Vehicle/driver monitoring and sensing is one family of applications which falls in the M2M (Machine-to-Machine) area.
Drivers, fleet owners, transport operations, insurance companies are stakeholders which need to have analytical reporting on the mobility patterns of their vehicles, as well as real-time views in order to support quick and efficient decisions towards eco-friendly moves, cost-effective maintenance of vehicles, improved navigation, safety and adaptive risk management.
Vehicle sensors do continuously provide data, while on-the-move, which are stored and processed in order to provide valuable information to stakeholders. Applications identify speed violations, abnormal driver behaviors, and/or other extraordinary vehicle machine conditions, produce statistics per driver/vehicle/fleet/trip, correlate event with map positions and route, assist navigation, monitor consumptions, and perform many other reporting and alerting functions.
The main functionalities required by the use case are:
- Analytics with variant types of aggregation logic: Average and absolute values in time (speed, fuel, kms, etc.)
- Trip-level analysis: Identification of trips, the relevant trip data and production of statistics on a trip basis
- Real-time monitoring and alerting: Production of current vehicle position and route on the map. Production of real-time event notification upon the identification of certain value conditions or upon geo-referenced data (i.e. vehicle is near a certain Point-of Interest)
While current implementations satisfy the initial business needs, they will not be able to respond to the growing needs, and the main concern is to follow new solutions that can scale efficiently. An appropriate design compensates with both the need of analytical queries in the huge raw data volume, as well as with the real-time information required.The design preferred produces table views on both the real-time and past analytical data; when necessary, it combines the query results from both the real-time view (stored in-memory) with analytical data views (from columnar stores). High throughput and quick processing of event data streams is supported by complex event processing. Responses also carry data from relational databases. Combining all these frameworks (complex-event processing, in-memory, columnar and relational) need the unifying layer of CoherentPaaS, which is especially useful when a query function is actually facing the need to provide the union of results across two different stores; one storing analytical data and the second storing the real-time data.
By utilizing key aspects of the CoherentPaaS platform:
- The use of different technologies over the same programming and transaction layer will satisfy important business needs that need to be addressed by multiple data stores with a rich variety of functionality enabling to perform all kinds of query processing.
- Holistic Transaction Manager assures consistency between different data stores
- Complex-event processing simplicity is enhanced, since the applications will be written based on queries that are declarative as opposed to the current programmatic APIS available such Storm etc. that take longer to be written and require skillful programmers.
- Ability to scale in cloud environments