential to sell them with the optimum price before deterioration. Products with estimated low life-time can be managed differently, for example dynamic pricing can take place.

978-1-5386-1469-3/18/$31.00 ©2018 IEEE

Gary et al. [32] mentioned that the data concerned with the time and temperature of the product would enable setting dynamic expiration dates for the products. In their research, they concluded that the deterioration of perishable products is directly related to the environmental conditions; and so, keeping track of the environment conditions of products (time and temperature specifically) enables cost saving through identifying the products that are still in good condition after the primarily determined expiry date (static expiration date). This in addition to identifying the products that deteriorated and reached bad condition before the static expiration date. They used technology-enabled expiration date (which is based on RFID and sensors that can measure the environmental conditions). In addition, they used a model introduced in 1982 by Labuza that can calculate the deterioration rate with respect to time. However, it cannot be generalized for all perishable products categories; so more customized model is needed. Moreover, they depended on another model proposed in 1984 by Labuza as well, that calculates the deterioration rate depending on the temperature. TOXDTECT is an European project, which was started in 2013 and was closed in 2016 [33]. The project developed [34] an intelligent packaging system to detect meat quality and estimate its remaining life-time with reference to food safety standards, through analyzing special volatile organic components using sensors in the packaging. Estimating the remaining life time, facilitates having more accurate estimation of the amount of inventory to be produced, that can fulfill demand and at the same time does not exceed its needs and can be sold before outdating. This would enhance demand management, which is illustrated in section I.A as one of the key points of supply chain management. Moreover, knowing the exact remaining life time can enhance the inventory management through checking alternative options for the product that can lead to selling it with the optimum price before deterioration E. Products Re-routing Depending on Remaining Life Time Current models do not only stop at estimating the remaining life time of perishables, but take it further to make use of proactive analytics, which does not only predict the remaining life time, but is also capable of recommending what should be done based on this prediction. Accordingly, this would take the management process to a completely enhanced cycle. David et al. [35] proposed a model that uses information collected in real-time while transporting perishable products, to extend material planning model and reach an extended material planning model (EMRP). The EMRP uses calculations of the remaining shelf life time that is based on unexpected environmental changes and compare it with the remaining transportation time for the product; to help in making informed decisions on the best route and retail channel for it. Haass et al. [36] ran simulations to understand the effect of using intelligent containers for transporting bananas and claimed that they would contribute in reducing food wastes. The intelligent containers were composed of several sensors


6

and were capable of adjusting route in case of occurrence of any issue in the cooling system of the container. The containers used the machine-to-machine capability enabled by IoT technology to communicate with each other and switch routes, where containers with inventory with shorter life time, go to the closer destinations and the other containers go to the farer destinations. It is obvious that re-routing products that faces environmental changes that shortens their remaining lifetime increases the probability of selling the product before its expiration, because it arrives earlier to the retailer and so reducing outdating costs. In other words, dynamically calculating the remaining life-time of products can be considered too valuable, in which prescriptive analytics can enable better control and management of the inventory and the order fulfilment, through deciding on the best customer/channel to deliver the products to, with the best quality and reasonable remaining life-time. Furthermore, the return management will be enhanced through decreasing returns. IV. DISCUSSION The discussion in the previous section revealed the significant positive impact of integrating IoT into inventory optimization and control. There is a consensus that IoT opens new opportunities for businesses to enhance their traditional inventory management models in various and rapidly increasing directions. Table I compares and appraises prominent related efforts as discussed in the previous section, based on a number of identified criteria identified from the literature, i.e., “Use cases”, “Areas in businesses to be enhanced”, “Concrete objectives” of approaches category, “Areas considered based on SCORE” and potential “Challenges”. The primary finding is that in all use cases cost could be optimized by utilizing IoT. Moreover, in most of the use cases the challenge of IoT integration would be its highest cost to add the hardware to each product separately, but it would be convenient if it is added per containers and integrated in large organizations. As revealed from the previous discussion, the most recent and prominent research efforts -that usually take an applied nature- utilizes IoT in only one direction. This points to the necessity of designing and building a comprehensive framework for an integrated IoT support in Supply chain with the main objective of strengthening inventory management with multiple objectives. This framework can be used as a reference for research and development in this area. In particular, the data collected using IoT technologies can give the organization a third dimension that could affect its management in different areas. In addition, this data becomes only valuable when we integrate it with analytics. However, all proposed models that use IoT address one or maximum two areas involved in managing the inventory. Therefore, a business model is needed that links almost all the areas involved in managing perishables, aiming to reach optimum profit and costs. The model mainly needs to focus on the areas involved in ‘source’, ‘deliver’ and ‘return’ in SCOR standard (cf. Section I.A); as ‘produces’ is sufficiently tackled in the literature.

