星地融合网络通过实现卫星网络与地面蜂窝网络的有机融合,可为用户提供灵活的网络接入服务,满足多场景用户应用需求.然而,卫星网络拓扑结构的动态变化、星地链路的复杂特性、多网和多用户资源调度粒度的显著差异性,以及资源共享模式的...星地融合网络通过实现卫星网络与地面蜂窝网络的有机融合,可为用户提供灵活的网络接入服务,满足多场景用户应用需求.然而,卫星网络拓扑结构的动态变化、星地链路的复杂特性、多网和多用户资源调度粒度的显著差异性,以及资源共享模式的多样性,均对星地融合网络用户关联及资源调度问题带来困难及挑战.本文针对星地融合网络下行链路增强型移动宽带用户(Enhanced mobile broadband User,EU)及高可靠低延迟通信用户(ultra-Reliable and low latency communications User,RU)数据传输场景,研究用户关联及资源调度问题.在满足用户传输需求、网络可用资源、资源共享模式等约束条件下,将联合用户关联及资源块调度问题建模为系统效用函数最大化问题.由于该优化问题为强耦合问题,不易直接求解,本文将原优化问题分解功率分配子问题、用户关联及资源调度子问题.针对功率分配子问题,利用拉格朗日部分松弛法对离散变量进行连续化,进而将功率分配子问题转换为凸优化问题,采用拉格朗日对偶算法进行求解.基于得到的功率分配策略,本文提出一种启发式算法确定用户关联模式及资源块分配策略,在对蜂窝网络可用时频资源状态进行评估的基础上,分别针对时频资源充足及时频资源不充足的情况设计用户关联及资源块调度策略.针对时频资源充足情况,根据用户资源调度模式,分别定义RU及EU的时空资源调度粒度,进而针对采用正交频分多址接入(Orthogonal Frequency Division Multiple Access,OFDMA)模式的RU及EU,基于KM匹配算法优化确定时频资源块调度策略;进而,针对采用穿孔模式的RU与EU,提出一种基于贪婪算法的EU资源块重调度策略,以满足用户最小传输速率;针对占用相邻子信道的EU优化确定OFDMA模式及非正交多址接入(Non-Orthogonal Multiple Access,NOMA)模式,以实现效用函数的提升.针对蜂窝展开更多
本地差分隐私具有不需要可信第三方、交互少、运行效率高等优点,近年来受到了广泛关注.然而,现有本地差分隐私集合数据频率估计机制未能考虑数据的隐私敏感度差异,将所有数据同等对待,这会对非敏感数据保护过强,导致估计结果准确度低....本地差分隐私具有不需要可信第三方、交互少、运行效率高等优点,近年来受到了广泛关注.然而,现有本地差分隐私集合数据频率估计机制未能考虑数据的隐私敏感度差异,将所有数据同等对待,这会对非敏感数据保护过强,导致估计结果准确度低.针对这一问题,定义了集合数据效用优化本地差分隐私(set-valued data utility-optimized local differential privacy, SULDP)模型,考虑了原始数据域同时包含敏感值和非敏感值的情况,在不减弱对敏感值保护的前提下,允许降低对非敏感值的保护.进一步,提出了符合SULDP模型的5种频率估计机制suGRR,suGRR-Sample, suRAP,suRAP-Sample和suWheel,理论分析证实,相对于现有的本地差分隐私机制,所提方案能够对敏感数据实现完全相同的保护效果,并通过降低非敏感数据的保护效果,实现了频率估计结果的准确度提升.最后,在真实和模拟数据集上评估了新的方案,实验结果证明了所提的5种机制能够有效降低估计误差,提升数据效用,其中suWheel机制表现最优.展开更多
Two utility-optimization dynamic subcarrier allocation(DSA) algorithms are designed for single carrier frequency division multiple access system(SC-FDMA).The two proposed algorithms aim to support diverse transmission...Two utility-optimization dynamic subcarrier allocation(DSA) algorithms are designed for single carrier frequency division multiple access system(SC-FDMA).The two proposed algorithms aim to support diverse transmission capacity requirements in wireless networks,which consider both the channel state information(CSI) and the capacity requirements of each user by setting appropriate utility functions.Simulation results show that with considerable lower computational complexity,the first utility-optimization algorithm can meet the system capacity requirements of each user effectively.However,the rate-sum capacity performance is poor.Furthermore,the second proposed utility-optimization algorithm can contribute a better trade-off between system rate-sum capacity requirement and the capacity requirements of each user by introducing the signal to noise ratio(SNR) information to the utility function based on the first utility-optimization algorithm,which can improve the user requirements processing capability as well as achieve a better sum-rate capacity.展开更多
