A Novel Forecasting Methodology for Sustainable Management of Defence Technology

By Dong-Sik Jang

A Novel Forecasting Methodology for Sustainable Management of Defence Technology

Technology forecasting is to find the systematized knowledge applied to manage technology efficiently in the future. Using the results of technology forecasting, we can do many important works in management of technology (MOT) such as technological innovation, technology valuation, road-mapping, new product development, etc.. In addition, technology forecasting is important job for R&D (Research and Development) planning and technology strategies in a company. The role of technology forecasting has been increased to diverse areas in MOT. New technology fields, as well as the traditional technology domains including information and communication technology (ICT), bio and medicine technologies, are emerging in the areas of technology forecasting. National defence technology is one of the very important issues in technology forecasting. Most companies and nations would like to improve their competitiveness in national defence. Furthermore, they want to sustain the technological competitiveness for national defence. Most research for technology forecasting were based on qualitative and quantitative approaches such as Delphi and patent analysis. The Delphi technology forecasting is to forecast the future technology of target domain by repeated survey from domain expert group. This provides more subjective and qualitative results for technology forecasting because the Delphi methodology is depended upon the experts’ experience and knowledge.

This study proposes a dynamic methodology for technology forecasting in national defence technology. The study introduces dynamic analysis of sustainable management of defence technology to overcome the limitations of static analysis. The technology growth curve is extracted based on the present and future technology levels as determined from the Delphi method. In order to verify the extracted growth curve, another technology growth curve is extracted based on the unexamined and registered patents from 1974 to present. These two growth curves are examined by regression analysis, and then the curves are adjusted according to the rate of increase in patents.

Proposed Method

An understanding of the stages and trends of technology development using time series analysis is essential for sustainable management of defence technology. Therefore, the proposed methodology in this research includes extracting the growth curve of defence technology analysis using the Delphi technique, and validating the process by the patent-based growth curve. A schematic of the proposed methodology is shown in Figure 1.

The proposed procedure consists of four curve steps from derive survey based growth curve to correct derived growth curves. This procedure performs key role for sustainable technology forecasting and valuation in our national defence technology.

Derivation of the Survey-Based Growth Curve
The Delphi results of target technologies, which were examined in the past, are collected. There are critical technologies such as C4I (command, control, communication, computer, and intelligence) and the weapon system as target technologies. These technologies have been analyzed using the Delphi method in 2003, 2007, and 2010. Each Delphi result of the technologies is evaluated relative to the most developed country’s technology level, which is designated as 100. In this study, we use the Gompertz model for forecasting growth trends of the defence technology because this model showed its better performances than other models in the growth curves from the results of previous researches.

Derivation of the Patent-Based Growth Curve
The annual numbers of patents related to the target technology are collected. This research extracts keywords from the definitions of the core C4I weapon system technologies; the keywords are then used to identify patents from 1974 to 2013. The growth curve is extracted on the basis of the annual accumulated number of patents. The Gompertz model is used again for the patent-based growth curve.

Verification of the Derived Growth Curves
Both the survey-based and the patent-based growth curves are verified to check for statistical significance. Since it verifies the statistical significance only, simple regression analysis is conducted to simplify the model. In this paper, we consider simple linear regression to avoid the overfitting of nonlinear regression model.

Correctly Derived Growth Curves
If there is any similarity between the survey-based and the patent-based growth curves, it is that the survey-based growth curves can be used as the basic data for the sustainable management of defence technology. However, it is impossible to have reliability in the survey-based growth curve if the two curves are completely different. That is, the Delphi survey, which requires the inefficient use of cost and time, has to be conducted all over again. Therefore, this research proposes a new concept of that growth curve that statistically satisfies both the survey-based and the patent-based growth curves.

Results and Analysis

Survey-Based Growth Curve
The Defence Agency for Technology and Quality (DTaQ) in South Korea has conducted a defence science and technology survey related to a system of 27 weapons. This research used the Delphi-expert survey results related to C4I system of 2003, 2010, and 2013. Among the 21 technologies of C4I, six technologies that had been continuously examined in 2003, 2010, and 2013 were selected as the technology levels in each year and future for the experiment.

Figure 2 presents the technology levels of the six technologies. In 2003, technology levels in 2003 and 2010 were predicted. In 2010, technology levels in 2010 and 2013 were predicted. Then in 2013, technology levels in 2013 and 2018 were predicted.

