Chang Ouk Kim

Quality-of-service oriented web service composition algorithm and planning architecture

In the next few decades, it is expected that web services will proliferate, many web services will offer the same services, and the clients will demand more value added and informative services rather than those offered by single, isolated web services. As the result, the problem of synthesizing web services of high quality will be raised as a prominent issue. The clients will face the trouble of choosing or creating composition plans, among numerous possible plans, that satisfy their quality-of-service (QoS) requirements. Typical QoS properties associated with a web service are the execution cost and time, availability, successful execution rate, reputation, and usage frequency. In engineering perspective, generating the composition plan that fulfills a clients QoS requirement is a time-consuming optimization problem. To resolve the problem in a timely manner, we propose a constraint satisfaction based web service composition algorithm that combines tabu search and simulated annealing meta-heuristics. As an implementation framework of the algorithm, we suggest a QoS-oriented web service composition planning architecture. The architecture maintains expert made composition schemas in a service category and assists the client as pure user to choose the one he/she wants to use. The main modules of the architecture are composition broker and execution plan optimizer. With the aid of the UDDI server, the composition broker discovers candidate outsourced web services for each atomic process of the selected schema and gathers QoS information on the web services. After that, the execution plan optimizer runs the web service composition algorithm in order to generate a QoS-oriented composition plan. The performance of the algorithm was tested in a simulated environment.

Forward-backward analysis of RFID-enabled supply chain using fuzzy cognitive map and genetic algorithm

Supply chain is a non-deterministic system in which uncontrollable external states with probabilistic behaviors (e.g., machine failure rate) influence on internal states (e.g., inventory level) significantly through complex causal relationships. Thanks to Radio frequency identification (RFID) technology, real time monitoring of the states is now possible. The current research on processing RFID data is, however, limited to statistical information. The goal of this research is to mine bidirectional cause-effect knowledge from the state data. In detail, fuzzy cognitive map (FCM) model of supply chain is developed. By using genetic algorithm, the weight matrix of the FCM model is discovered with the past state data, and forward (what-if) analysis is performed. Also, when sudden change in a certain state is detected, its cause is sought from the past state data throughout backward analysis. Simulation based experiments are provided to show the performance of the proposed forward-backward analysis methodology.

Agent-based diffusion model for an automobile market with fuzzy TOPSIS-based product adoption process

Research highlights? An agent-based model simulates the market-share dynamics of competing automobiles. ? Consumer-agents base their adoption on fuzzy, multi-attribute TOPSIS method. ? Social influence is raised by the interaction among the consumer-agents. ? An empirical study was conducted to verify the performance of the proposed model. This paper focuses on the product diffusion in a competitive automobile market. Since purchasing a car is costly, the consumers in the market tend to behave like rational decision makers. They naturally compare the attributes of cars (e.g., brand preference, fuel economy, safety, comfort) and make overall decisions. In this paper, we propose an agent-based (AB) diffusion model consisting of tens of thousands of interacting agents. In the model, an agent represents a consumer and bases its multi-attribute decision-making on fuzzy TOPSIS. The decision-making process integrates three purchasing forces: experts product information provided by mass media, subjective weights on product attributes assigned by individual consumers, and social influence (i.e., information delivered from a consumers neighbors who have already adopted products). The AB model executes the agents and observes the collective behavior. In this sense, the model can assist in the analysis of the complex market dynamics. We conducted an empirical study to verify the performance of the AB model.

Adaptive product tracking in RFID-enabled large-scale supply chain

The Radio Frequency Identification (RFID) technology is gradually being adopted and deployed for product flow management in the supply chain. In order to track RFID-tagged products efficiently in the RFID-enabled, large-scale supply chain, this paper first presents the design of a product tracking system that can collaborate with the EPC Network, a suite of network services for RFID data management in the supply chain. Next, we explain a product monitoring procedure that is performed by comparing the actual path of a product with its planned path. Finally, we develop an adaptive product search algorithm based on a reinforcement learning technique to efficiently locate a product deviated from its planned path. Experiment results are provided to show the performance of the search algorithm.

