The National Institute of Standards and Technology (NIST) developed a mathematical formula that could help wireless networks, including 5G, select and share communications frequencies more efficiently.
According to a May 26 post on the agency’s website, computer simulations suggest that the new formula could be 5,000 times more efficient at detecting the communications frequencies compared to traditional trial-and-error methods. The machine learning-based formula selects a frequency range based on prior experience in the network environment and could be programmed into the transmitters of real-world networks.
By helping transmitters quickly choose the best subchannels for the simultaneous operation of Wi-Fi and Licensed Assisted Access networks, the formula enables transmitters to learn to maximize data rates by communicating with each other.
“This work explores the use of machine learning in making decisions about which frequency channel to transmit on,” NIST engineer Jason Coder said. “This could potentially make communications in the unlicensed bands much more efficient.”
By instituting a Q-learning technique, the formula can map environmental conditions to maximize the value that returns the signal. The technique also allows the algorithm to try different actions and learn which provides the best outcomes.
According to a NIST study, an “exhaustive effort” to identify the best communication frequently would take approximately 45,600 trials, but the formula could select a similar solution after just 10 trials.