Literature Analysis on the Role of Multi-agent in Improving Supply Chain & Logistics Functions
Introduction
Customers’ demands have shifted dramatically in today’s business environment. This tendency causes businesses to refocus their efforts on increasing the value of their products and services, which should be higher than that of their competitors. Higher-value products generate greater sales and increase customer happiness and retention (Delibašić & Mandic, 2012). The Supply chain objective is the management of supplier, manufacturer, and customer relations to enhance customer value while cutting the cost of product flow through the pipelines. It facilitates in the development of product value as well as market competitiveness (Delibašić & Mandic, 2012). Supply chains are a crucial component of every economic system and organization because they encompass all business processes related to obtaining raw ingredients from suppliers, manufacturing items of interest or services using raw ingredients and other relevant resources, and finally delivering the items of interest or services to customers.
The progression of globalization in any sector has increased demand and supply uncertainties, and also the possibility of potential disruption (Patil, 2015). Today’s corporate organizations are more focused on risk reduction for a robust supply chain and maximizing profits, but competitive market issues and instabilities, and the complexity of supply chains, have exacerbated issues for businesses (Patil, 2015). Supplier selection for sub-assemblies, components, and materials, number and location of production assemblies, the amount of capacity at each assembly, and the allocation of each market region is not an easy task. Also, Global supply chains are more complex to control than domestic supply chains due to higher transportation costs and the complication of decision-making due to inventory cost considerations and longer lead times (Patil, 2015).
Implementing supply chain management systems such as advanced planning systems (APS), enterprise resource planning (ERP), and e-commerce is a viable way to manage complex situations since it integrates all of the SCM components (Delibašić & Mandic, 2012). It synchronizes the business process with the logistics chain for improved coordination between resources. However, SCM system performance depends on the effective coordination of each entity to attain maximum profit. Businesses expand traditional SCM structure with independent virtually associated and time-sharing agents. With independent virtual associates and time-sharing agents, SCM expands traditional work structures. Agent technology based on SCM combines advanced architectures and simulation software such as artificial intelligence and distributed systems to act as autonomous and proactive agents that can interact with others.
Methods of Implementation
In volatile and unpredictable multi-agent environments, the agent works as a software system that interacts with the environment and is competent of adaptive and independent activities. Multiple inputs from different entities of the supply chain process have been designed into MAS for SCM. The outputs are divided into two categories: organizational and customer-related. Object-oriented classes, relevant identification, data attributes, and behavioral action rules can all be used to create computing agents. The environment is a source of information, sensors perceive it, and effectors respond in a specific way (Delibašić & Mandic, 2012). Agent behavior is examined using a basic table, logical rules, or other decision-making techniques such as neural networks, fuzzy logic, or artificial intelligence, depending on the type of mapped perceived information. In this paper we will discuss two frameworks (Delibašić & Mandic, 2012): Java agent development (JADE) and CMPS frameworks to automate SCM process and semantic web services in a parallel context.
CMPS
Consumer Agent (CAgent) and Service-Provider Agent (SPAgent) are the two basic types of agents in the CMPS framework, connected with service selection functionality. CAgent enables service requests to be submitted to the CMPS (Pala & Karakostasa, 2014). It is crucial in initiating the reliable service discovery procedure. In CMPS, the SP Agent group of agents is in charge of service providing. The core service module includes service descriptions, case-based reasoning (CBR), and Rule-Based Reasoning technologies depending on OWL ontologies as a foundation for developing various logistic service-related to the domain or global ontologies and semantic descriptions of various web services (Pala & Karakostasa, 2014).
JADE — Design & Procedure
Functions of the JADE-based system, unlike earlier SCM systems, are not generated from inputs and outputs or assumptions and are essentially non-functional requirements of the system. JADE’s major purpose is to make information sharing and collaborative planning less complicated. Entities can be introduced as parallel components to do this (Perera & Karunananda, 2016). During information sharing, entities such as raw material suppliers, manufacturers, distributors, and retail suppliers access a shared ontology as well as their own personal ontologies. Each agent is connected to distinct sub-containers in order to obtain their personal ontologies and is validated by the main container, which contains the system’s domain ontology (Perera & Karunananda, 2016). Although agent behaviors are interlinked with ontologies, however, agents connected with the main container are responsible for agent administration. For local decisions, agents must access a MySQL database, and they must access receive information in XML files for global decisions, while SCM negotiation is ensured using an ACL messaging approach (Perera & Karunananda, 2016). Let’s take consider the steel industry business network system where they want real-time tracking of manufacturing heavy steel components and seamless cross-company information flow linked with the manufacturing process (Helaakoski, Kipinä, Ojala, Peltomaa, & Iskanius, 2006). The system architecture was created depending on the needs and environment. Equal rights and duties for all companies in the corporate network, security and access control measures, various network architectures, and company-specific user interfaces. A service provider-based model, administering shared databases, company-level access controls, and information distribution (Helaakoski, Kipinä, Ojala, Peltomaa, & Iskanius, 2006). The system integrates JADE architecture with FIPA standard with ACL mechanism which enables ontologies for message content (notifications, unexpected changes) (Helaakoski, Kipinä, Ojala, Peltomaa, & Iskanius, 2006). The other examples of the companies are COM ELECTRON which implements JADE to maximize profit by increasing the number of transactions and agreements achieved after several rounds of negotiations (Oprea) and Meta MorphII which implements MAS architecture with ACL to support enterprise integration and supply chain management (Oprea). VIRT CONSTRUCT, an agent-based virtual enterprise, is being developed at the University of Ploiesti. Its goal is to send out order bids and conduct negotiations. The contract net protocol and a service-oriented negotiation model are implemented by VIRT CONSTRUCT (Oprea). JADE was used to conduct several simulations with various environmental parameters.
