Space Apps Challenge 2018

As humans get closer to living on Mars, it’s becoming more and more clear that we will likely be living in what is called a Closed Ecological Life Support System (CELSS). This is a system that mimics the biosphere of the Earth, where plants, animals, and bacteria live in an equilibrium where each provides the nutrients that the others need to live. In the Space environment, the system will be designed around the nutrient requirements of the humans that inhabit it. Unfortunately, our CELSS design is too complicated to cover in the alloted presentation time, but we are happy to describe it in much greater detail after the talk.

One of other challenges in space exploration is the scarcity of human time and labor available. On the ISS, the vast majority of astronaut time is currently eaten up with station maintenance, leaving little time for science, and personal relaxation. In long duration space travel, this will lead to not only psychological issues (as seen in earlier space missions) but also the danger of stress-driven mistakes. The solution to this is to automate as much of the CELSS and other infrastructure as possible so it doesn’t need constant human monitoring.

To do this we have also designed a hierarchical, extendible and robust sensor monitoring system. To start, we have examined the scope and details of such a system for a balanced closed ecological system that would support a single human. We have created a suite of sample data and a UI for the monitoring of that data through a simple dashboard interface and provide for safe ranges of natural variation in the system, and alarms when the system begins to creep outside those safe ranges. The astronaut can easily switch between different dashboards to check on different systems, such as the Cabin Atmosphere, Growth Chambers, Bioreactors, and other Infrastructure that enables and maintains the whole system. Modern, low size, weight, mass and power sensors will enable a huge distributed system with good redundancy in data collection, allowing for the safest system possible.

However, even simply creating massive sensor network brings its own set of technical challenges. When there are hundreds or even thousands of sensors, the ability to provide power and communication to all of them becomes very challenging. Without proper design, it is easy to turn a space habitat into an unmanageable mass of wires. Wireless communication, while seeming like a simple answer, has difficulty when dealing with such large numbers of sensors. Also, many space environments place strict limitations on stray EM emissions for safety and/or scientific reasons. Also, the harsh radiation levels in space induce regular failures in commercial grade electronics, requiring redundant systems, further increasing the number of sensors and communication overload.

Our proposed solution to these problems is a highly redundant star topology network. Sensors that are mounted in a single location such as a single growth chamber are all connected to central processing hub via an addressable wired system such as CAN bus. These central nodes and attached sensors would be used in duplicate or triplicate to monitor systems so that individual hardware failures do not disrupt data collection. Where possible, the sensor nodes will utilize low power microprocessors such as the MSP430, allowing them to run off of small batteries for years without recharge. This minimizes the number of physical wires while also minimizing RF communication nodes. In cases where physical wiring is impossible, alternate wavelength bands and protocols such as LowPAN can be used to minimize conflict with high data rate and priority networks.

Each sensor hub will be a more powerful computer roughly equivalent to a Raspberry Pi. These hubs can do more computationally intensive tasks such as machine vision and machine learning algorithms to separate nominal data from data which actually require human intervention. These hubs will use high speed wired or wireless communication to a central computer system and also sending alerts to wearable devices on astronauts.

Because of the hierarchical structure of the network, it is highly scalable as a space colony grows over time and can even incorporate highly distributed sensor data from things like very distant scientific and environmental sensors.

Ultimately, this sensor network design should easily grow to match the needs of a space colony while providing a huge level of automation and safety.