Research and Advances
Computing Applications Virtual extension

Emerging Technologies to Support Supply Chain Management

Posted
  1. Introduction
  2. Data Capture and Transmission
  3. Optimization—The Backbone of SCM
  4. Modeling Languages
  5. Handling Complexity
  6. Artificial Intelligence and SCM
  7. Conclusion
  8. References
  9. Author

New technologies that turn raw data into information and knowledge are changing the way the firms operate. In this context, various forms of supply chain management (SCM) applications are among the enabling technologies transforming how business markets operate. IT research agencies expect SCM applications sales to triple by 2004 [9], in part owing to developments in related technologies that support SCM applications. Prominent vendors in SCM applications market include i2 Technologies, SAP AG, Oracle, and Invensys, which produce a range of hardware and software components that span communication, optimization, and modeling systems. The hardware and software components that support SCM applications can be collectively referred to as the supply chain infrastructure. In this article, we outline important developments in SCM and supply chain infrastructure, including technologies in optimization and modeling systems, which have had a remarkable imprint on supply chain decision-making. Our observations are indicative in nature and shed light on the trajectory these developments are taking, but are not meant to be comprehensive and do not encompass all the facets of supply chain management and SCM applications.

Back to Top

Data Capture and Transmission

In the area of communication, automatic data capture (ADC) technology is fast becoming an important tool to support business transactional information and supply chain processes. ADC systems, which are predicted to grow by almost 16% per year [9], include bar code scanning, voice recognition, and radio frequency data capture (RFDC) systems. The devices supporting ADC systems range from scanners, keyboards, PCs, laptops, servers, PDA devices, cell phones, pagers, and vehicle-mounted instruments. One cannot hope to know the status of items flowing through the system without collecting data in a usable form as ADC systems do. Bar code scanning systems are currently integral to the supply chain infrastructure of many firms, and radio frequency identification (RFID) is gaining increasing acceptance too. While RFID systems are more expensive than bar codes, they can be read at very high speeds, and can collect 40 times the data collectible through traditional bar codes, which can help minimize or eliminate information lag: the gap between an item being sold or shipped and records showing it in need of replenishment. The elimination of this gap would ultimately reduce replenishment time itself.

RFDC is also extending beyond data capture as it becomes technically and commercially feasible at the transmission end, transmitting and even managing information generated by bar codes and warehouse management systems (WMS) across the supply chain. Further, recent development efforts include new standards that enable communication between radio frequency terminals of different suppliers. Emerging technologies in this area will also allow "wired" devices like telephones to operate on the same data network as a radio frequency handheld unit, but without the need to dial into local telephone lines. With increased frequency allocations and efficient coding of information over wireless channels, mobile and wireless networks are likely to become the networks of choice over the wired networks [10].

Clearly, the devices and supporting technologies used for data capture and transmission have increased considerably. ADC is just a part of the infrastructure. Once data is tracked and transmitted, it must be recorded and analyzed to make intelligent decisions. At the corporate level, enterprise resource planning (ERP) acts as the central nervous system of an enterprise [8]. At the next level, supply chain planning and optimization software analyze data and transactions to give managers a range of decision choices. Recent developments in optimization software and modeling language have generated an explosion in implementation and use of SCM applications across companies and industrial sectors.

Back to Top

Optimization—The Backbone of SCM

Managers have increasingly found the latest versions of optimization software capable of providing reasonably good solutions to industrial problems in a short time span, owing to advances in algorithmic techniques and heuristics in artificial intelligence (AI) and computer science. Optimization software now embeds advanced algorithmic techniques such as constraint programming, a declarative, relational styled programming tool originally developed in the 1980s and used solely in the computer science arena, and now being used complimentarily with mathematical programming to solve management and business problems. IT’s constraint component addresses the issues of satisfiability, while the programming level component assembles the steps to solve the problem [6]. The remarkable aspect of this doctrine is that decision variables of a mathematical programming problem can be treated as programming language variables within a computer-programming environment. This flexibility allows the user or manager to program different solution strategies—algorithmic techniques available for solving a particular class of problem—for mathematical programming problems. Constraint programming systems are versatile in that they facilitate the selection of an appropriate strategy based on the user’s understanding of the underlying problem structure.

