A data utility model for data-intensive applications in fog computing environments

Abstract

Sensors, smart devices, and wearables have been widely adopted in recent years, bringing to the production of a vast amount of data which can be shared among several applications as input for their analysis. Data-intensive applications can benefit from these data but only if data are reliable and timely, and if they fit the requirements of the application. Designing data-intensive applications requires a trade-off between the value obtained by the analysis of the data, which is affected by their quality and volume, and the performance of the analysis that can be affected by delays in accessing the data and availability of the data source. In this chapter, we present a Data Utility model to assess the fitness of a data source with respect to its usage in a data-intensive application running in a Fog Computing environment. In this context, data are provided using a Data-as-a-Service (DaaS) approach, and both data storage and data processing can be placed in a cloud resource as well as in an edge device. The placement of a resource affects the quality of the service and the data quality as well. On this basis, the Data Utility model provides a support for making decisions on the deployment of data-intensive applications according to the impact of the task location, and on the selection of proper data sources as input for the application according to the application requirements, taking into consideration that both tasks and data can be moved from the edge to the cloud, and vice versa, to improve the efficiency and the effectiveness of applications.