09/29/03 Feature Story

Debasis Mitra and his team are using mathematical principles to solve problems that can be measured in dollars and cents.

The Bell Labs' Mathematical Sciences Research Center (MSR) is answering the question that math students have asked themselves through the ages -- "Will I ever be able to apply this knowledge in the real world?"

And the answer is a resounding "Yes!"

The MSR's approximately 35 researchers are putting their expertise to use and partnering with Lucent's Supply Chain Networks (SCN) organization to tackle problems that can be measured in dollars and cents, in addition to productivity gains.

"The Math Center has always dealt with the 'what if' questions of science," said Debasis Mitra, vice president, Mathematical Sciences Research, in Murray Hill, N.J. "But we're also using mathematical principles to solve problems and to support the decisions and strategies in other parts of Lucent's business."

Collaboration between Bell Labs mathematicians and their business unit colleagues has been a staple of the company for a long time, according to Mitra. He describes his team, which includes three Bell Labs Fellows and seven distinguished members of technical staff, as "problem solvers by nature." Their work with SCN has given them a new outlet and way to impact the business, he said. "Whether it's molecules or manufacturing -- moving light or moving inventory -- a lot of the same models, principles and ways to look at data apply."


	Whether it’s molecules or manufacturing -- moving light or moving inventory -- a lot of the same models, principles and ways to look at data apply. Debasis Mitra

With the formation of SCN two years ago, Lucent's supply chain function has been re-invented, and our move to a virtual manufacturing model has been a major strategy shift for the business. "These events have resulted in a whole set of new and interesting problems that Bell Labs mathematicians and statisticians are keen to work on and solve," said Mitra.

"A fundamental ingredient in the success of Lucent's supply chain and product design chain transformation has been our strong partnership with Bell Labs and the MSR team," said SCN's Dave Ayers, vice president, Product Design Chain Solutions, in Whippany, N.J. "They provide a strong leverage of analytical design to a wide range of critical challenges that impact Lucent's full stream costs, from test and manufacturing yields to sophisticated demand planning improvements."

Teaming with SCN
In 2001 Mitra and SCN leadership began discussions to formalize some of the ad hoc business planning work between the two groups. "Strengthening the relationship between the MSR and SCN made sense because a majority of the issues concerning uncertainty in demand, supply and manufacturing processes are mathematical problems at their core," Mitra said.

Qiong Wang, Francis Zane and Yuliy Baryshnikov of the Industrial Mathematics and Business Analysis team are working with SCN to ensure that Lucent has the appropriate inventory levels in its supply chain.

The nature of the engagements between the MSR and SCN is based on the complexity, scale and urgency of a given problem. Most projects are dedicated, long-term projects, but MSR researchers also are available for on-demand, high-stakes problem solving.


	A Long Tradition


	Mathematical research has a long and distinguished history at Bell Labs. In 1922 the Mathematics Center was formed. Prior to this, Bell System mathematicians were members of separate engineering departments. The Mathematics Center, one of the original departments when Bell Laboratories started in 1925, served the parts of the business that had no mathematicians of their own. Throughout the years, consulting and collaboration with other areas of the business have remained major functions of the Mathematics Center. Today, the Mathematical Sciences Research Center conducts both applied and fundamental research in the core mathematical disciplines -- mathematics, statistics, electrical engineering, operations research and computer science -- to support Lucent's business needs. "Our strength in fundamentals enhances Lucent's brand value and also allows us to respond to Lucent's ever changing technical and business needs," said Debasis Mitra, vice president, Mathematical Sciences Research. "Our interactions with the market segments and the sales and services organizations attest to the widespread need for mathematics and statistics in every part of Lucent's business."

Many of the problems being tackled by Mitra's team and SCN deal with inventory -- determining what levels of supplies, components and finished products Lucent and our suppliers should keep on hand at any given time. According to the MSR's Francis Zane, technical manager, in Murray Hill, the MSR is working with a team led by Steve Dennis, senior manager, SCN Virtual Manufacturing Services, in Hunt Valley, Md., and Jack Yin, SCN project manager, in Whippany, N.J., on Lucent's "buffer" strategy.

"This strategy helps us determine what sort and amount of inventory we need to ensure we can provide our customers with what they want, when they want it," said Zane. "We also must make sure that we're not holding inventory that is too costly to the business."

To ensure that Lucent has the appropriate inventory levels in its supply chain, the MSR conducts a statistical analysis on a product-by-product basis. "We have models that show optimum inventory levels and lead times for products from all parts of the company," said Zane. "The goal is to develop a single view of how much inventory we're going to keep by product and component."

Working with Whippany-based Michael Massetti, director, Supplier Management, and his team, Zane and his colleagues also are identifying where materials should be held in the supply chain. "Often, it's to our advantage to hold inventory further upstream -- with our suppliers, or our suppliers' suppliers," said Zane.

According to Jim Preston, SCN Global Operations Support, also in Whippany, "The MSR is a powerful resource available to support logical approaches to business issues. The work of the MSR in developing an easy-to-use model that optimizes the balance between inventory, delivery performance and lead time belies the underlying complexity."


	The MSR is a powerful resource available to support logical approaches to business issues. Jim Preston

In addition, Zane and his colleagues work closely with Bob Svenson, director, Supply Chain Design and Optimization, in Murray Hill, and his team on high-level modeling of Lucent's supply chain function to improve the structure of the supply chain, reduce costs and be more responsive to our customers.

Making Sense of the Data
The MSR also has worked with SCN to identify the right point in time to shift from manufacturing one product to another (phase-in, phase-out), as well as on the impact of lead-time reductions on inventory and on the impact of forecast cancellations.

"Sometimes we forecast that customers will buy a complete system -- a switch, for example -- but they end up buying components of the same switch," explains Zane. "The data then shows that people aren't buying the product forecasted, but in actuality they're buying the same product in a different form. In this case, it appears as if we've over-forecasted one product, while under-forecasting another."


	In the past, we may have subjected components to batteries of tests. But after analyzing data sets, we now realize the testing criteria can be loosened without impacting the performance of the product. Diane Lambert

Presented with this problem by the SCN Demand Planning team, the MSR set out to provide a clearer picture of what the data actually showed. "Without a granular view of the data, it looks like an error in our forecasting. The challenge was to identify the right level of data abstraction so that it could be managed and understood, while keeping it low enough that it can capture the impact of common components," said Zane.

Defining New Metrics
In the area of statistics, the MSR's Diane Lambert, director of Statistics and Data Mining Research in Murray Hill, and her team look at how data should be collected, modeled and reported. Lambert and her colleagues -- part of a multi-functional team that also included Lou Manzione's Network Hardware Integration Research department and Pete Hall's SCN engineering group -- recently defined new metrics to better track manufacturing costs and established processes whereby manufacturing information can be used to improve the component selections and product costs.

"In the past, we may have subjected components to batteries of tests. But after analyzing data sets, we now realize the testing criteria can be loosened without impacting the performance of the product. We have been able to lower the threshold of some of our product testing based on a new understanding of the test measurements," said Lambert.

This work led to a significant improvement in manufacturing yields for DWDM products for the Optical Networking Group. For their work, the team was presented with a Central Bell Labs Teamwork Award by Research and Advanced Technologies President Jeff Jaffe.

The Right Tools
Developing tools that help SCN evaluate, understand and better manage the supply chain end-to-end will benefit Lucent's bottom line, according to Mitra. "Data can be useless unless presented or modeled in a way that drives the right actions. These tools will help our engineers make sense of the steady stream of data they already have."

— by Joe Massarelli