If you want a faster smartphone, you just have to buy a new one (or play around with ROMs or settings). However, if you want a faster Internet connection you are at the mercy of the operator; If you’ve decided it’s profitable, you may want to install fiber optics, but if not you really can’t do much more. So lately we are seeing developments focused on take advantage of lines that are already available instead of adding new ones; This is the case of the record speed in copper cables that we discussed not long ago.
Improving the veteran TCP / IP
Today it is time to talk about a new technique that does not focus on improving cable transmission, but on improve the protocol with which most data is transmitted on the Internet: TCP / IP. Developed by researchers at MIT, Aalborg University and Caltech, it focuses on improving one of the common bottlenecks in any connection: the nodes through which the information passes. When we send data through the network to an external server, we do it in the form of TCP packets in a very concrete and detailed way; Since the 1970s we have used this protocol, a necessary standard for everyone to speak the same language.
Despite the fact that it is still very reliable and useful, it is beginning to show its age in some details in which it is too strict. For example, the obligation that once a connection is made between two machines, TCP packets will always follow the same path; in the same way, it is mandatory that the recipient receives the packages in the same order in which they were sent. Therefore, nodes between transmitter and receiver can only receive packets, save them, and send them back without modifying them.
This method has many positive points, but also other negative ones depending on the situation. For example, if an attacker manages to listen to the transmission, it will receive the packets in order and it will be very easy to read them. The solution proposed by the researchers is to abandon this concept for a more current one, in which nodes are smart and able to create packages, send them another way, or even recode them. The researchers tested this new protocol, and found an improvement of 5 to 10 times the speed of the standard protocol.
To ensure safe packet delivery, nodes create first-degree equations depending on the packet content (which is treated as a number) and random numbers; Each package includes a header with the coefficients, and with the unknown variable (x) representing the content of the packages. Thus when the recipient receives all the packets, all he has to do is add the coefficients and solve the equation to conclude that they belong to the same transmission; So it doesn’t matter when you receive them, since it is the equation that guarantees that they belong to the same data set, not the fact that they have arrived behind each other as usual. This prevents an attacker from easily reading some of the information by capturing certain packets, as he is forced to capture all of them; something that will be more difficult now since the packages may go different ways.
The creators of this protocol hope that it will matter in the future of the Internet of Things, 5G connections, and network-scattered storage. They intend to make it completely accessible, although its implementation will depend on the numerous patents that regulate the current market.
Source | Aalborg University | Gizmag