安全DNS协议DNS-over-HTTPS(DoH)的标准化和部署应用,使DoH隧道成为一种新的隐蔽性网络威胁并受到广泛关注。在云网络环境中对大规模DoH业务流量中潜在的隧道流量进行甄别,需要同时兼顾计算效率和准确率。针对当前基于机器学习的DoH隧...安全DNS协议DNS-over-HTTPS(DoH)的标准化和部署应用,使DoH隧道成为一种新的隐蔽性网络威胁并受到广泛关注。在云网络环境中对大规模DoH业务流量中潜在的隧道流量进行甄别,需要同时兼顾计算效率和准确率。针对当前基于机器学习的DoH隧道检测算法特征效率低、计算复杂度高的问题,设计了一组数据包块长度特征并提出了一种基于最大相关最小冗余(max-Relevance and Min-Re-dundancy,mRMR)特征筛选算法和随机森林算法的低维快速DoH隧道检测方法,该方法通过特征筛选选取对DoH隧道检测任务贡献大的特征,并使用随机森林分类器进行DoH隧道检测任务。实验结果表明,该方法在仅使用10维特征的情况下,达到了与使用24~34维特征的其他算法相当的准确率,可有效降低部署应用的计算复杂度,更好地适应大规模DoH业务流量分析的应用场景。展开更多
随着无线技术的快速发展,无线设备呈现爆炸式增长趋势,导致频谱资源日益稀缺,雷达和通信频段不断重叠。为了避免无线通信和雷达感知之间的相互干扰,学术界广泛研究了通信感知一体化(Integrated Sensing and Communication,ISAC)技术,并...随着无线技术的快速发展,无线设备呈现爆炸式增长趋势,导致频谱资源日益稀缺,雷达和通信频段不断重叠。为了避免无线通信和雷达感知之间的相互干扰,学术界广泛研究了通信感知一体化(Integrated Sensing and Communication,ISAC)技术,并且重点关注了正交时频空(Orthogonal Time Frequency Space,OTFS)信号。OTFS信号具备实现无线通信与雷达感知一体化的潜力。然而,分数多普勒会抬高OTFS多普勒旁瓣,引起多普勒弥散效应,不仅在通信数据与通信数据之间、通信数据与雷达数据之间产生严重干扰,还将导致微弱目标被强目标旁瓣淹没,进而影响雷达探测概率和通信信道估计精度,恶化整体性能。针对分数多普勒导致的OTFS性能下降问题,提出了基于原型滤波器的OTFS通感一体化信号设计方法。通过原型滤波器调理多普勒旁瓣,在不显著损失多普勒分辨率的同时,抑制多普勒弥散效应,提升检测概率,降低误码率。针对OTFS互相关匹配滤波信道估计算法计算复杂度高等问题,进一步提出了利用恒虚警率(Constant False Alarm Rate,CFAR)检测进行信道估计的思路,在降低计算复杂度的同时,稳健检测出了同一时延、不同多普勒的多个目标,保障了信道估计和目标检测性能。依据理论分析和仿真实验可知,本文可将分数多普勒条件下的通信误码率降低2个数量级。展开更多
A novel nonlinear multi-input multi-output MIMO detection algorithm is proposed which is referred to as an ordered successive noise projection cancellation OSNPC algorithm. It is capable of improving the computation p...A novel nonlinear multi-input multi-output MIMO detection algorithm is proposed which is referred to as an ordered successive noise projection cancellation OSNPC algorithm. It is capable of improving the computation performance of the MIMO detector with the conventional ordered successive interference cancellation OSIC algorithm. In contrast to the OSIC in which the known interferences in the input signal vector are successively cancelled the OSNPC successively cancels the known noise projections from the decision statistic vector. Analysis indicates that the OSNPC is equivalent to the OSIC in error performance but it has significantly less complexity in computation.Furthermore when the OSNPC is applied to the MIMO detection with the preprocessing of dual lattice reduction DLR the computational complexity of the proposed OSNPC-based DLR-aided detector is further reduced due to the avoidance of the inverse of the reduced basis of the dual lattice in computation compared to that of the OSIC-based one. Simulation results validate the theoretical conclusions with regard to both the performance and complexity of the proposed MIMO detection scheme.展开更多
