COMET: : a component-based real-time database for vehicle control-systems

Abstract: Vehicle control-systems have evolved from small isolated controllers to complex distributed computer-systems. These systems include nodes spanning from simple 8-bit micro-controllers with a minimum of memory to complex 32-bit processors with vast resources. The main motivation for this evolution is the need for increased functionality in motor vehicles. Examples of such functionality include momentary fuel consumption measurements, anti-spin systems, and computerized diagnostics of vehicle-status. The control of the increased functionality requires the handling and maintenance of larger volumes of data, and has created a need for a uniform and efficient way to access and maintain this data. A real-time database management system could satisfy this requirement but an extensive survey of commercial and experimental database management systems has shown that there is currently no database system suitable for vehicle control systems available. In today's systems, data management is performed in an ad-hoc fashion at a low level of abstraction, using internal data-structures, e.g., shared variables and structures. This approach requires that the consistency of the data is maintained by the application, by, for example, resolving data access conflicts through the use of semaphores. This thesis presents a flexible and configurable database management system designated COMET, suitable for embedded systems and in particular, vehicle control-systems. To be able to handle the varying requirements imposed by different systems, COMET emphasizes configurability and tailorability, by adopting a component-based architecture.The result of this research is the implementation of COMET BaseLine, which is an instance of COMET suited to a particular vehicle control-system. The required behaviour of this database is based on requirements gathered from a case study performed at Volvo Construction Equipment Components AB in Eskilstuna. To fulfill these requirements, a concept called database pointers has been introduced and implemented. Database pointers provide controlled direct access to individual data elements in the database, efficiently and temporally deterministic, providing at the same time a high level of abstraction.