The existing studies, concerning the dressing process, focus on the major influence of the dressing conditions on the grinding response variables. However, the choice of the dressing conditions is often made, based on...The existing studies, concerning the dressing process, focus on the major influence of the dressing conditions on the grinding response variables. However, the choice of the dressing conditions is often made, based on the experience of the qualified staff or using data from reference books. The optimal dressing parameters, which are only valid for the particular methods and dressing and grinding conditions, are also used. The paper presents a methodology for optimization of the dressing parameters in cylindrical grinding. The generalized utility function has been chosen as an optimization parameter. It is a complex indicator determining the economic, dynamic and manufacturing characteristics of the grinding process. The developed methodology is implemented for the dressing of aluminium oxide grinding wheels by using experimental diamond roller dressers with different grit sizes made of medium- and high-strength synthetic diamonds type AC32 and AC80. To solve the optimization problem, a model of the generalized utility function is created which reflects the complex impact of dressing parameters. The model is built based on the results from the conducted complex study and modeling of the grinding wheel lifetime, cutting ability, production rate and cutting forces during grinding. They are closely related to the dressing conditions （dressing speed ratio, radial in-feed of the diamond roller dresser and dress-out time）, the diamond roller dresser grit size/grinding wheel grit size ratio, the type of synthetic diamonds and the direction of dressing. Some dressing parameters are determined for which the generalized utility fimction has a maximum and which guarantee an optimum combination of the following： the lifetime and cutting ability of the abrasive wheels, the tangential cutting force magnitude and the production rate of the grinding process. The results obtained prove the possibility of control and optimization of grinding by selecting particular dressing parameters.展开更多
文摘星地融合网络通过实现卫星网络与地面蜂窝网络的有机融合,可为用户提供灵活的网络接入服务,满足多场景用户应用需求.然而,卫星网络拓扑结构的动态变化、星地链路的复杂特性、多网和多用户资源调度粒度的显著差异性,以及资源共享模式的多样性,均对星地融合网络用户关联及资源调度问题带来困难及挑战.本文针对星地融合网络下行链路增强型移动宽带用户(Enhanced mobile broadband User,EU)及高可靠低延迟通信用户(ultra-Reliable and low latency communications User,RU)数据传输场景,研究用户关联及资源调度问题.在满足用户传输需求、网络可用资源、资源共享模式等约束条件下,将联合用户关联及资源块调度问题建模为系统效用函数最大化问题.由于该优化问题为强耦合问题,不易直接求解,本文将原优化问题分解功率分配子问题、用户关联及资源调度子问题.针对功率分配子问题,利用拉格朗日部分松弛法对离散变量进行连续化,进而将功率分配子问题转换为凸优化问题,采用拉格朗日对偶算法进行求解.基于得到的功率分配策略,本文提出一种启发式算法确定用户关联模式及资源块分配策略,在对蜂窝网络可用时频资源状态进行评估的基础上,分别针对时频资源充足及时频资源不充足的情况设计用户关联及资源块调度策略.针对时频资源充足情况,根据用户资源调度模式,分别定义RU及EU的时空资源调度粒度,进而针对采用正交频分多址接入(Orthogonal Frequency Division Multiple Access,OFDMA)模式的RU及EU,基于KM匹配算法优化确定时频资源块调度策略;进而,针对采用穿孔模式的RU与EU,提出一种基于贪婪算法的EU资源块重调度策略,以满足用户最小传输速率;针对占用相邻子信道的EU优化确定OFDMA模式及非正交多址接入(Non-Orthogonal Multiple Access,NOMA)模式,以实现效用函数的提升.针对蜂窝
文摘本地差分隐私具有不需要可信第三方、交互少、运行效率高等优点,近年来受到了广泛关注.然而,现有本地差分隐私集合数据频率估计机制未能考虑数据的隐私敏感度差异,将所有数据同等对待,这会对非敏感数据保护过强,导致估计结果准确度低.针对这一问题,定义了集合数据效用优化本地差分隐私(set-valued data utility-optimized local differential privacy, SULDP)模型,考虑了原始数据域同时包含敏感值和非敏感值的情况,在不减弱对敏感值保护的前提下,允许降低对非敏感值的保护.进一步,提出了符合SULDP模型的5种频率估计机制suGRR,suGRR-Sample, suRAP,suRAP-Sample和suWheel,理论分析证实,相对于现有的本地差分隐私机制,所提方案能够对敏感数据实现完全相同的保护效果,并通过降低非敏感数据的保护效果,实现了频率估计结果的准确度提升.最后,在真实和模拟数据集上评估了新的方案,实验结果证明了所提的5种机制能够有效降低估计误差,提升数据效用,其中suWheel机制表现最优.
