With climate change showing real and measurable effect on our daily lives with increase in extreme natural calamities, planning for a sustainable future has become a nessasity. To solve a problem of this scale, one needs to understand it fully first. To understand the extent of our effect on a climate a easy to use and distribute measuring units will be crucial. We will be tacking this problem.
A end-to-end Pollution Detection system with Machine Learning Predictions will be designed. It will consists of two major part. Client-Side Pollution Box which will be compact and fully integrated with various sensors and modules to detect Air, Water, Light, Noise Pollution Parameters and Server-Side Cloud Processing Center which will use ML based systems to find patterns and correlations between various pollution parameter and how they are affected with weather conditions. The project will provide a easy to replicate system which can be used by concerned authorities in the Metropolitan areas to monitor and curb pollution on a real-time basis.
Pollution Box will contain a Camera, Microphone, Air Quality and Gas Sensor, Water TDS and Turbidity sensor, which will be used to create a Holistic Pollution Parameter which will be used to build a fully contained pollution index.
Camera will be used to scan the night sky and provide a light pollution metric which can be used to further plan the areas street light consumption. Microphone will be used to provide a all-day look at the noise pollution and can issue mental health warning if exceeding the researched paramter. Air Quality and Gas Sensors can be used to provide a accurate Air Quality Index. Whereas resistivity based TDS calculation can identify increase in toxicity and salinty of water on a real-time basis.
Making the Pollution box compact and easy to use will make it easier for the authorities to make a mesh of these IoT enabled boxes that can give a better resolution in detecting the problem areas and solving the issue at a larger scale and thus ensuring the sustainble future.
At the server end the data will be processed and Machine Learning based program will find patterns and correlation between different parameters to further our understanding on the effects of pollution.

Project Proposal

1. High-level project introduction and performance expectation

With Climate change showing real and measurable effects on our daily life, it has become necessary to bring about a positive change and fast. Most countries failed to meet their Paris 2020 Accords. Each day we wait we are making the problem much bigger and harder to solve. It has now become crucial to change our unsustainable practices fast. But we are not very good at solving the problems we don't understand. To understand the extent of climate change we need to build much better infrastructure to measure our effects on the climate and how these effects are correlated with many larger climatic factors. Our project for this year's Innovate FPGA Challenge is thus related to providing that infrastructure. The modern civilization is dynamic and ever-changing so is our effect on the surrounding environment. But our most of our cities lack the measurement units to quantify the pollution we are causing to pinpoint causes and understand the pattern on a large scale. Most of them have less than 5 Pollution Measurement centers throughout, that too are focused only on air pollution. What we need is a compact and scalable system to increase this data collection to inform the authorities of the patterns thus helping them make better policy decisions. Our proposal is to design a fully integrated and compact pollution box, which will have sensors capable of measuring various parameters for all major types of pollutions such as Air, Water, Light, and Noise pollution and Cloud infrastructure using ML and AI systems to predict and find patterns and correlation between these parameters and natural and sociological conditions. The proposed system will contain two major parts:

1. Pollution Box (Client-Side): A compact IoT-enabled pollution detecting center that can be distributed to the normal public which can measure continuously and provide real-time metrics to the central pollution server. The pollution box will contain the following sensors to measure said parameters:

• Camera: will be used to observe the night sky to calculate visible stars and ambient light to calculate Light pollution measurement.
• Microphone: will be used to detect sounds throughout the day to calculate noise pollution levels.
• MQ135: This will be used to detect ammonia, sulphur and smoke in the air.
• PM2.5 Sensor: This will be used to measure particulate matter in the air.
• Turbidity Sensor: It will measure dispersion of Infrared light through the water, which is directly related to the impurities in the water.
• Using electrodes through the water and measuring resistivity, we can calculate TDS of the water.

It will also contain:

• Wifi module: To send the data to the server
• Temperature Sensors
• Humidity Sensors

2. Cloud Processing Center (Server-side): A central processing server that collects data from large area of pollution boxes and complies the data. The data is used by ML System to find patterns and form correlations which can be published as open-access to the authorities.

This system will ensure a large amount of data collection at a much better resolution, hence helping in finding problem areas and fixing them before they become major health and environmental hazards. The targeted users of this system will be the normal citizens on a locality level or city level. The data could be used by the Government authorities or NGOs to better understand the real-time situation. 

