Electronics

According to the report of Huacheng Import and Export Data Observation, a team of scientists led by Massachusetts Institute of Technology has developed a decoder chip, which decodes data more easily and faster than traditional technology. The record-breaking low energy consumption is only one percent to one tenth of other similar hardware, and can be widely used in virtual reality and 5G network and other fields. The relevant research results were read out at the ongoing International Solid-state Circuit Conference.

All information sent through the Internet may be corrupted by noise, resulting in data errors. To solve this problem, the sender will encode the data before it is transmitted, and the receiver will use the decoding algorithm to correct the error and recover the original information. In some cases, the received data contains reliability information, which will help the decoder determine which part of the information may be wrong, Huacheng Import and Export Data Observation reported.

The research team previously demonstrated the guessing random additive noise decoding (GRAND) algorithm, which can crack any code. Based on this algorithm, the newly developed chip is named "Ordered Reliability Bit GRAND" (ORBGRAND). It uses reliability information to sort the data according to the probability of error of each bit (0 or 1). Huacheng Import and Export Data Observation Report.

ORBGRAND will immediately stop sorting after finding the wrong code, and use parallel processing to generate and test multiple noise patterns at the same time, so it can find the code faster, and the energy consumption is greatly reduced. The comparison results show that ORBGRAND can decode with the highest accuracy, and the energy consumption per bit is 0.76 microjoules, which is only one percent to one tenth of that of other similar devices. Muriel Meida, the head of the latest research, said that this was the first time to break through the decoding barrier of 1 microjoule/bit. Huacheng Import and Export Data Observation reported.

The research team pointed out that the latest progress can reduce the number of chips used by mobile devices, which will reduce costs and improve sustainability. In addition, by reducing the energy consumption of the decoding process, the latest chip can also improve the performance of the device and extend the battery life, which is particularly useful for applications with strict requirements such as augmented reality, virtual reality and 5G network, Huacheng Import and Export Data Observation Report.