MIT's New Theory Could Bring More Reliable Communication Protocols

The mechanism of virtual communication is primarily based on the communication protocols implemented in the services. The protocols currently in use are efficient but requires precise information to be provided in limited bits. A small error may lead to failure in entire communication. Massachusetts Institute of Technology (MIT) researchers led by an adjunct professor of Electrical Engineering and Computer Science Madhu Sudan who is also a principal researcher at Microsoft Research New England wants to bring the human ability of flexible communication to computers.


In few recent papers Madhu Sudan with his colleagues described the theoretical limits to the degree of imprecision that communicating computers can tolerate. He explains what is the kind of problem that human communication can to solve but not the machines. The research is aimed at coming up with such abilities in communication schemes. The factor that the team is investigating is compression. An earlier paper, in 2011, he considered a hypothetical case where devices in communication shared an almost infinite codebook that assigned a random string of symbols (like a serial number) to every possible message that devices might send.

This experiment helped the researchers to get statistics on the problem of compression. The question here is the number of symbols that a device need to send to other to identify a particular message. The answer is obtained from probability and interpretation of string of symbols. The team was able to show that good amount of compression is still possible as long as protocol designers can make assumptions on the ranges of probability. They also generalized their model to new context. The team lead explained that performing checksum can help in cloud computing models where huge data duplication exists.

Source: #-Link-Snipped-#