Reliability optimization by AI-based fault mitigation in 5G/6G
View Patent ↗An unsolved problem in 5G-Advanced and especially 6G is message fault mitigation without a costly retransmission. Methods are disclosed for the receiver to analyze each message element's received waveform signal to detect characteristic features of interference and noise, such as excessive amplitude or phase variation within the message element or excessive deviation from the predetermined modulation levels of the modulation scheme, and to provide that data to an AI model trained in message fault correction. The AI model can then identify the faulted message elements, and attempt to correct them according to the likely intent or meaning of the message based on the non-faulted message elements, and on the bit sequences of previously received non-faulted messages, and other criteria that the AI model may apply. By repairing the message upon receipt, the costs in time, transmission power, and background noise generation may be avoided. Next-generation users will enjoy the improved reception.
1. A method for a wireless receiver to determine a corrected value of a faulted message element, the method comprising:
a) receiving and demodulating a message comprising message elements, each message element occupying a single resource element of a resource grid, each message element comprising a waveform signal modulated according to a modulation scheme;
b) receiving and demodulating an error-detection code associated with the message, the error-detection code comprising a hash or digest or parity construct of the message;
c) determining, according to the error-detection code, that at least one message element is faulted;
d) providing, as input to an artificial intelligence (AI) model, data about the waveform signal of each message element of the message and data about the error-detection code; and
e) determining, as output from the AI model, a corrected version of the message;
f) wherein the determining that at least one message element is faulted comprises determining that the message, as demodulated, is inconsistent with the error-detection code, as demodulated.
2. The method of claim 1 , wherein the message is received according to 5G or 6G technology.
3. The method of claim 1 , wherein the waveform signal of each message element comprises an amplitude and a phase, and the data about the waveform signal comprises the amplitude and the phase of each message element.
4. The method of claim 3 , wherein the data about the waveform signal of each message element further comprises:
a) a variation or fluctuation of the amplitude within the message element; and
b) a variation or fluctuation of the phase within the message element.
5. The method of claim 1 , wherein the data about the waveform signal of each message element further comprises a difference between a modulation parameter of the message element and a closest predetermined modulation level of the modulation scheme.
6. The method of claim 5 , wherein the modulation parameter comprises the amplitude or the phase of the waveform signal.
7. The method of claim 5 , wherein:
a) the modulation scheme comprises an I branch and an orthogonal Q branch, wherein the waveform signal of each message element comprises a sum of the I branch and the Q branch of the message element; and
b) the modulation parameter comprises an amplitude of the I branch or an amplitude of the Q branch.
8. The method of claim 5 , wherein the data about the waveform signal of each message element further comprises a difference between the waveform signal of the message element and an average of the waveform signals of a selected subset of the message elements.
9. The method of claim 1 , wherein the data about the waveform signal of each message element further comprises a difference between a frequency of the waveform signal and a predetermined subcarrier frequency.
10. A method for a wireless receiver to demodulate a message, the method comprising:
a) receiving a message comprising message elements, each message element comprising a single resource element comprising a waveform signal, the waveform signal modulated according to a modulation scheme;
b) determining, according to an error-detection code associated with the message, that the message is corrupted;
c) providing, as input to an artificial intelligence (AI) model, for each message element:
i) an average amplitude of the waveform signal within the message element;
ii) an average phase of the waveform signal within the message element;
iii) a variation or fluctuation in the amplitude of the waveform signal within the message element, relative to the average amplitude; and
iv) a variation or fluctuation in the phase of the waveform within the message element, relative to the average phase; and
d) determining, as output from the AI model, for each message element, a probability that the message element is faulted.
11. The method of claim 10 , further comprising determining, as further output from the AI model, for each faulted message element, a corrected value of the faulted message element.
12. The method of claim 10 , further comprising determining, as further output from the AI model, a corrected version of the message.
13. The method of claim 10 , further comprising:
a) providing, as further input to the AI model, bit sequences or character sequences of non-faulted messages previously received by the wireless receiver; and
b) determining, as further output from the AI model, whether the message deviates from those bit sequences or character sequences.
14. The method of claim 13 , further comprising:
a) when the message deviates from the bit sequences or character sequences of previously received non-faulted messages, determining, as further output from the AI model, a version of the wireless message that avoids deviating from the bit sequences or character sequences of the previously received non-faulted messages.
15. The method of claim 10 , further comprising determining, as further output from the AI model, an intent or meaning of the message according to non-faulted message elements of the message.
16. The method of claim 10 , further comprising, when the message violates a format or convention, determining, as further output from the AI model, an altered version of the message compliant with the format or convention.
17. The method of claim 10 , further comprising calculating, according to the error-detection code, a corrected value for each faulted message element.