ID 原文 译文
14365 采用基于节点层级号的信标时隙分配机制，根据邻居表和时隙分配信息计算能够提前进入到载波监听多路访问时隙( Carrier SenseMultiple Access，CSMA) 的时间，减少了信标时隙的浪费，提高了信道利用率。 The beacon timeslot allocation mechanism based on node layer number is adopted，according to neighbor tableand timeslot allocation information，the time when CSMA timeslot can be entered in advance can be calculated，which reduces the waste of beacon timeslot and improves the channel utilization．
14366 同时采用基于拓扑信息的信标帧高效广播机制，删除不大于自身层级号的节点和大于且在两跳范围内的非子孙节点的时隙分配信息，降低了网络控制开销。 At the same time，the efficient broadcast mechanism of beacon frame based on topology information is used，the main idea isto delete the timeslot allocation information of nodes whose layer number is not greater than itself and thetimeslot allocation information of non-descendant nodes whose layer number is greater than itself within twohops，which ultimately reduces the network control overhead．
14367 仿真结果表明，ELDM-MAC 协议在信道利用率、平均时延和控制开销等方面都优于 IEEE1901． 1 MAC 协议，更适用于宽带电力线通信网络的实际应用场景。 The simulation results show that ELDM-MACprotocol proposed performs better than IEEE1901． 1 MAC protocol in channel utilization，average delay andcontrol overhead，which can be preferably applied in practical scenarios of BPLC networks．
14368 最小均方误差( Minimum Mean Square Error，MMSE) 检测算法，虽然能在大规模多输入多输出系统中获得接近最优的线性检测性能，但是涉及高维矩阵求逆运算，难以在实际应用中快速有效地实现。 Minimum mean square error( MMSE) detection algorithm is near-optimal for massive multipleinput multiple-output( MIMO) systems，but it involves matrix inversion with high complexity． Thus，it is difficult to apply quickly and effectively in practice．
14369 提出了块高斯－赛德尔( Block Gauss-Seidel，BGS) 低复杂度信号检测算法，将 MMSE 检测器的滤波矩阵先进行分块预处理，构造分裂矩阵，再通过迭代求解发送信号向量估计值，以提高算法检测性能。 In this paper，a low complexity Block Gauss-Seidel( BGS)algorithm based on block matrix is proposed for improving the traditional Gauss-Seidel( GS) algorithm．
14370 仿真结果表明，BGS 迭代算法在调制方式为 64QAM、用户侧的天线数量设置为 16、基站侧的天线数量设置为 256 时，迭代 2 次后就能快速接近 MMSE 检测性能。 The simulation results show that the proposed BGS iterative algorithm can approach the MMSE detection performance with 2 iterations，when the modulation mode is 64QAM，and the number of antennas on the userterminal and on the base station is set to 16 and 256，respectively．
14371 在设置近似初始值后，BGS 算法的性能得到了进一步的改善。当调制方式为 256QAM 时，设置近似初始值的 BGS 算法在迭代 2 次后就能逼近MMSE 算法的误码率( Bit Error Ｒatio，BEＲ) 性能曲线，此时算法的复杂度仍然保持在 O( K2) 。 When the initial value is set，the performance of the BGS algorithm is further improved． When the modulation mode is 256QAM，the BGS algorithm can approximate the bit error rate( BEＲ) performance curve of the MMSE algorithm with 2 iterations，and the complexity of the algorithm remains at O( K2) ．
14372 针对正交频分多址( Orthogonal Frequency Division Multiplexing Access，OFDMA) 异构网络中用户关联和功率控制协同优化不佳的问题，提出了一种多智能体深度 Q 学习网络( Deep Q-learningNetwork，DQN) 方法。 To solve the problem of poor joint optimization of user association and power control in orthogonalfrequency division multiplexing access( OFDMA) heterogeneous networks，a multi-agent deep Q-learningnetwork( DQN) method is proposed．
14373 首先，基于用户关联和功率控制最优化问题，构建了正交频分多址的双层异构网络系统模型，以实现智能决策。 First，a two-layer heterogeneous network system model of OFDMA isconstructed to realize intelligent decision-making based on user association and power control optimization．
14374 其次，根据应用场景和多智能体 DQN 框架的动作空间，对状态空间和奖励函数进行重构。 Secondly，the state space and reward function are reconstructed according to the application scenario andthe action space of the multi-agent DQN framework．