Remote Crowd Estimation

Inspiration

How often is it that us as students cannot find a study/meeting space in a building? The initial motivation behind this project to give students data of how many people there are inside a specific room so they do not have to walk all the way to the building only to find no empty space.

We expand the idea of sensing traffic movement of crowd to fast deployment of first response. We added smoke sensor which will detect the air quality level of a specific room. Imagine a situation where there's a fire inside a building - by having these devices in each room we could isolate the location of the fire from the smoke sensor and also use the WiFi sniffer to detect how many people is inside that room to be evacuated in a timely manner during critical situation.

If time permits, we would like to expand this project to a full-fledge mesh network in National Forest to detect forest fire.We would use the smoke detector to detect fire, and an RF sensor to detect if there is any trekker in those area to people who needs evacuation. This is very useful because there isn't any internet in those area - hence, providing this system could give first responder idea where to focus in combating these forest fires.

What it does

We have developed an IOS app which can track the estimate number of devices/people in a room remotely.

How we built it

We have decided to use the number of unique devices connected to WiFi and Air Quality sensor as metrics to estimate the number of people in an area/room. To do this, we have used ESP8266 WiFi chips to operate in sniffing mode to monitor the WiFi traffic and look at the number of different MAC addresses. The MQ-135 air quality sensor is used to estimate how much carbon dioxide and methane gas is in an area. We compile these data and send it through Xbee modules (which utilizes Zigbee for mesh networks) to a coordinator on Raspberry Pi, which then uploads these data to a server where the IOS app pulls from.

Challenges we ran into

Before finally deciding to settle on using the Zigbee protocol for a mesh network with Xbee modules, we spent a lot of time figuring out what is the best way to upload these data to a server. Before using Zigbee, we tried using the ESP8266 chips to form a star network with MQTT protocol, however, we ran into issues trying to operate it with University WiFi. After that, we also considered having each ESP8266 modules sending data to a cloud server without sending it to a central hub (Raspberry Pi). We should be able to achieve it using AWS IOT which utilizes the MQTT protocol. Sadly, because of our short of time and the fact that ESP8266 has few well documented open-source AWS libraries, we decided to move on with a more feasible implementation for this timeline. One of the other challenges we faced was the fact that IllinoisNet WiFi mostly runs in the 5G band, while IOT devices are made to only operate in the 2.4G band. This made estimations of devices less accurate.