Novel network technology combined with advances in hardware development have permitted the enabling of distributed real-time systems and have shortened the time-to-market period. Distributed frameworks, also known as middleware, are often used to integrate enterprise systems, shorten the development time, and reduce complexity. However, to deploy standard middleware in robotics and control applications, we have to deal with the challenge of producing predictable outputs. Most real-time applications in these areas are developed in ad hoc manner, and as such, it is hard to migrate them to new platforms. To overcome this issue while minimizing development effort and increasing reusability for distributed real-time systems, we propose a control framework for distributed real-time systems based on standard middleware specifications. The control framework is composed of asynchronously running task modules, which can be located on either the local or the remote machines. The task modules are connected by an event channel, which uses the publish/subscribe communication method. We also have developed an adaptive event channel in order to meet real-time system requirements and to produce predictable outputs. Detailed development of the control framework along with the adaptive event channel are assessed through a set of experimental results.