Real Time Perception Method for Full Link State in Low Latency Channels
Miaozhuang Cai, Xingyuan Fan, Zhengyang Peng, Junyi Chen · 2025
In order to solve the difficulties of dynamic monitoring and real-time feedback of the entire link state in low latency communication scenarios, a real-time perception method for the entire link state in low latency channels is proposed. Firstly, the maximum correlation kurtosis deconvolution algorithm is used for signal denoising, and channel noise is effectively suppressed by optimizing filter coefficients; Secondly, based on the improved Mexican hat wavelet transform, signal feature extraction is achieved, and the feature representation ability is improved by adjusting the time-frequency domain parameters; Finally, a Bayesian probabilistic neural network model is constructed, utilizing Parzen window technology and Bayesian decision theory to achieve real-time classification and perception of the entire chain state. The experimental results show that the proposed method exhibits significant advantages in full link state perception under low latency channels, with a stable false alarm rate of less than $0.5 \%$ and a processing delay controlled within 0.15 seconds. Its performance is significantly better than the compared methods.