BASMATI Mass Event Application in Karlsruhe
From 16th till 22nd of July BASMATI had its ultimate test on DAS FEST in Karlsruhe. Therefore, a crowd monitoring application was prepared for cloud deployment.
The final Use Case
The application chosen is dealing with asymmetric density – a well known issue on mass-events. This phenomeny is often caused by external effects (like weather, running-order and other events) and self-reinforcing movement of visitors. Those factors in combination cause local and temporal congestion bearing a high risk of injury and panic. It cannot be avoided by limited access and is extremely hard to perceive by a human observer. Therefore, the application used Raspberry PI based passive detection of WIFI signals coming from visitors’ mobile devices and connectes to a BASMATI-managed cloud backend. This backend should derive the estimate position of the users and visualize critical regions using heat-maps. Thereby it enables a timely response. The evaluation should show, how the application components are automatically deployed and scaled up in a scenario with fluctuating demand.
Setup and operation of the infrastructure
The setup of the infrastructure started already before the actual testing. It was extensively supported by the festival organizer – The Karlsruhe Event GmbH. To enable reliable connection a cable-based communication between the central container and the sensors was used. Therefore approximately 2 km of ethernet were laid during the festival setup. The cloud connection was established via a satellite link. Over the whole pre-festival the BASMATI members tested and refined the application. They calibrated the signal processing and trilateration engine live on DAS FEST to prepare it for the festival weekend. As in the last years peak numbers of 150.000 visitors were expected during the headliners performances. Moreover, a rain forecast during one of the headliners gigs on Saturday promised interesting results for the festival organizer.
The BASMATI test turned out to be a complete success. Only from Sunday 15:00 till 23:30 the backend received a total of about 7,6 million signals collected by the Raspberry PIs. They were transferred to the BASMATI platform and real-time processed by the managed application. BASMATI automatically adjusted its computing and memory resources to the actual demand, provided a real-time monitoring of all sensors and application components and produced a meaningful visualization. Besides proven reliability and stability the cloud-based design of the application excelled by its cost-efficiency. The main expenses resulted from components also required for any local deployment. Due to the demand-driven automated adjustment, the backend processing cost were reduced to a minimum. The data collected, which is completely anonymized will most probably be published as Open Data enabling data scientists using it for their scientific work.
We say thank you to the organizers, to the DAS FEST Team and to the visitors!