Service-oriented platform for real-time optimization and execution of RFID-enabled smart container loading

Abstract

Logistics plays a more and more important role in manufacturing company to sharp competitive in today’s supply chain integration. As the end of logistics in manufacturer, container loading intellectualization has attracted more and more focuses. Harmonious coordination between planning and execution ground high performance of the container loading.

Several challenges exist in business operations. Firstly, container loading planning is time-consuming. Several case-based constraints exist in the optimization defined by different customers and the Customs. Secondly, execution driven by container loading is inefficient. Operators may use unclear or unreadable paper-based guidance during execution which is inconvenient to handle. Besides, due to various situations, shipping documentations are complex. Afterwards Order-picking, a critical part in executions, is a costly activity. Thirdly, coordination during whole container loading process is another challenge. Information sharing between planning and execution is delayed and inaccurate. The whole procedure usually cannot be monitored by managers in a remote office. Problems cannot be addressed without timely provision of data. This research targets to establish a smart container loading platform based on real life practices and constraints. Unlike common warehouse operation, warehouse executions are driven by container loading.

To overcome the challenges, an integrated solution is proposed with three main parts. The first one is to establish RFID-enabled real-time smart environment where device-level data is collected timely with bidirectional communications under Wi-Fi occasionally connected situation. Secondly, Service-oriented Architecture (SOA) is applied in rule-based computation and heterogeneous information sources integration. In rule-based computation, constraint is computed and wrapped as rule. It is a kind of service which is restored in service repository according to its properties. Under different cases, several rules are addressed and invoked in a specified sequence to realize different business logics. And the platform should communicate with heterogeneous sources within short time. Systems wrap application programming interfaces (API) as web services which can be invoked flexibly at any time. Data fetch process is divided into several threads which extract data concurrently. Thirdly, container loading optimal model is provided based on practical constraints. Based on various rules, shipments are clustered as groups before loading. To load containers more equally, heuristic algorithm is provided with reducing container capacity continuously until other constraints are violated or more shipment groups are separated.

Four contributions have been made in this study. Firstly, a five-layer system architecture is proposed for small and medium enterprises (SMEs) applications especially in occasionally connected network environment, rule-based computations, and efficient data exchanges. Secondly, several practical constraints of loading planning are addressed and a heuristic algorithm is designed. Shipments can be equally loaded while satisfying most conflicted constraints. Thirdly, a solution for auto-ID enabled warehouse execution is provided. The solution is driven by container loading equipped with real-time information capturing and processing. Fourthly, a guidance can be refereed for real life implementations in SME. The guidance summarizes system customization procedures with reengineered resources.