Skip to Content

Most of the products you use every day have traveled a long way before they got to your hands. Take your phone as an example; it was produced in a factory, stored in a warehouse, sent to a store, and finally obtained by you. As the product travels from one location to the other, a massive amount of data gets generated.

However, many manufacturers have trouble using these data for their benefit. These companies generate a high volume of products with expiration dates and dynamic retail demands, which often increases their operational costs and failure rates.

To help businesses improve efficiency, Hyong Kim’s team is developing Smart Logistics, software that uses large data to optimize logistics operation and detect anomalies.

“What we’re trying to do is understand the data and do two things: find the optimal way to reduce the costs, and understand the operation so that if something goes wrong, we’ll be able to pinpoint what happened,” said Kim, a professor of electrical and computer engineering.

We’re trying to understand the operation so that if something goes wrong, we’ll be able to pinpoint what happened.

Hyong Kim, Program Director, Carnegie Mellon – CMKL | Thailand

Kim’s team has examined the information flow in complex networks, which laid the foundation of Smart Logistics. Engineers used to examine individual nodes in the network to determine their problems, but it was not scalable. To improve on previous methods, Kim’s team has built a map of the information flow and acquired a holistic view of the network. From there, they could spot which part of the system does not work properly.

Products, in a sense, are similar to information. No matter what kind of information we deal with, there are only three events associated with it: processing information, storing information, and moving information. While manufacturers work with physical goods instead of digital data, these basic events remain the same.

The key to deploying products, then, is understanding the global operation instead of studying the entities individually. Kim’s team aims to reconcile the data and pick the relevant ones to grasp the bigger picture. To characterize the logistics application and create behavior models, they will examine both the information flow and the operational relationships among the warehouses and distribution centers. In this way, they can develop algorithms to optimize large scale logistics operations and detect operation failures.

As the director of Carnegie Mellon – CMKL | Thailand in Pittsburgh, Kim is working with ThaiBev, Thailand’s largest beverage manufacturer and one of the founding sponsors of the program. Two of ThaiBev’s main challenges are their high inventory level in the warehouse and the significant number of factories and items they need to handle.

“There are three parts of the business that we’re expecting,” said Teerapan Luengnaruemitchai, the vice president of ThaiBev. “First, we’re expecting to save the logistics costs by lowering the inventory level. Second, we want to save transportation costs by adjusting the routes so that we can ship the products from the right locations to the customers. And thirdly, we need to get alerted about exceptional events quicker so that we can respond to the incidents faster.”

While ThaiBev strives to lower its stock level, it also needs to ensure that it has enough products. As Kim’s team develops Smart Logistics, they are working on predicting consumer demand to find that tricky balance. Once they reconcile the data and validated their results, they will determine the best way to deploy this technology to help ThaiBev achieve its goals.

The other members of Kim’s team include Akkarit Sangpetch and Orathai Sangpetch, the co-director and vice president of Carnegie Mellon – CMKL | Thailand. Kim is also working with several Ph.D. and master’s students from this program. Also called CMKL, Carnegie Mellon – CMKL | Thailand is a collaboration between CMU and King Mongkut’s Institute of Technology Ladkrabang (KMITL), a leading engineering university in Thailand.