Results
Advantages and Disadvantages
A successful communication and negotiating process between suppliers and manufacturers lead to better communication networks that allow customers’ needs to be met. MAS is a low-cost, standard communication infrastructure that consists of distinct agents that connect in a real-time, open environment while ensuring transaction security (Delibašić & Mandic, 2012). We could achieve a more effective supply of resources and a decrease in total expenses, time, and resource savings because MAS is better at deploying automated agents save on transit and integration, and information. It lessens the bullwhip effect, lowers user requirement fluctuations, and enhances reserve and resource prediction (Delibašić & Mandic, 2012). On fulfilling these demands MAS relatively lowers overall operational costs. Most crucially, MAS uses resource input to detect key information, get inventory control for instance checking warehouse stock availability and communicating information to other agents, and preserve system security (Delibašić & Mandic, 2012).
However, cooperation in an open environment, distributed allocation of resources, task distribution, agent interoperability, privacy concerns, and overall system resilience are major issues of the multi-agent system approach (Oprea). Non-agent legacy systems faced difficulties and required efficient software development methodologies for agent-based systems. It is important to choose specific model negotiation depending on the application domain that is formulated by a multi-agent. The behavior of a MAS model and design could be improved by applying a robust learning algorithm (Oprea). As discussed above, many companies implement MAS architecture for logistic, resource allocation, telecommunication, and goods market management. TELE TRUCK is a telecommunications-based system used for online dispatching in a supply web’s logistics management node. Here, many sorts of negotiation techniques are utilized to allocate transportation operations within the shipping company chain (Oprea). TELE TRUCK real-world transportation scheduling can be addressed by multi-agent. Truck drivers, trucks, and trailers are self-contained entities with their own goals. These objects function as intelligent agents that can plan, make goals, and communicate in order to supply resources for transportation plans through various negotiating tactics (Oprea).
System Analysis
Steel Net controls cross-organizational information, material flows required to manufacture and deliver goods that are better, and more economical than the competitors. It implemented MAS architecture, collected personnel feedback, prioritize changes and make improvements in future test iterations (Helaakoski, Kipinä, Ojala, Peltomaa, & Iskanius, 2006). Their main focus was usability, reliability, and user-friendliness.
Potential improvement has been observed in SCM communication due to MAS implementation. In this example, the negotiation and communication process was represented between raw material and manufacturing agent. There are five agents involved in quoting different prices. Agent 1 quoted 35 over 25 and second quoted 45 over 35 offer, but after the third agent price moves down and the manufacturer was able to receive the product at a competitive rate through the negotiation process (Perera & Karunananda, 2016).
Discussion
I would have preferred MAS for more flexible and dynamic supply chain system implementation. Many big organizations like P&G also tended to move towards MAS for an effective mechanism to manage operational processes to satisfy the customer. MAS supports the characteristics of competitive product pricing, quality, variations in client expectations, and fast delivery (Nachar). P&G requires an information support system to respond quickly and effectively to client needs, as well as to design, prototype, manufacture, test, and deliver high-quality products to market at a low cost in the minimum possible time, thus MAS fulfills all these factors required for company’s growth. (Nachar) Companies are implementing reusable legacy system modules to manage the supply chain and logistic operations. Now, Microsoft provided Robotic Process Automation (RPA) feature to automate tasks on their own machines for legacy applications (Ghosh, 2021). It will remove the chance of human error during the negotiation and make the best offers in the market without human intervention. Also, in logistics, there is no need to record data manually and make the paperwork. It will automate the process and be capable to store data through pdf, image, and text files just by scanning.
Conclusion
Supply chains are a crucial component of every economic system for the development of product value as well as market competitiveness. MAS helps organizations with scheduling, enterprise coordination, information sharing, order fulfillment, collaborative production planning, provider selection, remanufacturing, and resilience. However, implementation of MAS is complex, requires appropriate learning and agent training process to make an efficient supply chain management system. I think we should also involve legacy application benefits to enhance and customize the product according to specifications.
Reference
Delibašić, B., & Mandic, K. (2012). Application Of Multi-Agent Systems In Supply Chain Management. Management Journal of Sustainable Business and Management Solutions in Emerging Economies, 75–84.
Ghosh, D. (2021, February 22). Looking Back on Power Automate’s 2020. Retrieved March 15, 2022, from Microsoft: https://powerautomate.microsoft.com/en-us/blog/looking-back-on-power-automates-2020/
Helaakoski, H., Kipinä, J., Ojala, K., Peltomaa, I., & Iskanius, P. (2006). Agent technology for supporting real-time supply chain management. International Journal of Agile Systems and Management, 260–275.
Nachar, M. (n.d.). Business Operations in P & G: Strategies for Achieving Competitive Advantage. 1–10.
Oprea, M. (n.d.). APPLICATIONS OF MULTI-AGENT SYSTEMS. Applications of Multi-Agent Systems, 240–270.
Pala, K., & Karakostasa, B. (2014). A Multi-Agent-Based Service Framework for Supply Chain. The 5th International Conference on Ambient Systems, Networks and Technologies (ANT-2014), 53–60.
Patil, M. (2015). Challenges for Supply Chain Management in Today’s Global. European Journal of Business and Management, 61–63.
Perera, L., & Karunananda, A. (2016). Using a multi-agent system for the supply chain management. International Journal of Design & Nature and Ecodynamics, 107–115.