A Bounded-time Service Composition Algorithm for Distributed Real-time Systems

Handling the dynamics of future service-based distributed systems in real-time is a complex problem; a number of state transitions or reconfigurations take place that must be handled in real-time; this requires to impose some bounds to the structure of the system to ensure timely operation. We present a model for real-time reconfiguration based on a service model using the concept of service implementations that are actual versions of a specific functionality or service. Over this model, we present an algorithm for service composition that provides a feasible solution compliant with the provided application quality of service (QoS) criteria that consists of a set of service implementations. This algorithm executes in linear time by drawing a clear separation between the composition algorithm and the real-time analysis of the service implementation paths; this is a key idea that allows to have the linear time service-based composition algorithm as a simple straight forward graph search guided by some values or heuristics related to the application QoS. Our solution targets real-time systems being, therefore, appropriate for timely reconfiguration. The proposed solution is evaluated using an profiling tool. We provide experimental results of this tool showing the suitability of the approach and the proposed concepts.