978-1-5386-1469-3/18/$31.00 ©2018 IEEE

Next, in Section V, we draw on the results of this comparative analysis by introducing the requirements for a comprehensive framework for an integrated IoT into Supply chain for perishable inventory management. V. REQUIREMENTS FOR A COMPREHENSIVE FRAMEWORK TO INTEGRATE IOT FOR PERISHABLE INVENTORY MANAGEMENT From the previous discussion, it is inevitable that IoT plays a pivotal role in improving the efficiency of inventory management throughout the supply chain network. This is mainly realized from the ability of extracting knowledge from the massive amount of heterogeneous data continuously collected in real-time, which provided visibility over the supply chain networks and assists informed decision making. The study presented in Section IV reveals the necessity of having a comprehensive framework for inventory management using IOT. This section builds on the findings of this analytical study by identifying and listing the set of requirements that are essential for such a framework. Req1: Data Connection Layer: this layer sets up the data sensing, collection, ingestion and pipelining steps to the centralized cloud-based data storage. The main challenge in exploiting Big Data and data collected from several sources is focused around identifying the ways of handling the diversity, heterogeneity and complexity of the data, where traditional mechanisms used for smaller datasets, e.g., manual integration or manual curation of data, are not applicable anymore, due to the volume and velocity of Big Data. This syntactic and semantic incompatibility usually results in data redundancy and inconsistencies that signiﬁcantly affect the quality of the sensed data, and subsequently, the quality of the decisions taken based on this data. Semantic Web technologies (e.g. Ontologies) are the means to deal with these issues. The integration of heterogeneous data throughout the supply chain network by means of Semantic Web technologies is one of the major problems which needs to be tackled in this layer. This includes the development and integration of relevant semantic ontologies which form the backbone of data representation and annotation. The ultimate goal of this layer is to achieve a plug-and-play compatibility of heterogeneous data sources due to different applications and hardware used by each entity/ value chain in the supply chain. Req2: Data Storage and Management Layer: this layer provides a scalable, available, reliable and widely accessible data storage medium which is capable of handling massive amounts of supply chain data. This layer can be implemented using a mix of various data storage systems. For example, it can utilize scalable cloud-based relational database services (e.g. Amazon RDS, SQL Azure) for storing structured data while it will rely on cloud-based NoSQL storage services (e.g., Amazon DynamoDB, Google Datastore) for storing and processing semi-structured and unstructured data sources.


7

TABLE I.

IOT IMPACT ON INVENTORY MANAGEMENT Impact on Inventory Management

Use Cases that uses IoT

Inventory location and quantity tracking

Inventory storage environment monitoring and control Production machinery damages and error prediction Shelf-life prediction and remaining realtime estimation Products rerouting depending on remaining lifetime

Prominent research efforts

Areas in businesses to be enhanced

Areas in supply chain management to be enhanced

[22], [28], [38], [39], [40], [41]

 Enhance customer service  Determine Exact delivery time  More efficient logistics (exact inventory quantity and place is known)  Determine storage conditons  Estimate inventory qulaity  Decrease deterioration rate

 Customer service management  Supplier relationship management  Supplier relationship management  Customer service management

[29], [30], [31], [42], [43]

 Detection of machine failiars  Primary estimation of maintenance dates

 Manufacturing flow management

[32], [33], [44], [45], [46], [41]

