基于深度无监督学习的多小区蜂窝网资源分配方法

针对多小区蜂窝网络资源分配所要求的低能耗、高速率和低延时问题,提出一种基于深度无监督学习的多小区蜂窝网络资源分配方法.首先,构建基于无监督学习的深度功率控制神经网络,通过约束处理输出优化的信道功率控制方案以最大化能量效率的期望;然后,构建基于无监督学习的深度信道分配神经网络,通过约束处理输出优化的信道分配方案,并联合前期训练好的深度功率控制神经网络拟合输出优化的信道功率,进一步优化能量效率的期望.仿真结果表明,所提出的方法在保证低计算时延的同时可获得优于其他算法的能量效率和传输速率.

Deep unsupervised learning based resource allocation method for multi-cell cellular networks

Aiming at the problem of low energy consumption, high speed, and low latency required for resource allocation for multi-cell cellular networks, a deep unsupervised learning based resource allocation method is proposed. Firstly, an unsupervised learning based deep power control neural network is constructed to output an optimized channel power control scheme by constraint handling, so as to maximize the expectation of energy efficiency. Then, an unsupervised learning based deep channel allocation neural network is constructed to output an optimized channel allocation scheme by constraint handling, and the unsupervised learning based deep power control neural network trained well previously is combined to fit and output the optimized channel power control scheme to further optimize the expectation of energy efficiency. The simulation results show that the proposed method can obtain better transmission rate and energy efficiency than other algorithms while ensuring low computational latency.