Towards a customized vehicular maintenance based on 2-layers data-stream application

Demand-Driven Data Acquisition for Large Scale Fleets

“…The presented demand-driven data acquisition makes large fleets accessible to individual data consumers more efficiently than previous methods. Instead of acquiring a static set of sensors from each vehicle, such as [16,[18][19][20][21][22][23][24][25][30][31][32], we minimize the scope of data acquisition for each vehicle individually. This optimization facilitates access to subsets of the total fleet and sensor capacity, as the remaining data do not need to be transmitted and processed.…”

“…The development of systems to collect vehicle data is often a prerequisite for answering research questions based on the data obtained [16][17][18][19][20][21]. For example, the authors of [16] have developed driving behavior analysis methods based on continuous streams of vehicular data.…”

“…The sensor selection depends on the sensors required by the data consumers and can be continuously adjusted even after the data subject gave consent. In other works, data acquisition takes place as a static routine that is decoupled from the requirements of individual data consumers [16,[18][19][20][21][22][23][24][25][30][31][32].…”

“…Apart from [23], the presented systems are not integrated natively with the vehicle. Consequently, the authors had to utilize the On-Board Diagnostics (OBD) interface to acquire sensor data from the vehicle [16][17][18][20][21][22]25]. The OBD is a physical interface within the car that exposes a small fraction of the data available on the Controller Area Network (CAN) [40].…”

“…While some works use OBD as a sole vehicular data source [16,18,20,21], others also obtain data from their own sensors [17,22,25,43] or use them exclusively [19]. A combination of different data sources can also occur on a higher level by aggregating data from existing telematics systems and smartphones [24].…”