 Sell older products first

 Returns management  Demand management

[35], [36], [47], [48]

 Delivery plans and distribution i.e. deliver older products first

[22], [23], [24], [25], [37]

 Returns management  Orders fulfillment

Req3: Analytics Layer: This layer will facilitate a number of engines to provide the analytical functions. Depending on the task requirements, this layer can use one or multiple engines to execute the analytics jobs. For example, machine learning engines (e.g., Apache Mahout, SystemML, BigML) will be concerned with the process of constructing and building adaptive models and algorithms that learn from data as well as adapt their performance as data changes over time when applied to one population for another. A predictive modeling engine will support various statistical and mathematical models to make predictions that are based on historical data. The pattern matching engine will provide tools to identify shapes and patterns in data, perform correlation analysis and clustering of data in multiple dimensions. This layer can make use of various big data analytics systems such as: Hadoop stack, Spark stack, Mahout and R. In general, analytics techniques can be generally categorized into the following categories [49]:  R3.1: Descriptive analysis is used to explain what is happening in a given situation. These techniques are commonly used to answer questions of the form “What has

978-1-5386-1469-3/18/$31.00 ©2018 IEEE

Concrete Objectives

 Minimize logistics costs  Increase Profit  Minimize outdating costs

 Increase Productivity  Minimize machine failures and errors  Minimize outdating costs  Minimize returns  Minimize outdating costs  Optimize Profits

Supply chain areas to be enhanced based on SCORE division

Deliver / Source

Deliver / Source

Challenges

In most of perishables, it is expensive to be done per product rather than a container of products

Integrating the data collected from all the supply chain and ensuring technological interoperability between the suppliers, manufacturers, wholesalers and retailers

Make

Expensive to be done per product on shelf Return

Real-Time actions taken by smart vans and containers Deliver / Return

happened?” Common techniques used for this include descriptive statistics with histograms, charts, box and whisker plots, or data clustering. For example, display the number of outdating inventory per month/year.  R3.2: Diagnostic analysis is used to understand why certain things happened and what are the key drivers. For example, why has a specific delay happened? Common techniques for diagnostic analysis are clustering, classiﬁcation, decision trees, or content analysis. For example, why demand decrease for specific product.  R3.3: Predictive analysis is used to predict what will happen in the future. It is also used to predict the probability of an uncertain outcome. For example, it can be used to predict whether a perishable product will deteriorate faster than usual. Statistics and machine learning offer great techniques for prediction. For example, estimate the order level and demand to be produced.  R3.4: Prescriptive analysis is used to suggest the best course of action to take to optimize your decision outcomes. Typically, prescriptive analysis combines a


8

predictive model with business rules. Techniques such as decision trees, linear and non-linear programming, and Monte Carlo simulation are effective here. For example, recommend the optimum path for the product to undergo, after a change in the expected remaining life time was detected, to ensure minimum costs.  Req4: Presentation Layer: this layer will use tools (e.g., Tableau, Infogram Plotly) for building user-friendly dashboards and applications that display the results of the analytics engine. The supported dashboards need to support various visualization schemes and be able to dynamically display and update the results of the analytics jobs. In addition, Augmented Reality (AR) technology may also be incorporated at this layer. Furthermore, these results provide decision makers and operators with crucial profound insight, thus enabling them to make timely informed decisions.

manufacturing network. The blueprints models are being extended to meet the requirements revealed by the analytical study presented in this paper. The framework will support a feedback loop for the continuous improvement of supply chain processes and components. ACKNOWLEDGMENT Gratitude is to be offered to Late Prof. Dr. Osman Ibrahim, who unfortunately is no longer with us, although he persuaded and motivated us to prepare this paper, before passing away. REFERENCES [1]

[2] [3]

