In service-oriented architectures (SOAs), workflow quality of service (QoS) composed of a hierarchically organized set of services ultimately depends on the reliability of the individual atomic services at the very bottom of the service orchestration and composition food chain.

While for some even this (not too complex) insight may be novel, this paper already solves the next step, namely how to dynamically adapt running workflows to degrading or failing atomic services in order to guarantee a certain overall QoS level. This is of special concern in the area of wireless sensor networks (WSNs), where several identical (inexpensive) sensors may provide the same function (for example, temperature in a particular room) to a higher-level workflow (for example, monitoring of a smart home). In case some sensors fail, the workflow engine executing the monitoring process needs to dynamically identify and invoke another (still reliably working) sensor in the same room as a substitute in order to guarantee a high availability (that is, QoS) of the whole monitoring process.

The authors present a component architecture for such a workflow management system (WfMS) and also include all algorithms. In this setting, the component responsible for locating the faulty sensors employs a certain bit-map based Bloom filter for determining certain sensor attributes (for example, membership: “sensor is located in room 101” or “QoS of this sensor below a threshold”).

Finally, the authors convincingly show by virtue of several computer simulation experiments that due to the Bloom filter and the hierarchical setup their WfMS is more efficient (for example, space, computing cost) and faster in detecting and restoring workflow QoS, especially in larger workflows.

I can definitely recommend this easily accessible, comprehensive, and thorough exposition (sufficient for actual implementation) to all researchers and specialists who really want to understand comparable systems or even want to build one.