According to the team of researchers at MIT and the Universities of Boston and Maynooth in Ireland, the microprocessor can decode any type of code, regardless of its structure. The universal decoding algorithm called Guessing Random Additive Noise Decoding (GRAND), can eliminate several logical processes that are complex for a common processor.
Research led by Muriel Médard, a professor at MIT, pointed out that the chip has a way of thinking about these codes and creating a store of specific codes. “An important breakthrough with GRAND is that, through repetitive hashes, you can build new ones and add them to the end of the original data.”
As the encoded data travels through the network, it is affected by noise or power interrupting the signal, which is generated by other electronic devices. However, when this encoded data and the noise that affected them reach their destination, the algorithm manages through specific code stores to restructure the stored information.
Instead, GRAND works by guessing the noise that affected the message and uses the noise pattern to reduce the original information. The microprocessor generates a series of ordered noise sequences, subtracts them from the received data and checks if the resulting keyword is in the stored information.
Although it looks like a random process, GRAND’s algorithm actually works by means of a probability structure, where it allows the algorithm to guess what it might be. “In a way it works like a human when it tries to solve problems, look for different solutions and choose the best one,” Médard explained.
It should be noted, the device is also designed to seamlessly change the storage of original codes, as it contains two static random access memory chips. That is, one can decipher code words, meanwhile, the second can load a second codebook to change its decoding back and forth.
According to the team, they tested the GRAND processor and found that it can decode any code up to 128 bits without presenting any problems with a few microseconds of latency. Now, they expect GRAND to be able to respond to more complex problems, such as working with longer and more complex noise patterns or increasing system performance and energy savings.
