Caesar works to improve internet connectivity during disaster response
Just last week, the Midwest was hit with storms that took down power lines and left tens of thousands without power for days. Getting aid to people whose homes had been destroyed or who needed medical attention, along with setting up ongoing communications, was a crucial and sometimes lacking relief effort. Illinois ECE affiliate faculty member Associate Professor Matthew Chapman Caesar’s latest research looks at how to get vital internet connectivity to disaster-stricken areas using drones.
In an event like last week's Derecho that tore through parts of the Midwest, Caesar's drone base stations would be able to get connectivity to those who needed it, even if roads were blocked like this one. Photo submitted by Angela Anthony.
“There are a lot of environments where you need internet connectivity. It may be a disaster scenario and you have emergency responders who need to help people or you may have battlefield environments where people need to conduct various operations,” said Caesar, associate professor of computer science. “A lot of people don’t know how important it is to have network connectivity in these situations. There are so many infrastructures that rely on network connectivity.”
Given that wires can’t be connected to everyone’s devices under normal circumstances, let alone in a disaster, deploying wireless communication is a necessity. However, providing coverage where it’s needed for the right amount of people is a significant challenge. Caesar and collaborators are working to make the infrastructure behind network connectivity mobile.
“Wouldn’t it be great if the infrastructure was robotic and automatically put stations where you need them?” said Caesar. “We’re building infrastructure where you can put base stations with wireless transmitters on drones and fly them out to where you need them. If there’s a big flood or hurricane, we can deploy services quickly in that area.”
Existing wireless networks, such as those used by cell phones or in wifi networks of a house, are built around stable base stations. In a crisis situation, these base stations can be knocked out and unavailable for the people who need them. Another issue with traditional base stations is their lack of flexibility when it comes to workload. If a lot of people are trying to use a traditional network at once it can become overloaded and jammed. Caesar’s design would allow for more network capacity to be added as needed.
“We came up with a design for a drone that could autonomously fly to areas where more coverage was needed,” said Caesar. “The hard problem with this is you have to figure out where you need coverage before you need it, how do you know where people are going to be?”
The team used artificial intelligence and a machine learning technique called deep learning to develop an algorithm. The drones were flown around with sensors to observe and analyze an environment. After observing patterns over time, the drones predict where to fly based on the information. If they make a correct decision and provide good coverage, they are rewarded; if they fly to the wrong area or don’t provide good coverage, they are penalized.
In addition to applications related to national disasters, there are other ways this technology can help people around the country. One of them is particularly relevant to current times.
“During the pandemic, there are a bunch of schools shut down and there are students who don’t have internet access but need to watch lectures online,” said Caesar. “This is a way we can help address the social inequity that comes out of this, where there are places with high-quality internet there are places where there isn’t’ and they can’t learn.”
Another less dire situation is when many people at large festivals or sporting events are trying to access the same network at one time. To meet this temporary demand, the drones could be flown in for a few days at a time and then move on to the next location when the need decreases.
Read the original article on the CSL site.