Forest fire has become one of the big challenges in recent days. Even with the highly evolved technology, we are struggling to avoid forest fire. Forest fire is causing plenty of loss to the vegetation and animals habitats.
Internet Of Trees is an FPGA based system, which shall be used for collect the data of environmental conditions, early fire detection, animal conservation, tree conservation in the forest.
CNN Algorithm will be running on the FPGA board, which will process the image to identify the fire.
when the fire is detected, the auxiliary circuit of the smoke detection/Co2 level sensor will be used to confirm the detection is not false detection.
This camera-based system also will be used to monitor the animal movements, image-based tree statistics, rain data. Using analog sensor boards we can detect the environmental conditions.
Microsoft Azure will be used to deploy analytics algorithm and cloud software. Using Microsoft Azure we also deploy warning systems to firefighters, Real-time video monitoring systems on the cloud.
By deploying CNN-based sensor applications using FPGA accelerates the detection speed, Accuracy of the detection, and ease of connection to the cloud.

Demo Video

[URL: https://youtu.be/c6WrtN2sn3Y]

Project Proposal

1. High-level project introduction and performance expectation

Purpose of the design

Forests are vital to life on Earth. They purify the air we breathe, filter the water we drink, prevent erosion, and act as an important buffer against climate change. According to statistics in the US, in the year 2020, there were 58,950 wildfires compared with 50,477 in 2019, according to the National Interagency Fire Center. About 10.1 million acres were burned in 2020, compared with 4.7 million acres in 2019. So it is our responsibility to save the forest from a forest fire.

The main purpose of the proposed FPGA-based Internet Of Trees design is to save the forest from a forest fire, illegal tree cutting, monitor the environmental conditions of the forest to take necessary measures to save the forest.

Application scope

This design will be used in the Forests, Farms to protect the trees, crops from fire. The camera in the design will be used to collect the image data of forests or Farms, CNN Algorithm will be running on the FPGA board, which will process the image to identify the fire, illegal human entry into the forest, tree counts, etc. When the fire is detected, the auxiliary circuit of the smoke detection/Co2 level sensor will be used to confirm the detection is not false detection. There is a crowd blockchain-based authentication using the Azure cloud to double-check the detection. Microsoft Azure will be used to deploy analytics algorithms and cloud software. Using Microsoft Azure we also deploy warning systems to firefighters, Real-time video monitoring systems on the cloud.

Targeted users

Private Farms, Federal Governments, mining agencies

2. Block Diagram

Top-Level Block Diagram

Functional Block Diagram

3. Expected sustainability results, projected resource savings

Expected Sustainability

An intelligent forest management system will save the natural inhabitants of plants and animals, which intern will contribute to energy conservation, global warming, natural calamities.

4. Design Introduction

Purpose of the design

Application scope

5. Functional description and implementation

Hardware Description

DE10-Nano Cyclone V SoC FPGA Board

Intel Cyclone V based SoC board is the heart of "Internet of forest system" which includes HPS based system which will help our project to integrate IoT based stacks and in the FPGA fabric of cyclone V we are implementing WiFi Control Module, Ethernet Control Module, Fire Identification & Tree, Animal detection Algorithm State Machine, Convolution Neural Network, D8M Camera Control, and GAS Detection FSM.

ESP-WROOM-02

ESP-WROOM-02 has inbuilt a Tensilica L106 32-bit RISC processor with a wireless stack inbuilt. This will be used to interface the "Internet of Tree FPGA" system to the Microsoft Azure cloud. It communicates to the FPGA state machines via the SPI interface.

EVAL-CN0357-ARDZ - Electrochemical GAS detection

EVAL-CN0357-ARDZ is an Electrochemical gas detection sensor used to detect toxic gases. this hardware is used as a confirmation system of a forest fire. CNN will do preliminary forest fire detection based on an image processing algorithm. This sensor will be integrated into the FPGA board using the SPI interface.

Tera ASIC D8M Camera

D8M Camera is an 8MP camera with a GPIO interface. The images from this sensor are fed to the CNN algorithm for forest fire detection.

Local PC/ Monitor GUI

Through the UART interface, DE-10 interface to a local monitoring system for the video monitoring, status monitoring when the system installed in the control centers, otherwise trough azure same can be monitored.

Solar power for the remote deployment

Since this device will be mostly deployed in the forest or remote area, a Solar power system, which is sufficient to cater to the power requirement of the FPGA and sensor boards will be attached.

Software description

Convolution Neural Network

A convolution neural network is an image processing algorithm that is used to identify the various features of the image. Since this algorithm will be implemented on the FPGA there is an added advantage of parallel implementation and faster computation. In our system, this will help to identify the forest fire, Animals, Human, tree counts.

Azure IoT

Azure IoT module is responsible for the conversion of the processed data to Azure IoT compatible packets. This will also include authentication, alert mechanism, and database read-write state machines.

Ethernet Control Module

An alternate way to connect the board to Azure IoT is via ethernet. Using the SGMII interface the board shall be connected to Azure IoT. We will be using Mega Wizard IP for implementing this.

Fire Identification & Tree, Animal detection Algorithm State Machine

This module takes the inputs from the CNN and does the Fire identification, Tree, Animal detection. Also generates alerts for data for the Azure IoT.

D8M Camera Control

This module takes care of reading the data from the camera, writing to FIFO, and generation of the camera control signals.

GAS Detection FSM

This module communicates to the GAS sensor using the SPI interface. It is also responsible for calculating the chemical composition of the environment and gives out the data to the fire detection algorithm.

WiFi Control Module

The wifi control module converts the messages to be transmitted to the cloud to suitable formats, and send to ESP-WROOM-02, via the SPI interface. This module is also responsible for configuring the ESP-WROOM-02

6. Performance metrics, performance to expectation

Performance Metrics

Area of foreset saved by one unit of "Internet of Trees".
Number of Animals saved by "Internet of Trees"
Number of trees saved by "Internet of Trees"
Speed of fire detection by CNN.
Power consumption has to be very low since this deployment will be in the forest and the system is powered by solar.

7. Sustainability results, resource savings achieved

Sustainability Results

This system will save the large area of forest which is affected by forest fire, illegal tree cutting, and animal hunting

FPGA based Internet Of Trees for Smart Forest