Some well-known constraint programming applications or systems are embedded into optimization software, such as PROLOG [2] and ECLiPSe [12], to solve large and complex problems with impossibly long solution times. A practical use of such systems involves exploring a hybrid approach to solve problems without writing the application again. The architecture of a constraint programming system is not suitable for finding the optimal solution, but for quickly arriving at a feasible solution of large and complex problems in a spectacularly short time.

The challenge still lies in developing models to support supply chain decision-making using algorithmic techniques like constraint programming. To take advantage of such techniques, the manager still needs to know how to use the system and to develop appropriate models capturing the business situation. It can be a formidable task to model a complex situation using constraint programming, involving a working knowledge of computer programming. Companies are becoming aware of the need for supply chain managers to have expertise in using such systems, while efforts to make constraint programming systems easier to use are also in the works.

Back to Top

Modeling Languages

Another issue of importance in the context of mathematical programming is the generation of the mathematical formulation, or prototype, in a format understandable by the optimization software. Real life business situations involve large-sized problems and rapid generation of the problem prototype is a key concern for the manager since he or she cannot afford to spend an inordinate amount of time on this activity. Therefore, scope of optimization has expanded greatly to include languages for model generation and managing data. Modeling languages like AMPL [4], which develop formulation formats in linear or matrix forms, interface with optimizing software like CPLEX [5], MINTO [7], and XPRESS [3]. Real-life problems often run into thousands of variables and constraints that capture the multiple dimensions of decision-making. The modeling languages make the development of problems feasible within a short span of time. The growing interest in optimization of supply chain planning problems can also be attributed substantially to these incredible technological advances in the modeling language domain.

Back to Top

Handling Complexity

A discussion of optimization must naturally include the time incurred in problem solving. In this context, we need to discuss issues involving the growth of the problem solution time as a function of the size of the problems. It takes different time amounts to solve different problem instances of the same size by using any particular algorithm (say W). The solution times may vary depending upon the algorithm we use. Since different problem instances of the same size (say size p) may require different solution times using any particular algorithm (say W), the term "solution time" refers to the longest time that any instance of size p requires to get solved by employing algorithm W [11]. Different algorithms can solve the same problem instance of a particular size, but algorithms differ in how long they take to solve a particular problem instance of a specific size. The longest time needed to solve a specific problem instance as a function of the size is called the time complexity function (TCF) or simply the complexity of the algorithm. When we speak of the complexity of the problem, we mean the complexity of the most efficient algorithm, known or unknown, which solves it. In the context of optimization, a commonly asked question is how much computing time, as a function of constraints and variables, is required to solve a certain class of mathematical programming problems. The time required to solve problems like the traveling salesman problem, the vehicle routing problem, and the cutting stock problem increases exponentially with the size of the problem. In fact, to arrive at an optimal solution for a fairly large-sized problem of this nature would take years. Supply chain decision-makers must tackle such issues almost on a daily basis—but such situations involve almost instantaneous decision-making. The decision-makers cannot afford to search for optimal solutions, but must quickly access a feasible solution.

Much research has been conducted on rapid solution, an initiative that has benefited in no small measure by the developments in the processing power (millions of instructions per second or MIPS) of computing machines and modeling languages. Currently, optimization of mathematical programming models on the PC is significantly faster than it was on mainframes of several decades ago. This evolution, complemented by the developments in constraint programming and the algorithms in artificial intelligence have greatly facilitated supply chain management decision-making.

Back to Top

Artificial Intelligence and SCM

AI techniques have been put to use in multiple segments of the supply chain. Expert systems embed AI techniques for scheduling in cellular manufacturing systems (CMS) and related manufacturing environments. Many sophisticated business intelligence commercial applications also embed AI algorithms to perform neural networking, clustering, and classification techniques, enabling intelligent decision-making in data mining and online analytical processing. The algorithmic techniques implanted in business intelligence commercial applications offered by companies such as IBM, Cognos, SAS, Oracle, and SPSS run at the back end and give users relevant decision choices in user-friendly interfaces. Decisions involving customer profiling, new product development, retail marketing, and sales patterns are immensely refined using business intelligence tools. Also, as such decisions have an impact on the overall supply chain processes, it is important that business intelligence tools also be linked to SCM applications [8]. Moreover, such decisions are strategic or tactical in nature, and may result in reconfiguration of supply chain network itself, a change in the information exchange across the supply chain, or a modification of business processes. For instance, a change in the target customer segment for a product leads to a change in the distribution system for that particular product. Likewise, a decision to shift the production of an item to a different manufacturing center would affect the raw material supply network for that item.

