A novel high-voltage light punch-through（LPT） carrier stored trench bipolar transistor（CSTBT） with buried p-layer（BP） is proposed in this paper.Since the negative charges in the BP layer modulate the bulk electr...A novel high-voltage light punch-through（LPT） carrier stored trench bipolar transistor（CSTBT） with buried p-layer（BP） is proposed in this paper.Since the negative charges in the BP layer modulate the bulk electric field distribution,the electric field peaks both at the junction of the p base/n-type carrier stored（N-CS） layer and the corners of the trench gates are reduced,and new electric field peaks appear at the junction of the BP layer/N drift region.As a result,the overall electric field in the N drift region is enhanced and the proposed structure improves the breakdown voltage（BV） significantly compared with the LPT CSTBT.Furthermore,the proposed structure breaks the limitation of the doping concentration of the N-CS layer（NN CS） to the BV,and hence a higher NN CS can be used for the proposed LPT BP-CSTBT structure and a lower on-state voltage drop（Vce（sat）） can be obtained with almost constant BV.The results show that with a BP layer doping concentration of NBP = 7 × 10^15 cm^-3,a thickness of LBP = 2.5 μm,and a width of WBP = 5 μm,the BV of the proposed LPT BP-CSTBT increases from 1859 V to 1862 V,with NN CS increasing from 5 × 10^15 cm^-3 to 2.5 × 10^16 cm^-3.However,with the same N-drift region thickness of 150 μm and NN CS,the BV of the CSTBT decreases from 1598 V to 247 V.Meanwhile,the Vce（sat） of the proposed LPT BP-CSTBT structure decreases from 1.78 V to 1.45 V with NN CS increasing from 5 × 10^15 cm^-3 to 2.5 × 10^16 cm^-3.展开更多
针对无线和电力线通信混合组网的信道竞争接入问题,提出了一种基于深度强化学习的电力线与无线双模通信的MAC接入算法。双模节点根据网络广播信息和信道使用等数据自适应接入双媒质信道。首先建立了基于双模通信网络交互和统计信息的双...针对无线和电力线通信混合组网的信道竞争接入问题,提出了一种基于深度强化学习的电力线与无线双模通信的MAC接入算法。双模节点根据网络广播信息和信道使用等数据自适应接入双媒质信道。首先建立了基于双模通信网络交互和统计信息的双模通信节点数据采集模型;接着定义了基于协作信息的深度强化学习(deep reinforcement learning,DRL)状态空间、动作空间和奖励,设计了联合α-公平效用函数和P坚持接入机制的节点决策流程,实现基于双深度Q网络(double deep Q-network,DDQN)的双模节点自适应接入算法;最后进行算法性能仿真和对比分析。仿真结果表明,提出的接入算法能够在保证双模网络和信道接入公平性的条件下,有效提高双模通信节点的接入性能。展开更多
基金Project supported by the National Science and Technology Major Project of China （Grant No. 2011ZX02504-003） and the Fundamental Research Funds for the Central Universities, China （Grant No. ZYGX2011J024）.
文摘A novel high-voltage light punch-through（LPT） carrier stored trench bipolar transistor（CSTBT） with buried p-layer（BP） is proposed in this paper.Since the negative charges in the BP layer modulate the bulk electric field distribution,the electric field peaks both at the junction of the p base/n-type carrier stored（N-CS） layer and the corners of the trench gates are reduced,and new electric field peaks appear at the junction of the BP layer/N drift region.As a result,the overall electric field in the N drift region is enhanced and the proposed structure improves the breakdown voltage（BV） significantly compared with the LPT CSTBT.Furthermore,the proposed structure breaks the limitation of the doping concentration of the N-CS layer（NN CS） to the BV,and hence a higher NN CS can be used for the proposed LPT BP-CSTBT structure and a lower on-state voltage drop（Vce（sat）） can be obtained with almost constant BV.The results show that with a BP layer doping concentration of NBP = 7 × 10^15 cm^-3,a thickness of LBP = 2.5 μm,and a width of WBP = 5 μm,the BV of the proposed LPT BP-CSTBT increases from 1859 V to 1862 V,with NN CS increasing from 5 × 10^15 cm^-3 to 2.5 × 10^16 cm^-3.However,with the same N-drift region thickness of 150 μm and NN CS,the BV of the CSTBT decreases from 1598 V to 247 V.Meanwhile,the Vce（sat） of the proposed LPT BP-CSTBT structure decreases from 1.78 V to 1.45 V with NN CS increasing from 5 × 10^15 cm^-3 to 2.5 × 10^16 cm^-3.
文摘针对无线和电力线通信混合组网的信道竞争接入问题,提出了一种基于深度强化学习的电力线与无线双模通信的MAC接入算法。双模节点根据网络广播信息和信道使用等数据自适应接入双媒质信道。首先建立了基于双模通信网络交互和统计信息的双模通信节点数据采集模型;接着定义了基于协作信息的深度强化学习(deep reinforcement learning,DRL)状态空间、动作空间和奖励,设计了联合α-公平效用函数和P坚持接入机制的节点决策流程,实现基于双深度Q网络(double deep Q-network,DDQN)的双模节点自适应接入算法;最后进行算法性能仿真和对比分析。仿真结果表明,提出的接入算法能够在保证双模网络和信道接入公平性的条件下,有效提高双模通信节点的接入性能。