An active product state tracking architecture in logistics sensor networks

Sensor technologies are being introduced as a means to collect the accurate and online information of products in logistics networks. Products with RFID (Radio Frequency Identification) tags can guarantee timely product location visibility. Also, additional sensors can measure location dependent attributes such as temperature and humidity. Representative product centric approaches such as EPC Network and the Dialog system make it possible to secure the item level product information automatically and fast. However, they leave room for more advanced services, especially for active product state tracking service that monitors the locations and attributes of products in a timely manner and triggers exception handling when the constraints associated with the product states are violated. Using state transition model, temporal data model, and publish/subscribe model, this paper proposes an active product state tracking system architecture which is able to track products even when they are enclosed in a box, a pallet, or a container. A simulation based experiment is provided to evaluate the performance of the proposed system.

Situation reactive approach to Vendor Managed Inventory problem

In this research, we deal with VMI (Vendor Managed Inventory) problem where one supplier is responsible for managing a retailers inventory under unstable customer demand situation. To cope with the nonstationary demand situation, we develop a retrospective action-reward learning model, a kind of reinforcement learning techniques, which is faster in learning than conventional action-reward learning and more suitable to apply to the control domain where rewards for actions vary over time. The learning model enables the inventory control to become situation reactive in the sense that replenishment quantity for the retailer is automatically adjusted at each period by adapting to the change in customer demand. The replenishment quantity is a function of compensation factor that has an effect of increasing or decreasing the replenishment amount. At each replenishment period, a cost-minimizing compensation factor value is chosen in the candidate set. A simulation based experiment gave us encouraging results for the new approach.

Service level management of nonstationary supply chain using direct neural network controller

In recent supply chain management, as the online use of inventory data becomes available with the development of Radio Frequency Identification (RFID) technology, it is now possible to monitor the performance measures in a timely fashion. Customer service level is a key performance measure that can be computed as the percentage of times that customer orders electronically received are fulfilled by on-hand inventory. Online monitoring of the service level enables the management paradigm to progress toward the closed loop based control which keeps revising the operation policy to reach a target service level. This paper proposes a closed loop supply chain control based on a direct neural network controller. Unlike the simulation based optimizations which usually need a demand forecasting and an early warning model, our proposed approach has the strength that it can maintain the target only by using the actual ones measured online. For the direct neural network controller, an amplification function which increases the learning speed by augmenting the learning error is proposed. Simulation based experiments were performed to test the performance of the controller against two kinds of unstable customer demand curves.

Neural network controller with on-line inventory feedback data in RFID-enabled supply chain

The item-level visibility which can be secured by RFID technology can help the inventory records of a supply chain correspond closer to the actual inventories. More accurate and timely tracking of chain-wide inventories provides a great potential for optimised on-line control of supply chains. In this paper, we develop an on-line neural network controller that optimises a three-stage supply chain. With the inventory data feedback from an RFID system, the neural network controller minimises the total cost of the supply chain rapidly while satisfying a target order fulfilment ratio. As a test bed of the neural network controller, we develop the beer game model of the supply chain. We demonstrate through simulation-based experiments that the neural network controller shows the highest performance when the inventory data is secured from item-level RFID data.

Multi-agent based distributed inventory control model

We consider a multi-stage inventory control problem with nonstationary customer demand under a customer service-level constraint. We propose a multi-agent based model for distributed inventory control systems. In this model, the agent at the first stage is called a retail agent and those at the remaining stages are called supply agents. The retail agent makes an effort to satisfy a target customer service level by adjusting its order release time according to the changes of customer demand trends. On the other hand, each supply agent tries to control its order release time so that product supply from its upstream agent is synchronized with the order request from its downstream agent. A cooperative demand estimation protocol and a distributed action-reward learning technique are developed to satisfy the target customer service level under nonstationary situations. A simulation based experiment was performed to evaluate the performance of the proposed multi-agent model.

Case-based myopic reinforcement learning for satisfying target service level in supply chain

In the last decade, driven by global competition in the marketplace, many companies have taken initiatives to revamp their supply chains in order to increase responsiveness to changes in the marketplace. The renovation of inventory control system is central to such an effort. However, experiences in industry have shown that the control of inventory in supply chain is not an easy task because of uncertainties inherent in customer demand. In this paper, we propose a reinforcement learning algorithm appropriate for the nonstationary inventory control problem of supply chain that has a large state space. Traditional reinforcement learning algorithms such as learning automata and Q-learning have the difficulty of slow convergence when applied to the situations with large state spaces. To resolve the problems of nonstationary customer demand and large state space, we develop a case-based myopic reinforcement learning (CMRL) algorithm. A simulation-based experiment was performed to show good performance of CMRL.