BACKGROUND AND OBJECTIVES: Mobile and ubiquitous devices are everywhere, generating an exorbitant amount of data. New generations of healthcare systems are using mobile devices to continuously collect large amounts of different types of data from patients with chronic diseases. The challenge with such Mobile Big Data in general, is how to meet the growing performance demands of the mobile resources handling these tasks, while simultaneously minimizing their consumption. METHODS: This research proposes a scalable architecture for processing Mobile Big Data. The architecture is developed around three new algorithms for the effective use of resources in performing mobile data processing and analytics: mobile resources optimization, mobile analytics customization, and mobile offloading. The mobile resources optimization algorithm monitors resources and automatically switches off unused network connections and application services whenever resources are limited. The mobile analytics customization algorithm attempts to save energy by customizing the analytics processes through the implementation of some data-aware schemes. Finally, the mobile offloading algorithm uses some heuristics to intelligently decide whether to process data locally, or delegate it to a cloud back-end server. RESULTS: The three algorithms mentioned above are tested using Android-based mobile devices on real Electroencephalography (EEG) data streams retrieved from sensors and an online data bank. Results show that the three combined algorithms proved their effectiveness in optimizing the resources of mobile devices in handling, processing, and analyzing EEG data. CONCLUSION: We developed an energy-efficient model for Mobile Big Data which addressed key limitations in mobile device processing and analytics and reduced execution time and limited battery resources. This was supported with the development of three new algorithms for the effective use of resources, energy saving, parallel processing and analytics customization.
BACKGROUND AND OBJECTIVES: Mobile and ubiquitous devices are everywhere, generating an exorbitant amount of data. New generations of healthcare systems are using mobile devices to continuously collect large amounts of different types of data from patients with chronic diseases. The challenge with such Mobile Big Data in general, is how to meet the growing performance demands of the mobile resources handling these tasks, while simultaneously minimizing their consumption. METHODS: This research proposes a scalable architecture for processing Mobile Big Data. The architecture is developed around three new algorithms for the effective use of resources in performing mobile data processing and analytics: mobile resources optimization, mobile analytics customization, and mobile offloading. The mobile resources optimization algorithm monitors resources and automatically switches off unused network connections and application services whenever resources are limited. The mobile analytics customization algorithm attempts to save energy by customizing the analytics processes through the implementation of some data-aware schemes. Finally, the mobile offloading algorithm uses some heuristics to intelligently decide whether to process data locally, or delegate it to a cloud back-end server. RESULTS: The three algorithms mentioned above are tested using Android-based mobile devices on real Electroencephalography (EEG) data streams retrieved from sensors and an online data bank. Results show that the three combined algorithms proved their effectiveness in optimizing the resources of mobile devices in handling, processing, and analyzing EEG data. CONCLUSION: We developed an energy-efficient model for Mobile Big Data which addressed key limitations in mobile device processing and analytics and reduced execution time and limited battery resources. This was supported with the development of three new algorithms for the effective use of resources, energy saving, parallel processing and analytics customization.