Back to Top

Conclusion

The discussions in this article relate to developments in areas of communication devices and software, optimization, constraint programming, and artificial intelligence, and their role in supply chain decision-making. These developments are further complemented by parallel improvements in Web-based technology and component-based software development. These advances have made faster development of SCM-related applications possible. This raises the issue of linking such applications together, and with ERP systems and databases. Extensive work is addressing the compatibility challenges to make different applications work together seamlessly.

A recent research initiative at University of Maryland has identified major issues concerning supply chain infrastructure (SCI) [1]. The research has shown that SCIs have produced significant benefits but their implementation is fairly expensive and complex, since firms need systems in sync with their size and line of business. In this regard, firms look towards distributed applications and prefer to adopt scalable solutions that allow them to distribute SCM functionality to their decentralized and spatially dispersed decision-making units. Another challenge is for firms to take thorough advantage of the investment costs in hardware, software, and equipment that constitute the SCI. This challenge motivates firms to initiate policies and incentives to encourage managers to use the new technology in the most productive and efficient manner. Indeed, it is a formidable task to align the management control systems to the new SCI. But this exercise is necessary since compatibility between these two arenas is vital for supporting intelligent decision-making. At the end of the day, the fundamentals that have guided business through the decades still rule. Numerous technological solutions are now available to firms, but decision-makers and manager will not use them unless they support business transactions and decision-making, and furnish lasting benefits.

Back to Top

Back to Top

    1. Ball, M.O., Boyson, S., Raschid, L., and Sambamurthy, V. Scalable supply chain infrastructures: models and analysis. In Proceedings of the 2000 NSF Design & Manufacturing Research Conference (Vancouver, 2000).

    2. Colmerauer, A. An introduction to PROLOG III. Commun. ACM 33, 7 (1990), 70–90.

    3. Dash Associates, 2000. XPRESS—MP Optimizer Sub-routine Library. XOSL Reference Manual. Dash Associates, Blisworth, U.K.

    4. Fourer, R., Gay, D., and Kernighan, B. AMPL: A Modeling Language for Mathematical Programming. Scientific Press, San Francisco, 1993.

    5. ILOG, 2001. ILOG CPLEX Reference Manual. ILOG, Gentilly, France.

    6. Lustig, I., and Puget, J. Program does not equal program: constraint programming and its relationship to mathematical programming. Interfaces 36, 1 (2001), 29–53.

    7. Nemhauser, G., Savelsbergh, M., and Sigismondi, G. MINTO, a Mixed INTeger Optimizer. Operations Research Letters 15, 1 (1994), 47–58.

    8. Stadtler, H., and Christopher, K., Eds. Supply Chain Management and Advanced Planning: Concepts, Models, Software and Case Studies. Springer, Berlin, 2000.

    9. Supply Chain Yearbook: 2000. Manufacturing.Net, Massachusetts, 2000.

    10. Varshney U., Vetter R. Emerging mobile and wireless networks. Commun. ACM 43, 6 (2000), 73–81.

    11. Vasavis, S. Complexity theory. In Encyclopedia of Optimization I, Floudas, C. and Poradalos, P. Eds. Kluwer Academic Publishers, 2001.

    12. Wallace, M., Novello, S., and Schimpf, J. ECLiPSe: A platform for constraint logic programming. ICL Systems Journal 12, 1 (1997), 159–200.

Join the Discussion (0)

Become a Member or Sign In to Post a Comment

The Latest from CACM

Shape the Future of Computing

ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.

Get Involved

Communications of the ACM (CACM) is now a fully Open Access publication.

By opening CACM to the world, we hope to increase engagement among the broader computer science community and encourage non-members to discover the rich resources ACM has to offer.

Learn More