VI. CONCLUSION AND FUTURE WORK Internet of Things (IoT) opens new horizons for the efficient management of inventory concerning both perishable and imperishable products. However, as the case with any new emerging technology, the opportunities it promises come with a vast of challenges that need to be studied and efficiently solved for full advantage. This article focuses on studying the integration and interplay of IoT in different areas related to inventory management and specifically perishables. This article contributes by an analytical study that investigates, classifies, compares and analyses the role of IoT for efficient perishable inventory management throughout the supply chain network. This comparative analysis reveals the opportunities and challenges in a number of identified use cases, which acts as a roadmap for research and development in this area. The main finding of this analytical study is the necessity of having a comprehensive framework for inventory management by utilizing IoT. Based on this, we have identified a list of requirements that entail the major building blocks of such a framework. Current and future work efforts are ongoing in a number of concurrent and related directions with the ultimate goal of drawing such a framework, which realizes the requirements identified in Section V. The framework is being build and validated iteratively using a number of real-life case studies as part of H2020 FOF project ICP4Life. The framework is being designed with a preventive focus, which prevents problems before their occurrence and (semi-)automatically takes recommended recovery actions to prevent/mitigate the impacts of identified discrepancies. The framework is being built to be founded on a well validated and experimented knowledge base based on the novel concept of manufacturing/production blueprints as reported in [49,50], which integrates and interrelates knowledge of the whole

978-1-5386-1469-3/18/$31.00 ©2018 IEEE

[4]

[5]

[6]

[7]

[8] [9]

[10]

[11] [12] [13]

[14]

[15] [16]

[17]

A. Marder, “10 Shocking Inventory Management Statistics,” 2015. [Online]. Available: http://blog.capterra.com/inventorymanagement-statistics/. [Accessed: 27-Sep-2017]. Joost, “The Three Types Of Inventory And Reasons For Carrying Them,” OperationsManager. . Postscapes, “Best Internet of Things Definition.” [Online]. Available: https://www.postscapes.com/internet-of-things-definition/. [Accessed: 24-Sep-2017]. Gartner, “About Gartner.” [Online]. Available: https://www.gartner.com/technology/about.jsp. [Accessed: 12-Nov2017]. J. Rivera and L. Goasduff, “Gartner Says a Thirty-Fold Increase in Internet-Connected Physical Devices by 2020 Will Significantly Alter How the Supply Chain Operates,” Gartner, 2014. [Online]. Available: http://www.gartner.com/newsroom/id/2688717. [Accessed: 13-Aug-2017]. A. Meola, “How IoT logistics will revolutionize supply chain management,” Business Insider, 2016. [Online]. Available: http://www.businessinsider.com/internet-of-things-logistics-supplychain-management-2016-10. [Accessed: 13-Aug-2017]. A. R. Hevner, S. T. March, J. Park, and S. Ram, “Design Science in Information Systems Research,” Des. Sci. IS Res. MIS Q., vol. 28, no. 1, pp. 75–105, 2004. N. R. Sanders, Supply Chain Management: A Global Perspective. United States of America: John Wiley & Sons, Inc., 2012. K. L. Croxton, S. J. García‐ Dastugue, D. M. L. Lambert, and D. S. Rogers, “The Supply Chain Management Processes,” Int. J. Logist. Manag., vol. 12, no. 2, pp. 13–36, 2001. H. Stadtler, C. Kilger, and H. Meyr, Eds., Supply Chain Management and Advanced Planning. Concepts, Models, Software, and Case Studies, Fifth. London: Springer-Verlag Berlin Heidelberg, 2015. JobAdder, “Preferred Supplier Agreement.” [Online]. Available: https://jobadder.com/glossary/psa. [Accessed: 11-Nov-2017]. Supply Chain Council, “Supply Chain Operations Reference (SCOR) Model Overview - Version 10.0,” 2010. M. Ben-Daya, E. Hassini, and Z. Bahroun, “Internet of things and supply chain management: a literature review,” Int. J. Prod. Res., vol. 7543, no. November, pp. 1–24, 2017. M. Ben-Daya, E. Hassini, and Z. Bahroun, “Internet of things and supply chain management: a literature review,” Int. J. Prod. Res., vol. 7543, no. November, pp. 1–24, 2017. S. Li, L. Da Xu, and S. Zhao, “The internet of things: a survey,” Inf. Syst. Front., vol. 17, no. 2, pp. 243–259, 2015. D. Bandyopadhyay and J. Sen, “Internet of things: Applications and challenges in technology and standardization,” Wirel. Pers. Commun., vol. 58, no. 1, pp. 49–69, 2011. L. Ramundo, M. Taisch, and S. Terzi, “State of the art of technology