Generally, the DTaQ conducts the defence science and technology survey on a three-year basis. This is a part of the sustainable management of defence technology, and the survey results are used as fundamental data for R and D planning of weapon systems and critical technology. In 2010, 2361 experts participated in 1485 technology examinations. In 2013, there were 707 experts conducting examinations for 1163 technologies.

Figure 3 is a list of growth curves extracted with the Gompertz model. The survey examination results were well-reflected in the derived growth curves since all the values of the sum of squared errors (SSE) were close to zero.

It is important to note that the growth curves vary depending on the examination of the data, even though the curves are derived from the same technology. It comes from the relative comparison with the most developed country’s technology level as the criterion. That is, the technology level is not examined continuously with absolute comparison, so it is necessary to extract growth curves each time when conducting dynamic analysis.

Patent-Based Growth Curve
As keywords were extracted from definitions of target technologies, patent data from KIPRIS had been retrieved to ascertain the number of patents per year. The status of retrieved patents is unexamined and registered during 1 January 1974 to 31 December 2013. In addition, the ended patents are excluded for make a comparison between common generations. We used the bibliographic data, abstract, scope of claims in the retrieved patent documents.  The growth curve was extracted based on the accumulated number of patents per year. Since patents applied in Korea go public 18 months after their application, patents that had been in effect for less than two years were used to construct the growth curve. For example, patents disclosed by 2001 were used to extract the growth curve, and it was compared with the survey-based growth curve.

Conclusions

This paper conducts a technology level evaluation for sustainable management of defence technology. In particular, this research proposes an evaluation methodology of dynamic analysis in order to overcome limitations of static analysis. This methodology has achieved decisive results in the following ways. First, it suggests a technology growth curve based on Delphi analysis, and a verification method to rectify the problem of inadequate reliability. Methods of static analysis have been applied to technology level evaluation cases, and it has been difficult for both performing time series analysis and understanding technology development phases. Existing studies have used the Delphi or survey analysis to collect data and forecast technology level with technology growth curve. The growth curves should have been verified because the curves are extracted with the Delphi method’s limited data. However, the existing studies have not proposed any definite solutions for ensuring reliability. Thus, this research proposes patent analysis to verify the growth curve. Keywords are extracted from the definitions of target technologies to construct growth curves, based on the number of patents per year. Then, regression analysis was performed to figure out whether there are any overlapping coefficient confidence intervals of the survey-based and the patent-based curves. As the patent-based statistical verification process is proposed, it is possible to verify the survey-based growth curve with high reliability. Second, an effective method is suggested to adjust technology level evaluation extracted by the Delphi technique.

The purpose of technology level evaluation is to establish an appropriate strategy for technology development. Thus, it is highly important to determine how technology develops over time. Although it is possible to know the technology development with the growth curve, the reliability of the growth curve is critical. This research proposes a new growth curve analysis by weighted sum of the survey-based and the patent-based growth curves. The proposed growth curve shows that regression coefficient confidence intervals of both the survey-based and the patent-based growth curves overlap in the 99% confidence level. Thus, the proposed growth curve represents both the growth curves statistically. The methodology suggested in this research can provide dynamic technology level data for sustainable management of technology. Therefore, it is expected to be useful for institutes or research laboratories where the technology securing strategy is studied. Traditional approaches to forecast defence technology were based on static forecasting. This is to forecast sustainable technology at a static point from the time period but a technology trend is constantly changing. Thus, in this paper we need a dynamic approach to forecast the future technology to overcome the problem of static forecasting in sustainability of defence technology.

This research contributes to the dynamic forecasting works for sustainable technologies in diverse domains, as well as the defence technology. Though our research has some limitations, this is worthy as a first trial for dynamic technology forecasting in national defence technology. This is an important issue in sustainable technology forecasting. For further research, it is necessary to analyse not only patents but also various quantitative data in order to verify and adjust it in a more precise way.

 

This is an excerpt of the journal article: A Novel Forecasting Methodology for Sustainable Management of Defense Technology, by Kim, Sungchul; Jang, Dong-Sik; Jun, Sunghae; and Park, Sangsung. Published: December 18, 2015 in Sustainability 7, 16720-16736 (http://www.mdpi.com/2071-1050/7/12/15844) under a Creative Commons Attribution License (CC BY 4.0). 

Dong-Sik Jang
Professor

Dr. Dong-Sik Jang is currently working with the School of Industrial Management Engineering at the Korea University, Seoul.