New System Can Decode Any Data Without the Need for Specific Hardware

Data sent over a network, including email and online gaming, can be altered by noise generated from sources such as electromagnetic interference, magnetism, and even Bluetooth devices. That information is coded and gets decoded on the receiving end using an algorithm to remove the adverse effects and restore them to their original form. Most of those error-correcting codes and decoding algorithms have been designed together and often require dedicated hardware to handle those particular codes.

Engineers from MITand Maynooth University (Ireland) have developed a silicon chip capable of decoding any code regardless of its structure with incredible accuracy, using an algorithm known as GRAND (Guessing Random Additive Noise Decoding). The breakthrough could increase efficiency for AR/VR, 5G, the IoT, gaming, and other connected devices that use a high volume of data.

Think of codes as redundant hashes, in this case, 1s and 0s, which are added to the end of the original data. Then visualize that those rules for the creation of those hashes are stored in a specific codebook. As the original data travels over a network, it encounters noise that disrupts that data, altering it from the original. As soon as it arrives at its destination, a decoding algorithm consults its codebook and uses the hash structure to guess the original information, which is then rebuilt.

GRAND works similarly but instead works by guessing the type of noise pattern that affected the data. The system generates a series of noise sequences in the order they are likely to occur, removes them from the original data, then checks to see if the resulting code is in the codebook. While GRAND is a significant first step in removing noise, the team plans to retool the chip to handle soft detections (less precise data) and tackle more complex codes.