基金Supported by the National Basic Research Program of China(No.61393010101-1)the Defense-related Science & Technology Pre-Research Project of Shipbuilding Institute(No.10J3.1.6)
文摘Two utility-optimization dynamic subcarrier allocation(DSA) algorithms are designed for single carrier frequency division multiple access system(SC-FDMA).The two proposed algorithms aim to support diverse transmission capacity requirements in wireless networks,which consider both the channel state information(CSI) and the capacity requirements of each user by setting appropriate utility functions.Simulation results show that with considerable lower computational complexity,the first utility-optimization algorithm can meet the system capacity requirements of each user effectively.However,the rate-sum capacity performance is poor.Furthermore,the second proposed utility-optimization algorithm can contribute a better trade-off between system rate-sum capacity requirement and the capacity requirements of each user by introducing the signal to noise ratio(SNR) information to the utility function based on the first utility-optimization algorithm,which can improve the user requirements processing capability as well as achieve a better sum-rate capacity.
文摘The existing studies, concerning the dressing process, focus on the major influence of the dressing conditions on the grinding response variables. However, the choice of the dressing conditions is often made, based on the experience of the qualified staff or using data from reference books. The optimal dressing parameters, which are only valid for the particular methods and dressing and grinding conditions, are also used. The paper presents a methodology for optimization of the dressing parameters in cylindrical grinding. The generalized utility function has been chosen as an optimization parameter. It is a complex indicator determining the economic, dynamic and manufacturing characteristics of the grinding process. The developed methodology is implemented for the dressing of aluminium oxide grinding wheels by using experimental diamond roller dressers with different grit sizes made of medium- and high-strength synthetic diamonds type AC32 and AC80. To solve the optimization problem, a model of the generalized utility function is created which reflects the complex impact of dressing parameters. The model is built based on the results from the conducted complex study and modeling of the grinding wheel lifetime, cutting ability, production rate and cutting forces during grinding. They are closely related to the dressing conditions （dressing speed ratio, radial in-feed of the diamond roller dresser and dress-out time）, the diamond roller dresser grit size/grinding wheel grit size ratio, the type of synthetic diamonds and the direction of dressing. Some dressing parameters are determined for which the generalized utility fimction has a maximum and which guarantee an optimum combination of the following： the lifetime and cutting ability of the abrasive wheels, the tangential cutting force magnitude and the production rate of the grinding process. The results obtained prove the possibility of control and optimization of grinding by selecting particular dressing parameters.
