flink-vehicle-telematics

Stream processing of simulated on-vehicle sensors data using Apache Flink

Problem statement • Running the program • Authors • License

Problem statement

In this project we consider that each vehicle reports a position event every 30 seconds with the following format: Time, VID, Spd, XWay, Lane, Dir, Seg, Pos.

Metric	Description	Domain
Time	timestamp when the event was emitted	integer
VID	identifies the vehicle	integer
Spd	speed in mph	integer [0, 100]
XWay	identifies the highway where the event was emitted	integer
Lane	identifies the lane of the highway from which the position report is emitted	integer [0, 4]
Dir	indicates the direction	0 for Eastbound and 1 for Westbound
Seg	identifies the segment from which the position report is emitted	integer [0, 99]
Pos	identifies the horizontal position of the vehicle as the number of meters from the westernmost point on the highway	integer [0, 527999]

The goal of this project is to develop a Java program using Flink implementing the following functionality:

Speed Radar: detect cars that overcome the speed limit of 90 mph
Average Speed Control: detects cars with an average speed higher than 60 mph between segments 52 and 56 (both included) in both directions.
Accident Reporter: detects stopped vehicles on any segment. A vehicle is stopped when it reports at least 4 consecutive events from the same position.

Notes:

All metrics must take into account the direction field.
A given vehicle could report more than 1 event for the same segment.
Event time must be used for timestamping.
Cars that do not complete the segment (52-56) are not taken into account by the average speed control. For example 52->54 or 55->56.
A car can be stopped on the same position for more than 4 consecutive events.
An accident report must be sent for each group of 4 events. For example, the next figure shows 5 events for the car with identifier VID=3:

900,3,0,0,0,1,51,306000
930,3,0,0,0,1,51,306000
960,3,0,0,0,1,51,306000
990,3,0,0,0,1,51,306000
1020,3,0,0,0,1,51,306000

Running the program

The main program receives two arguments: the path to the csv with the events and the destination folder to write the results. In this last folder, you will find three csv files, each one storing the output of three types of events detected as explained above. There is a sample file in the test/resources. We recommend to take a look at the test to see how the program works.

To run the app with the sample csv, run the following commands:

mvn package
$FLINK_HOME/bin/start-cluster.sh
flink run -c fediazgon.VehicleTelematics \
    target/flink-vehicle-telematics-1.0-SNAPSHOT.jar \
    src/test/resources/test_input.csv /tmp
$FLINK_HOME/bin/stop-cluster.sh

The first command will also run the tests. Finally, check the tmp folder for the results (accidents.csv, avgspeedfines.csv and speedfines.csv).

Authors

Fernando Díaz
Eva Gil

License