| （英） |
The performance of wireless communication greatly depends on the propagation environment.
Adaptive modulation and coding (AMC) is essential to adapt to each environment.
Parameters representing the communication environment include SNR, Doppler shift, line of sight/multipath components, and interference.
Such information should be estimated by individual signal processing using known reference signals.
If all information can be estimated instantaneously without use of known signals, channel capacity can be efficiently and maximally utilized by AMC and interference cancellation.
Deep learning is applied to resolve these issues.
By extracting various features from the received signal, information about communication environments can be estimated simultaneously.
This paper presents our recent works, deep learning aided communication environment estimation, and proposes a approach that can improve the estimation accuracy by giving additional processing to the received signal. |