Using Intel FPGAs to design these pollution boxes will give them better processing power to interface and processes data from various sensors and pre-process the data on real-time on the client-side to ensure real-time nature of the data flow. Also, the system can later be designed for its own ASIC to mass-produce and bring down the costs by designing a PCB for it. Using FPGA gives the flexibility for better and more tweakable design. Intel FPGAs using Quartus Software are easy to work with and providing support for cloud connectivity is also a plus point.

2. Block Diagram

3. Expected sustainability results, projected resource savings

Performance Parameters for the successful development of the system is as follows:

• It should be able to provide valid and calibrated parameters for all the mentioned sensors.
• It should process the image from the camera to find the correct parameters for light pollution.
• It should be able to upload 24x7 real-time data to the cloud server at fixed intervals.
• It should be compact and easy to distribute and ship.
• The container must be able to work without any external charging power in normal sunny day circumstances. 
• It should be able to collect useful data to further our cause of a sustainable future.

4. Design Introduction

As seen in the block diagram the Pollution Box will be a container with sensors fixed for easy use for the general public. The sensors will be calibrated using easy processes that can be replicated when needed. The Pollution Box will not need any external power in normal circumstances (reasonably sunny days) due to reliance on solar power. The Box will however need an external wifi access point to provide access to the central cloud server.  

The major part of the system we are designing is the ML algorithm on the cloud server-side. The data is not useful if not corrected and studied in the right way. We will design a Machine learning-based algorithm to find patterns and provide some insights into the correlation between the different pollution parameters. The server will get the data using a mesh of pollution boxes throughout the area. This can be used to make a heat map of sorts to find problematic areas and find solutions to solve them as possible. This combined database of pollution parameters can be very useful for academic as well as practical purposes. It can help us better understand our effect on this planet while also providing real insights so ground-level change can be brought and climate-change deniers in our political climate can be countered. 

5. Functional description and implementation

The Intel FPGA will be used with the Cloud connectivity kit and IoT cloud infrastructure to make this project successful. The sensors will be added separately as needed so will be the case enclosure. The case can be 3D printed or a 3D model could be constructed to make replication easier. 

The connection will be made as shown in the block diagram, the cutouts for the control buttons and LCD will be made and solar panels will be fixed on the top side of the container. Cutouts for exhaust and intake fans will also be made in this stage. The water turbidity sensing interfaces will also need some cutouts. 

After the case is ready the modules will be fixed according to the block diagram. 

And after testing the client-side will be completed.

Simultaneously work will be done to improve and develop the Machine learning capabilities of the cloud server-side. 

6. Performance metrics, performance to expectation

The design must pass the following performance parameters :

• 24hr continuous data monitoring.
• A fully contained module must be designed. It should be so simple that normal citizens can use it without any hassle.
• Must run 24hr without any external charging (sunny day)
• Onboard collection and processing of real-time images from the camera must be done.

Intel FPGA will be perfect for these jobs due to its fast processing capabilities and brilliant Quartus Software Suite. The familiarity with the intel platform makes it easier to prototype and test the design faster. Also excellent support of intel fpga forums available is also a major factor. With the FPGA cloud connectivity kit this idea can be realized more efficiently than using some general microprocessor SOC for this application.

7. Sustainability results, resource savings achieved

This system is fully sustainable and can in later stages manufactured using low-cost recycled plastic. It would give the authorities to make better and efficient policies to curb pollution and our effect on this climate. This would help give clear and presentable metrics which will help the general public to understand the extent of the problem much better. 
Having a heat map of sort of all areas in a locality and a holistic pollution index combining all Air, Water, Light, and Noise pollution will give a much better big picture of the pollution in the area. Focusing just on air pollution neglects the effect we are causing elsewhere.

8. Conclusion

It's time to take some action. We must not make the same mistakes we are making for the last 60 years. Unsustainable Fossil fuels have made a huge impact on our development in the last 80 years. But at a much larger scale, this impacted the climate in much more ways than we can imagine. We must act now. And decide our pathway to a sustainable future.

This project is a small step in that direction. Using the technologies for a step towards a sustainable future. Understanding is the best way to solve a problem and when problems are at the scale of our whole climate, understanding it will require a much more dynamic and holistic approach. This system gives a way to implement a solution to increase that understanding many folds.