Monitoring fuzzy temporal requirements for service compositions: motivations, challenges and experimental results

Service compositions are an important family of self-adaptive systems, which need to cope with the variability of the environment (e.g., heterogeneous devices, changing context), and react to unexpected events (e.g., changing components) that may take place at runtime. To this aim, it is fundamental to continuously assess requirements while the system is executing and detect partial mismatches or handle uncertainty. Detecting the entity of a violation is very helpful, since it can guide the way applications adapt at runtime. This paper is based on the FLAGS language we already proposed in our previous work to represent requirements as fuzzy temporal formulas and identify partial violations at the temporal level. The paper illustrates the advantages of using the FLAGS language to express the requirements of service compositions, and proposes a technique to monitor them at runtime. The experimental evaluation demonstrates that the monitoring technique is feasible and the overhead introduced in the running system is negligible.