9

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26] [27]

[28]

[29]

[30]

[31]

[32]

[33] [34]

in the food sector value chain towards the IoT,” 2016 IEEE 2nd Int. Forum Res. Technol. Soc. Ind. Leveraging a better tomorrow, pp. 1– 6, 2016. S. Mercier, S. Villeneuve, M. Mondor, and I. Uysal, “Time – Temperature Management Along the Food Cold Chain : A Review of Recent Developments,” vol. 16, pp. 647–667, 2017. W. Lifang, Y. Xiaohang, A. Jin, and D. C. Yen, “Smart supply chain management: a review and implications for future research,” Int. J. Logist. Manag., vol. 27, no. 2, pp. 395–417, 2016. L. Janssen, T. Claus, and J. Sauer, “Literature review of deteriorating inventory models by key topics from 2012 to 2015,” Int. J. Prod. Econ., vol. 182, pp. 86–112, 2016. L. Ramundo, M. Taisch, and S. Terzi, “State of the art of technology in the food sector value chain towards the IoT,” 2016 IEEE 2nd Int. Forum Res. Technol. Soc. Ind. Leveraging a better tomorrow, pp. 1– 6, 2016. C. N. Verdouw, A. J. M. Beulens, H. A. Reijers, and J. G. A. J. van der Vorst, “A control model for object virtualization in supply chain management,” Comput. Ind., vol. 68, pp. 116–131, 2015. P. J. Reaidy, A. Gunasekaran, and A. Spalanzani, “Bottom-up approach based on Internet of Things for order fulfillment in a collaborative warehousing environment,” Int. J. Prod. Econ., vol. 159, pp. 29–40, 2015. G. J. Hahn and J. Packowski, “A perspective on applications of inmemory analytics in supply chain management,” Decis. Support Syst., vol. 76, pp. 45–52, 2015. S. Mathaba, M. Adigun, J. Oladosu, and O. Oki, “On the use of the Internet of Things and Web 2.0 in inventory management,” J. Intell. Fuzzy Syst., vol. 32, no. 4, pp. 3091–3101, 2017. TechTerms, “Web 2.0.” [Online]. Available: https://techterms.com/definition/web20. [Accessed: 24-Sep-2017]. G. Gaukler, M. Ketzenberg, and V. Salin, “Establishing dynamic expiration dates for perishables: An application of RFID and sensor technology,” Int. J. Prod. Econ., vol. 193, pp. 617–632, 2017. H. 2020, “Demonstration of Innovative concept for high quality Tansport of Perishable Agro Food Liquids including a Model shift to Short Sea and River transport,” European Union, 2018, 2017. [Online]. Available: http://cordis.europa.eu/project/rcn/198536_en.html. [Accessed: 11Jan-2018]. M. Liu, J. Ma, L. Lin, M. Ge, Q. Wang, and C. Liu, “Intelligent assembly system for mechanical products and key technology based on internet of things,” J. Intell. Manuf., vol. 28, no. 2, pp. 271–299, 2017. I. Zinnikus et al., “Integrated semantic fault analysis and worker support for cyber-physical production systems,” Proc. - 2017 IEEE 19th Conf. Bus. Informatics, CBI 2017, vol. 1, pp. 207–216, 2017. J. Lee, H. A. Kao, and S. Yang, “Service innovation and smart analytics for Industry 4.0 and big data environment,” Procedia CIRP, vol. 16, pp. 3–8, 2014. G. Gaukler, M. Ketzenberg, and V. Salin, “Establishing dynamic expiration dates for perishables: An application of RFID and sensor technology,” Int. J. Prod. Econ., vol. 193, pp. 617–632, 2017. CORDIS, “Innovative Packaging for the Detection of Fresh Meat Quality and Prediction of Shelf-Life,” 2017. CORDIS, “Intelligent packaging detects meat quality,” Spain, 2017.

