VANETs and How can a vehicle chat to another one?

Me: "Can my car talk to my neighbors car?"
Tonimobile: Yes, I can! (on a robotic vintage voice)
Me: "WHAT?"

In a distant future, something like this dialog would be possible, but on a realistic near present, at least exchange some information between vehicles is a valid way of thinking.

On this days, where anything between a refrigerator and a drone have wireless access to networks like WiFi, WiMAX, 3G, 4G, GPS and others, thinking to have networks connected in cars, trucks and buses aren't absurd anymore. Think if you are on a highway, travelling, and on your car panel, information like weather forecasting, real-time news about your path and destiny, traffic mapping and many others. That would be great right? RIGHT! And more, if your car understand the information about an accident that occurred a few meters from you and prevent you to engage on that destructive and dangerous perimeter. Wow!

Someone could say: "Yeah, but with my iPhone I can get better things!". Of course you can, but using an iPhone driving is prohibited and dangerous. A clear LCD panel on the vehicle console would be a valid solution.

Another one could say: "OK, if I fix my tablet with one of those "tablet hangers" that stick to the glass is sufficient". Agreed, but Internet is not a structure available on each street, unless you have a satellite connection. Besides this, imagine the structure that a network of this kind, that covers all streets in your country would be (lot of antennas and cables mixed to the beautiful landscape of the mountains).
And this two "alternatives" still can't prevent you to crash your car in a truck.

That's where the concept of VANETs (Vehicle Ad-hoc Networks) lies. Use the cars, trucks and buses as the main transmission media. If every vehicle have an interface and a memory buffer, why not? Someone takes information from another vehicle, and pass to the next a few kilometers away. Simple? Conceptually, YES! On a real-word view? Definitely NO!

There are 2 basic modalities of communication on VANETs:

V2I - Vehicle to Infrastructure - Vehicle, this is Antenna. Antenna, this is Vehicle. Pretty simple. You can view it in a funny way, looking to a car as a huge smartphone with wheels. There is more complicated problems related, but is basically the same concept.

V2V - Vehicle to Vehicle - This is where the things get more interesting. Vehicle this ........ Vehicle th.... All right. How can I communicate with another vehicle and establish a solid link to pass my information timely, before I get miles and miles away. Now you're starting to see where's the problem.

There are many studies of the area, for great content distribution networks relying on car mobility models, simulators for vehicle network behavior, accident prevention, traffic monitoring, security and privacy issues..

How can I represent a network like this?

This VANETs are complex dynamic networks that varies during time. Let's imagine describing a network with circles and lines. Each circle is a car, each line represents a connection between cars. For a specific moment, there will be a set of circles related by lines. In another moment some of the cars could be turned off, and the lines that linked those cars will disappear. It's a random phenomena that can be described in a temporal way.

That's where the concept of Time-Varying Graphs fits. A graph is a mathematical formalism that allows to describe relations between thins. Avoiding a little bit the math things, let's rethink our problem.

Now we will name the circles as nodes, and the lines as edges. We have a very simple graph! Inserting information on the nodes and edges can describe a phenomenon like a network. On the example below, imagine 10 cars that can communicate with each other, and each connection takes the time that is represented to reach the other car. On "graph therms" 10 nodes, linked by weighted edges with integer values.

 

 

This representation is intrinsically static (does not change during time). How can we make it change during time? With some mathematic armory we introduce some parameters that allows us to change those edges over a variable (on our case, time).

On a simple way, each time will have a different "on demand" representation, according to the situation that the network is. Just imagine making a lot of graphs for each minute of the network and stacking all of those together, giving just a post-it with the timetag to another person search to a specific moment. That is pretty much what a time varying graph can represent in really simple therms.

Now we can describe a time varying chaos with a graph =]. This tool could solve many problems on mobility models if used adequately (specially one, the introduction of accidents on vehicle networks simulation - theme of my graduation work).

Of course there is a lot more to talk about it, but for a introductory post I will stop here. More posts on the subject coming soon.

Hope you enjoyed!