IBM is providing IT solutions in many industries to address their challenges. Today, there are growing demands for solutions to a wide range of problems in clients' businesses, far beyond what traditional IT systems can provide. Industry Solutions Research is developing advanced solutions for various industries, such as manufacturing, natural resource development, and public sector, informed by deep insights into the industries and clients, and leveraging such technologies as advanced software engineering, compiler design, analytics, and simulation.
- Resilience Engineering Resilience Engineering is a research area focusing on improving the resilience of businesses and services. To make their systems resilient, it is important to develop methodologies, technologies and solutions to recover quickly from an emergent situation, but at the same time it is also very important to design and implement resilient systems which can be utilized in emergent situations as well as in normal situations. To create such systems, we need to sense the environment, anticipate risks with predictive models, initiate actions to alleviate risks, and simulate actions to help human decision making during crises. For example, we are studying decision support technologies involving the effects of simulated weather on traffic conditions as well as more abstract aggregation and analyses of social information.
- Model-based Systems Engineering Products and services are increasingly built as multi-disciplinary systems. For example, a modern automobile has an internal combustion engine, various electromagnetic motors and sensors, a power transmission system, and many control software modules connected via its own local network. A transportation fare card system involves subsystems to interconnect automatic ticket gates, handle credit card payments, and track electronic money. An entire city can be regarded as a functional aggregation of energy, healthcare, transportation, homes, factories and many other systems and services. "Systems Engineering" is an interdisciplinary approach to model and design such complex "Systems-of-Systems" spanning mechanical engineering, electronics, chemistry, software, control engineering, analytics, and physics. Model-based Systems Engineering (MBSE) is being developed to cover the left-hand side or upstream of V-model development using SysML as a promising systems language to represent the interdisciplinary system components. The MBSE projects in Tokyo Research have many objectives and goals:
(1) Create a new model-based development (MBD) methodology in collaboration with industry-leading customers
(2) Build easy-to-use editing tools for complex systems modeling languages and diagrams, with user-friendly interfaces such as simple tabular forms
(3) Innovate a new system identification technology to capture systems as sufficiently precise models by analyzing large-scale observation data
(4) Support consistent and coherent repositories for heterogeneous Systems Engineering databases by connecting and managing them seamlessly.
- Co-simulation Hub When creating new hardware systems, testing on actual equipment is important to resolve design problems, usually with prototypes and rigorous quantitative testing. However, testing costs can be greatly reduced through collaborative virtual prototyping, which shifts the focus from testing on physical hardware to testing on virtual models running as simulations on computers. This kind of virtual testing supports globally distributed and efficient development processes among multiple R&D centers and joint ventures without any need to ship hardware around the world. Co-simulation (from cooperative simulation) is a simulation methodology that allows rapid simulations of a hardware system with heterogeneous components. In a co-simulation, the model components and subsystems are simulated independently and simultaneously by using various simulation tools and algorithms, and the data is synchronized at discrete communication points. IBM Research – Tokyo is working on a new co-simulation methodology to address four challenges:
(1) Scalability: Two or more simulation components can run on multiple computers in parallel, allowing new components to be added without increasing the simulation times;
(2) Accurate Simulations: There is a general tradeoff between simulation accuracy and speed, but our parallel execution methods improve both the speed and fidelity (accuracy);
(3) Real-time Simulations: Even with scalable and parallel simulations, the speed of a simulation is limited by the simulated speed of the slowest component. Rapid system prototyping often calls for integrated simulations of real hardware components and virtual models, but this means the simulated components must be able to keep up with real hardware running in real time. We are developing very fast real-time simulation algorithms by devising new mathematical model for parallel computations; and
(4) Co-simulation Hubs: Open hub interfaces can facilitate remote collaborations by allowing each simulation component to connect to others via a Co-simulation Hub platform.
- Engineering Information Integration The development of complex modern products, such as automotive and electronic devices, requires many experts from diverse disciplines, including experts in system specifications, software development, mechanical design, electronic design, prototyping, and quality assurance. Experts in different disciplines use different applications that store their data in different databases, which frequently prevents us from understanding the relationships among the data or the comprehensive definition of the entire system. This means we cannot understand the impacts of a change in one part of the design. In addition, we may lose track of the master data that was used in the architectural decisions, especially if the data is replicated and independently modified in each application.
The purpose of this research is to develop a collaboration infrastructure that enables traceability and impact analysis for various engineering artifacts by integrating engineering information stored and managed in distributed applications. For example, we are studying technologies for 1. capturing, representing, and manipulating data relationships across different applications, and 2. providing a virtually unified database of engineering information stored in distributed applications.