Middleware architecture for control an heterogeneous expert system

Expert Systems are tools used when it is required to take decisions, since they have artificial intelligence and human-like reasoning capacity, they need expertise and knowledge of human experts in the area where it will be used; they serve as support to the human experts to monitor and manage systems of various kinds. An Expert System that is used as an auxiliary tool in large systems or with lots of events, needs to be able to scale as the supervised system grows. This paper presents a Middleware architecture that allows to add to an Expert System the following capabilities: scalability, high availability and fault tolerance. The Middleware is designed to distribute the services of the system among multiple instances, ensuring that there is always an instance that can replace any other failed. In this way, an Expert System could have some features of the distributed systems. The Middleware encapsulates the communication mechanisms of the distributed system and saves the user from having to deal with this layer. In this proposal the Middleware is composed of TAO, which is an implementation of the CORBA specification for distributed systems, also incorporates the Boost library to enable a concurrent work on each of the modules of the system and uses an interface type REST to establish communication with the user.