Open Problems in the Distributed Control Systems Domain

Lonnie R. Welch


This special issue of Parallel and Distributed Computing Practices focuses on the topic of Engineering Distributed Control Systems (DCSs). The guest editorial discusses open problems in the DCS domain. The articles contained in this issue are selected papers from the Joint Workshop on Parallel and Distributed Real-time Systems, a forum for presentation of engineering techniques relevant to control systems that are distributed or parallel.

Many difficulties confront those who must engineer the emerging generation of real-time computing systems. Such systems have rigorous Quality of Service (QoS) objectives. They must behave in a dependable manner, must respond to threats in a timely fashion and must provide continuous availability, even within hazardous environments. Furthermore, resources should be utilized in an efficient manner, and scalability must be provided to address the ever-increasing complexity of scenarios that confront such systems. The difficulties in engineering such systems arise from several phenomena, one of the most difficult being the dynamic environments in which they must function. Unfortunately, the classical paradigms for engineering real-time systems do not address this phenomenon. New paradigms, techniques and tools are needed! This article provides a justification of this viewpoint.

Control systems perform actions in response to conditions detected in the environment that they inhabit. A typical control system consists of sensors that monitor external conditions and software that filters, correlates and evaluates the sensor data. Events detected by the evaluation software activate software that initiates responses via actuators and guides the actions to successful completion. The processing performed by a control system depends on the environment in several ways. The load of the sensing path is primarily a function of the number of items detected in the environment. The load of the actuation path depends on the rate at which environmental events occur and are detected by the sensing path. Thus, the characteristics of the control system's environment should be a fundamental consideration in the definition of a successful engineering paradigm.

Real-time control systems, as their name implies, function in the real-world. What is the characteristic of their environment? Occasionally, one may encounter a fully deterministic real-world system. One may also, perhaps, find a real-world system that predictably follows a statistical distribution. However, it is often the case that the real-world environment is dynamic; it cannot be modeled by a deterministic model, or even by a stochastic model based on a time-invariant statistical distribution.

The environments of many control systems are dynamic. An individual software entity may therefore have large variations in its execution times (we have observed variations of three orders of magnitude among lower bounds and upper bounds on execution times of software paths through real-world control systems!). This makes analysis of timing conformance very difficult. Therefore, paradigms for the engineering of real-world distributed control systems should allow for dynamic behavior.

Most of the existing paradigms are based on worst case assumptions regarding execution times and resource usage, and have assumed that all system behavior follows a statically known pattern. When applying the techniques to systems that function in unknown environments, it is sometimes impossible to obtain some of the parameters required by the models! Paradigms are needed to allow the modeling of systems that work in dynamic environments.

On a separate note let me add two comments about the developments in the journal. First, on the next pages you will find short self-introductions prepared by six members of the Editorial Board. We plan to proceed with such introductions until all of the members are presented. Second, a new initiative is being introduced in the Calls for Papers and Participation area: Software Reviews. Dr. Hong Shen and Dr. Domenico Talia will be running this section. If you are interested in participating by submitting software for review or reviewing it, please contact them directly.

Lonnie R. Welch
Ohio University



  • There are currently no refbacks.