文摘安全DNS协议DNS-over-HTTPS(DoH)的标准化和部署应用,使DoH隧道成为一种新的隐蔽性网络威胁并受到广泛关注。在云网络环境中对大规模DoH业务流量中潜在的隧道流量进行甄别,需要同时兼顾计算效率和准确率。针对当前基于机器学习的DoH隧道检测算法特征效率低、计算复杂度高的问题,设计了一组数据包块长度特征并提出了一种基于最大相关最小冗余(max-Relevance and Min-Re-dundancy,mRMR)特征筛选算法和随机森林算法的低维快速DoH隧道检测方法,该方法通过特征筛选选取对DoH隧道检测任务贡献大的特征,并使用随机森林分类器进行DoH隧道检测任务。实验结果表明,该方法在仅使用10维特征的情况下,达到了与使用24~34维特征的其他算法相当的准确率,可有效降低部署应用的计算复杂度,更好地适应大规模DoH业务流量分析的应用场景。
文摘随着无线技术的快速发展,无线设备呈现爆炸式增长趋势,导致频谱资源日益稀缺,雷达和通信频段不断重叠。为了避免无线通信和雷达感知之间的相互干扰,学术界广泛研究了通信感知一体化(Integrated Sensing and Communication,ISAC)技术,并且重点关注了正交时频空(Orthogonal Time Frequency Space,OTFS)信号。OTFS信号具备实现无线通信与雷达感知一体化的潜力。然而,分数多普勒会抬高OTFS多普勒旁瓣,引起多普勒弥散效应,不仅在通信数据与通信数据之间、通信数据与雷达数据之间产生严重干扰,还将导致微弱目标被强目标旁瓣淹没,进而影响雷达探测概率和通信信道估计精度,恶化整体性能。针对分数多普勒导致的OTFS性能下降问题,提出了基于原型滤波器的OTFS通感一体化信号设计方法。通过原型滤波器调理多普勒旁瓣,在不显著损失多普勒分辨率的同时,抑制多普勒弥散效应,提升检测概率,降低误码率。针对OTFS互相关匹配滤波信道估计算法计算复杂度高等问题,进一步提出了利用恒虚警率(Constant False Alarm Rate,CFAR)检测进行信道估计的思路,在降低计算复杂度的同时,稳健检测出了同一时延、不同多普勒的多个目标,保障了信道估计和目标检测性能。依据理论分析和仿真实验可知,本文可将分数多普勒条件下的通信误码率降低2个数量级。
基金The National Science and Technology Major Project(No.2012ZX03004005-003)the National Natural Science Foundation of China(No.61171081,61201175)the Innovation Technology Fund of Jiangsu Province(No.BC2012006)
文摘A novel nonlinear multi-input multi-output MIMO detection algorithm is proposed which is referred to as an ordered successive noise projection cancellation OSNPC algorithm. It is capable of improving the computation performance of the MIMO detector with the conventional ordered successive interference cancellation OSIC algorithm. In contrast to the OSIC in which the known interferences in the input signal vector are successively cancelled the OSNPC successively cancels the known noise projections from the decision statistic vector. Analysis indicates that the OSNPC is equivalent to the OSIC in error performance but it has significantly less complexity in computation.Furthermore when the OSNPC is applied to the MIMO detection with the preprocessing of dual lattice reduction DLR the computational complexity of the proposed OSNPC-based DLR-aided detector is further reduced due to the avoidance of the inverse of the reduced basis of the dual lattice in computation compared to that of the OSIC-based one. Simulation results validate the theoretical conclusions with regard to both the performance and complexity of the proposed MIMO detection scheme.
