To solve the problem that Non Line Of Sight (NLOS) state identification requires classification of known channel types, a channel environment classification algorithm based on Convolutional Neural Network (CNN) was proposed. Firstly, an Ultra-WideBand (UWB) channel was sampled, and a sample set was constructed. Then, a CNN was trained by the sample set to extract features of different channel scenes. Finally, the classification of UWB channel environment was realized. The experimental results show that the overall accuracy of the model using the proposed algorithm is about 93.40% and the algorithm can effectively realize the classification of channel environments.

To study spectrum allocation problem in Cognitive Wireless Mesh Network (CWMN), an immune-based multi-objective optimization algorithm was proposed. The problem was modeled as a multi-objective optimization problem to maximize total bandwidth and minimize the total number of occupied spectra. Antibody encoding, whole cloning operator and non-dominated anti-body selection operator that are suitable for solving the problem were designed. The simulations results show that the proposed algorithm can obtain the Pareto optimal solutions of CWMN spectrum allocation, which improves the total bandwidth and minimizes the occupied spectra, so it optimizes the spectrum allocation performance.