978-1-5386-1469-3/18/$31.00 ©2018 IEEE

[35]

[36]

[37]

[38] [39] [40]

[41]

[42]

[43] [44]

[45]

[46]

[47]

[48]

[49]

[50]

[51]

D. Bogataj, M. Bogataj, and D. Hudoklin, “Mitigating risks of perishable products in the cyber-physical systems based on the extended MRP model,” Int. J. Prod. Econ., vol. 193, no. April 2016, pp. 51–62, 2017. R. Haass, P. Dittmer, M. Veigt, and M. Lütjen, “Reducing food losses and carbon emission by using autonomous control - A simulation study of the intelligent container,” Int. J. Prod. Econ., vol. 164, pp. 400–408, 2015. E. Bottani, M. Bertolini, A. Rizzi, and G. Romagnoli, “Monitoring on-shelf availability, out-of-stock and product freshness through RFID in the fresh food supply chain,” Int. J. RF Technol. Res. Appl., vol. 8, no. 1–2, pp. 33–55, 2017. J. M. Ryan, Guide to Food Safety and Quality during Transportation: Controls, Standards and Practices, Second Edi. Elsevier Inc., 2017. M. L. Tamplin, “Integrating predictive models and sensors to manage food stability in supply chains,” Food Microbiol., pp. 1–5, 2017. R. Jedermann, U. Praeger, and W. Lang, “Challenges and opportunities in remote monitoring of perishable products,” Food Packag. Shelf Life, vol. 14, no. May, pp. 18–25, 2017. G. E. Biccario, V. F. Annese, S. Cipriani, and D. De Venuto, “WSNbased near Real-time Environmental Monitoring for Shelf Life Prediction through Data Processing to Improve Food Safety and Certification,” in 11th International Conference on Informatics in Control, Automation and Robotics (ICINCO-2014), 2014, vol. 1, no. September, pp. 777–782. S. T. March and G. D. Scudder, “Predictive maintenance: strategic use of IT in manufacturing organizations,” Inf. Syst. Front., pp. 1–15, 2017. L. L. Ferreira et al., “A Pilot for Proactive Maintenance in Industry 4.0,” in Control, 2017. M. Göransson, F. Nilsson, and Å. Jevinger, “Temperature preformance and food shelf-life accuracy in cold food supply chains – Insights from multiple field studies,” Food Control, 2017. D. Li and X. Wang, “Dynamic supply chain decisions based on networked sensor data: an application in the chilled food retail chain,” Int. J. Prod. Res., vol. 55, no. 17, pp. 5127–5141, 2017. J. K. Heising, G. D. H. Claassen, and M. Dekker, “Options for reducing food waste by quality-controlled logistics using intelligent packaging along the supply chain,” Food Addit. Contam. - Part A, vol. 34, no. 10, pp. 1672–1680, 2017. R. Jedermann, M. Nicometo, I. Uysal, and W. Lang, “Reducing food losses by intelligent food logistics,” Royal Society Publishing, vol. 372: 20130, 2014. Y. Li, F. Chu, and K. Chen, “Coordinated Production Inventory Routing Planning for Perishable Food,” IFAC-PapersOnLine, vol. 50, no. 1, pp. 4246–4251, 2017. P. Thomas Maydon, “The 4 Types of Data Analytics,” KDnuggets, 2017. [Online]. Available: https://www.kdnuggets.com/2017/07/4types-data-analytics.html. [Accessed: 12-Dec-2017]. M. P. Papazoglou and A. Elgammal, “The Manufacturing Blueprint Environment : Bringing Intelligence into Manufacturing,” 23rd ICE/IEEE Int. Technol. Manag. Conf., pp. 772–781, 2017. A. Elgammal, M. Papazoglou, B. Krämer, and C. Constantinescu, “Design for Customization: A New Paradigm for Product-Service System Development,” Procedia CIRP, vol. 64, pp. 345–350, 2017.
