Aarhus Transport Planning Gets a Boost with GLayer's Fast Processing and Analysis of Big Traffic Data

Glayer is a GPU-accelerated backend software designed for fast aggregation, filtering and visualisation of spatial data. GLayer is capable of performing analytical queries on large datasets with millions of data points in a matter of milliseconds.

This performance is made possible by leveraging thousands of lightweight processing cores of a GPU unit. Upon entering GLayer, a dataset is split into smaller parts. Because each is handled simultaneously by one of many GPU cores, the aggregation-based analytics can be performed incredibly fast. Even with the standard GPU hardware, like the one used in laptops, it takes less than 100ms to process tens of millions of records.

In addition to fast response time, GLayer provides graphically-rich output to aid visual analysis of processed data, in the form of dashboards and heatmaps. These front-end tools can serve a variety of purposes, from viewing real-time traffic to understanding where and when most road accidents happen. 

In BIPED, GLayer is used to support geospatial analysis of traffic data. One use case involves analysing data on average speeds in Aarhus obtained from TomTom.

Aarhus traffic use case

Aarhus wants to become climate-neutral by 2030. To achieve this goal, the city needs to offset its current footprint of about 1.3 million tons of CO2e, of which half comes from transportation. A macroscopic traffic model is currently being created as part of the BIPED project to provide scenario modelling for decarbonisation-oriented planning measures.

The model relies on several inputs. One is the manually enhanced traffic network from OpenStreetMap. Another is the origin-destination (OD) matrix, which was developed using big traffic data from TomTom, such as data from SIM cards and transportation records.

Information on average speeds and travel times was extracted from TomTom reports for selected road segments and time periods (15-21 January 2023, 15-21 April 2023, 15-21 July 2023, 30 October 2023 - 6 November 2023). The aim was to provide a representative picture of the traffic situation in the city during a full week (i.e. without public holidays) in each season (winter, spring, summer, autumn).

For each day of the week, a full traffic report covering all the main road segments in Aarhus was generated, with data points aggregated on an hourly basis. Spatio-temporal data was then parsed via a custom configurable script to enable its mapping onto a corresponding database entry.

Policy makers can access this data via GLayer’s interface without writing any code. They can filter data according to a desired period or day of the week to get aggregated insights for several or all road segments or view the traffic situation at a particular point in time in a single segment.

A Glayer dashboard showing traffic-speed variation in Aarhus

With the current data from TomTom, the traffic data from which the OD matrix was created covers the city’s traffic only partially (15-25%). Additional data acquisition is planned to further improve the model and provide accurate insights into city-wide traffic patterns that can be used as a basis for cross-sectoral ‘what-if’ simulations .

GLayer architecture

The GLayer Server software architecture is designed as a distributed, modular system that integrates diverse data sources to create and manage a high-performance, GPU-accelerated index (see the diagram below). Optimised for scalability, this index draws on parallel processing capabilities to handle large-scale datasets efficiently, and can operate either as a single node or in a multi-node configuration.

GLayer server architecture

At its core, the GLayer system features a robust data-ingestion layer that normalises and preprocesses incoming data from various sources, such as traditional relational databases via JDBC, NoSQL databases, with information then fed into the GPU-based indexing engine.

The project management user interface (UI) enables users to manage data sources, configure indexing parameters, and coordinate collaborative workflows seamlessly. Specifically, it serves three main purposes:

• Datastore configuration: data connectors can access data from permanent storages such as SQL databases (local or remote) or csv files

• Project configuration: a project can define how the dataset is going to be visualised and filtered. This includes data type definitions, aggregation strategies, filtering capabilities and cartographical outputs such as choropleth maps, heat maps, histograms etc.

• Project visualisation: The system includes a web-based visualisation client, providing users with an intuitive interface for exploring indexed data through dynamic, interactive visualisations.

The overall system is built with a microservices approach to ensure modularity, fault tolerance, and scalability, while the frontend takes advantage of modern web technologies for a responsive and user-friendly experience.

Besides the GUI frontend, GLayer supports integration with other systems via an Open Rest API. The API offers extended capabilities for accessing third-party data without having to alter the data source or reconfigure the underlying project.

GLayer documentation