Monitoring the visitors of Renkums Beekdal with new sensor technology

The Renkums Beekdal is a nice area to walk in. The Renkums Beekdal foundation asked the Science Shop to investigate the amount and distribution of visitors in this area of natural beauty.

Managers of natural areas want to know how many visitors visit their natural areas at certain time slots and what the spatial spread of the visitors is. They can use these data for maintaining the area, for promotion, for their policy on nature and recreation and in their consultation with the public, municipality, province and businesses. The problem is that they don’t have the time or budget to monitor the amount of visitors. The challenge is to deliver these data in an accurate way against low costs.

Counting visitors combining new technologies using passive infrared (PIR) sensors together with so-called LoRa has been developed for nature area Renkums Beekdal. LoRa (Long Range radio) is a specific telecomnetwork and transmits data in an accurate way and at low cost. This technique combines long range, low power consumption and secure data transmission. Although it offers a considerable potential, it also offers challenges. Thus, the optimal way to use this new technique in a nature area should be defined. First of all, sensor components had to be selected, which should be connected with the LoRa gateway. The Adafruit Feather M0 LoRa chipset was selected since it has complete package and is relatively cheap. The HC-SR501 PIR sensor also was selected as it is cheap, widely available, and allows for sensitivity tweaking on the sensor. As the sensors was placed in outdoor environment, a waterproof and vulnerable case also was created.

The Things Network (TTN) was chosen because it is an open source network. To test the data management process in order to record and save the visitors data, several steps has been done. The adafruit was prepared by connecting the TTN network with six pin using a wire bridge and also connecting it with the antenna. Then, every complete package sensor was registered in TTN network console. The most important step was programming the Adafruit using Arduino IDE since it defined the data form from the sensor. The programming also included the selected spreading factor and the battery consumption. Finally, LoRa gateway was adjusted in high sensitivity. The Adafruits were able to connect over a distance of 2.68 km from the gateway, because the amount of forest in between sensor and gateway was a limiting factor. The accuracy of the sensor was calculated for two sensors: beekdal_01, located at the “Kabouterpad” and beekdal_05, located north of the Keijenbergseweg, at a crossing near a gate and a bridge over the brook. The average accuracy of beekdal_01 was 60% and it increased to 83% when the false positives are removed. However, beekdal_05 has 2% higher than beekdal_01. It has 62% average accuracy and it only has one false positive with no visitors crossing.

Finally, three PIR-sensors worked during a period of ten days. After that, the battery had to be recharged. The average amount of visitors during those ten days was 64 visitors per day. To visualise data interactively and having the ability to show data in a map, Grafana dashboard was selected, using InfluxDB as data storage using Node-red as a programming tool for wiring together hardware devices, APIs and online services as part of the Internet of Things. All programs are opensource but rather difficult to install. However, to keep data flowing, the node-red server has to be running all the time, but this can still be improved.

The new technology is interesting for counting visitors in a nature area, but the results can be improved by buying less cheap material. New experiments with this technique.