Introduction
To implement short-term overtourism mitigation actions, crowding information should be made available in near real-time. Several approaches can be used for crowd detection, but it has long been observed that the number of detected mobile devices is highly correlated to the real number of people present in an area. As such, the best option regarding cost, precision, and the near-real-time availability of data required for managing tourism crowding effectively, while complying with privacy rights, is then the one based on sensing wireless communication traces in roaming devices, since most tourists carry a mobile phone to take pictures and record videos during their visits. This tool is supported on strategically deployed crowding sensors that performs real-time detection of trace elements generated by mobile devices from different wireless technologies, namely Wi-Fi and Bluetooth, through edge computing to determine the number of mobile devices in the sensors’ vicinity. Therefore, only the number of devices detected is sent to a cloud server, allowing the upload with LoRaWAN (Long Range Wide Area Network) in the absence of local Wi-Fi coverage, while granting user privacy since no personal information is sent to the server. This STT uses a variety of open-source software technologies, installed in off-the-shelf hardware (a Raspberry Pi board) available at affordable costs. For the operating system a Kali Linux distribution has been adopted. For the local database, a SQLite database was chosen for storing all gathered anonymised data, which requires low memory usage, a convenient feature for edge computing. For performing Wi-Fi detection, the required hardware is a Wi-Fi card that supports monitor mode, which allows the board to capture all network traffic in its proximity. The Alfa Network AWU036AC board provides high performance at a low cost, having two antennas for dual-band detection (2.4 GHz and 5 GHz). As for the sniffing software, the Aircrack-ng tool has been used, an open-source software with several different applications for detecting devices. For receiving all messages, the cloud server uses the Mosquitto MQTT, a lightweight message broker that implements the MQTT protocol. For the data ingestion of all measurements sent by sensors, a database is necessary, and it has to be lightweight, capable of querying data rapidly from timestamps, and also capable of providing support for data visualization platforms to observe the results in real time. The InfluxDB, a time-series database focused on IoT applications, fulfils these requirements. The counting accuracy was addressed in a more controlled environment, where the detections from sensors were compared with the real number of people, obtained through direct observation during public events. Furthermore, this STT has been validated over one year, with sensors placed at several indoor and outdoor spots, such as areas with a large pedestrian flow, internal and external passages between buildings, and places for prolonged stays. Crowding information allowed to identify crowding patterns and tendencies, and supported decision making. Furthermore, it was also possible to create notification policies, where alerts can be triggered if predetermined crowding thresholds are exceeded, by using several contact points such as e-mail, Telegram, Google Chat, Microsoft Teams, Slack, or PaperDuty, enabling users to make just-in-time decisions facing overtourism situations. These alerts can be easily configurable by using the Grafana tool, also used for 2D spatiotemporal visualization of crowding information.
Primary Goals
This task is aimed at tourism SMEs or other organizations that deal with potentially excessive tourism crowding levels, to allow optimizing their delivery and quality of service. It provides guidance on how to build a STT to monitor crowding levels in real time and/or estimate the outflow period. Either in closed space cultural or religious tourism scenarios or in open air events for tourists, it is crucial to monitor the occupation of the space, either to ensure a better visiting experience, for security reasons, health reasons, or simply because of workforce management considerations.
Key Features
This tool performs overtourism monitoring in real-time, based on a set of sensors, built with off-the-shelf hardware available at affordable costs, that performs detection of trace elements of mobile devices’ wireless activity, mitigating address randomization, while preserving privacy. Detected crowding values are put together in a cloud server either via Wi-Fi or LoRaWAN, depending on local conditions and availability. The crowding information can then be analyzed by destination managers to understand the crowding levels in areas where each sensor is placed in a clear and simple perspective, either by dashboards for temporal and spatial visualization of crowding information or using the raw data for custom-made integration. In addition, notification policies can be created when over-tourism situations occur, providing the ability to implement just-in-time mitigation actions required by the nature of these often sudden and unpredictable circumstances.
Who can use and benefit from tool?
This toolkit aims to scaffold the work of destination managers to identify areas of overtourism concern and implement measures to mitigate the negative impacts of tourism while promoting smarter tourism practices. By performing local monitoring of tourism crowding, an it’s ability to obtain near real-time data, allows short-term decision-making, such as reinforcing or freeing up human and technical resources assigned to managing tourist crowding on the ground. This can help ensure that tourism benefits both visitors and residents while preserving the natural and cultural resources that make destinations appealing. Mitigating overtourism with this tool benefits many stakeholders, in the following ways, such as: – Heritage managers – which can elude heritage degradation more effectively, thus retaining the authenticity of destinations; – Local authorities – with improved services by making just-in-time decisions and planning more effectively urban cleaning and public safety routines, as well as reducing operating costs; – Local businesses – with increased share of tourism income; – Tourists – with increased visit satisfaction, with fewer delays, and can provide more guarantees of safety and cleanliness; – Tourism specialists – which can speed up service delivery and quality of service; – Local residents – with reduced stress from over-occupation of personal space and privacy, as well as improved attitude towards tourists and tourism professionals.
How Can I use it?
SMEs or other entities can make use of this tool to enable overtourism mitigation actions. Several practical example can be provided, such as promoting the visitation to less occupied but equally attractive areas, applied in recreational, cultural, or religious spots, both in indoor scenarios like palaces, museums, monasteries, or cathedrals, or in outdoor ones such as public parks, camping parks, concerts, fireworks, or video mapping shows. Besides assuring a better visiting experience (e.g. by inviting tourists to less occupied areas / rooms), these actions are also necessary for security reasons (e.g., to prevent works exhibited in a museum from deteriorating or even being vandalized by exceeding room capacity), health reasons (e.g., preventing infection in pandemic scenarios by not exceeding the maximum people density specified by health authorities), or even for resource management (e.g. to reduce the intervention of security and cleaning teams).
The tool is accesible from this link – https://sites.google.com/iscte-iul.pt/resetting-project/home/smart-tourism-toolkits/Crowd_Monitoring_STToolkit
Why use this tool/application instead of something else that already exists?
There are other technologies available for crowd detection and monitoring, but they present several disadvantages. To detect the number of persons in a given area it is also possible to use: – sound-based approach’s – but precision and detection ranges achieved are low, and present privacy issues; – vision-based approach’s – high cost due to hardware and computationally demanding tasks that are not affordable for edge computing, and also present privacy issues; – use social media content – since geo-location information needs to be tagged in people’s posts, only a portion of this information can be accessed and utilized, making this approach less representative of the actual crowding level; – use metadata provided by mobile operator’s – this method is focused on retrieving crowding patterns and predictions from past gathered data, and not in real-time, and usually with a not detailed special resolution. The use of mobile data is usually the most mature and available alternative, but operator’s willing to monetize network data, at a high cost, may render this last approach unfeasible for many SMEs, without providing a detailed and in real-time solution.