1.
Pilot SC4
L. Selmi - BDE - SC4 Webinar
BDE SC4
02.12.2016

2.
Objective of the Pilot SC4
L. Selmi - BDE - SC4 Webinar
A scalable, fault-tolerant and flexible
platform based on open source frameworks
that can process unbounded data sets and
graphs.

3.
Microservice Architecture
L. Selmi - BDE - SC4 Webinar

4.
Message Broker
L. Selmi - BDE - SC4 Webinar
Apache Kafka is a high-throughput
distributed durable messaging
system
Apache Kafka

5.
Kafka Cluster
L. Selmi - BDE - SC4 Webinar
Apache Kafka

6.
Stream and Batch
Processor
L. Selmi - BDE - SC4 Webinar
Apache Flink is an open source
platform for distributed stream and
batch data processing.
Apache Flink

7.
Flink Cluster
L. Selmi - BDE - SC4 Webinar
Apache Flink

8.
Storage and Indexing
L. Selmi - BDE - SC4 Webinar
PostGis is a spatial database that
stores the road network data.
Elasticsearch is a distributed open
source document database built on
top of Apache Lucene. It stores the
result of the workflow.

9.
Elasticsearch Cluster
L. Selmi - BDE - SC4 Webinar

10.
Pilot Architecture
L. Selmi - BDE - SC4 Webinar

11.
BDE Components
L. Selmi - BDE - SC4 Webinar

12.
The FCD Pipeline
L. Selmi - BDE - SC4 Webinar

13.
Visualization
L. Selmi - BDE - SC4 Webinar
The pilot SC4 can process
real-time FCD data for
map-matching and classify
a road segment according
to the traffic level.

14.
Distributed computing: the theoretical
minimum
L. Selmi - BDE - SC4 Webinar
Minimum requirement for fault-
tolerance and scalability
● Cluster of 3 nodes (Docker swarm)
● 4 CPU cores x node
● 1 (Flink) worker x node
● 1 (Flink) slot x CPU core
Max parallelism = 12

15.
Parallelization: map-match subtasks
L. Selmi - BDE - SC4 Webinar
1. source()
2. mapMatch()
3. keyBy()/window()/apply()
4. sink()
The subtasks can be distributed in
slots with different parallelism (e.g.
from 1 to 12)

16.
Parallelization: Flink dataflow
L. Selmi - BDE - SC4 Webinar
A slot can process all the subtasks in a
pipeline

17.
Parallelization: input and output data
L. Selmi - BDE - SC4 Webinar
device_id timestamp lat lon speed orientation transit
The mapMatch subtask keeps the time
order so that the next task
keyBy(road_seg)/window(15’)/apply() will
return the correct average speed and
number of vehicles within the time window
for each road segment.
road_seg_id start_date num_vehicles avg_speed

18.
Pilot Cycle 2 Targets
L. Selmi - BDE - SC4 Webinar
● Extend the functionalities
● Improve the technology
● Lower the boundaries

19.
Cycle 2 - Extend the functionalities
L. Selmi - BDE - SC4 Webinar
Short-term traffic forecasts
1. Map-match 44 Gb of historical
Floating Car Data from CERTH
(Thessaloniki)
2. Train a model (using ANN)
3. Make predictions using the model
and the near real-time data

20.
Cycle 2 - Improve the technology
L. Selmi - BDE - SC4 Webinar
● Improve the map-matching
algorithm
● Parallelize the processing of the
historical data
● Finalizing the “dockerization” of the
components

21.
Cycle 2 - Lower the boundaries
L. Selmi - BDE - SC4 Webinar
● Set up different visualizations for
traffic monitoring and forecasting
● Visualize the traffic pattern in a
road segment
● Visualize a location of a vehicle and
the matched road segment (for
tests)

22.
Thanks
L. Selmi - BDE - SC4 Webinar
BDE project website:
https://www.big-data-europe.eu/
Code repository:
https://github.com/big-data-europe
Contact:
luigi.selmi@iais.fraunhofer.de