随机型协同时隙分配模型

为解决机场概率容量条件下的协同时隙分配问题,提出了随机型协同时隙分配策略. 根据机场容量预报的更新模式,分别建立了静态模型和动态模型,模型以各种概率容量情景下的总航班延误损失期望最小为功效性目标,以各种概率容量情景下的平均旅客延误时间期望最小为公平性目标,在满足有效性约束的基础上,为进场航班分配时隙;并采用多目标遗传算法NSGA-Ⅱ求解模型. 以某机场运行数据为例进行仿真验证,与随机型地面等待经典模型相比,本文模型实现了机场概率容量条件下时隙资源的功效、公平、有效分配.     

IP路径吞吐量探测

随着Ｉｎｔｅｒｎｅｔ网络的发展,网络处于急速膨胀阶段．研究了ＩＰ网络的运行参数之一——ＩＰ路径吞吐量的探测算法．通过在特定路径上突发传送一组测试分组,在不影响现有通信的情况下,测试此特定路径当时可以承载的网络流量．结果表明,此算法具有简单和自适应的特点     

MIMO-NOMA系统中短包传输的功率分配方法

多输入多输出非正交多址接入(multiple input and multiple output-non-orthogonal multiple access,MIMO-NOMA)技术能够增加设备连接数量,有效提高系统的频谱效率,针对超可靠低时延应用业务中需要短包传输,而多用户簇的MIMO-NOMA系统同一簇用户对内远用户的吞吐量较低的问题,利用奇异值分解和最大比合并的方法消除簇间干扰,考虑近用户解码远用户的块错误概率低于给定阈值条件下,利用交替迭代算法联合优化簇内用户的发送功率和最大可达传输速率,实现远用户吞吐量的最大化。仿真结果表明与短包传输MIMO正交多址接入(orthogonal multiple access,OMA)系统相比,在保证块错误概率低于10^-5的条件下,远用户的吞吐量提高了32%。     

Comparison of zooplankton community structure between impacted and non-impacted areas of Paranaguá Bay Estuarine Complex,south Brazil

We analysed the seasonal distribution of the zooplankton community in an anthropogenically impacted area (Paranaguá Bay) and a non-impacted area (Laranjeiras Bay) of the Paranaguá Bay Estuarine Complex. Large phytoplankton (> 50 μm) and zooplankton were collected every two months, between August 2003 and June 2004. The phytoplankton community was numerically dominated by diatoms (78%) and dinoflagellates (19%). Zooplankton abundance varied between 670 and 100,716 individuals m^–3, with a dominance of copepods, mainly the calanoids Acartia lilljeborgii, Acartia tonsa and Pseudodiaptomus acutus. A clear seasonal pattern was observed: copepods were significantly more abundant during the rainy than in the dry season. Significant differences in abundance between the two bays were detected only for cirripede larvae, which were more abundant in Paranaguá Bay. This lack of difference between the two areas was probably a consequence of the water circulation along the estuary, which may have diluted and dispersed the pollutants from Paranaguá Bay to other areas of the estuary.     

融合预训练的港口吞吐量LSTM预测模型

港口吞吐量时序变化数据量较小且变化快,传统LSTM神经网络在此类数据上易出现过拟合,导致模型预测性能不佳。针对此问题,本文提出融合预训练与LSTM时序模型,通过预训练捕获任务领域的全局信息,再用LSTM模型精确描述各个港口的吞吐量变化规律,以提升模型对全部港口吞吐量预测的准确性。以天津港等15个中大型港口过去二十一年的月吞吐量为实验数据,以BP、ARIMA、传统LSTM等预测模型和目前流行的GNN-LSTM模型为比较基准进行仿真实验,结果显示本文所提出的融合预训练的LSTM模型能有效解决LSTM神经网络的过拟合问题,整体预测准确率高于所有基准模型。与传统LSTM模型相比,基于预训练的LSTM的MAE指标平均降低45.2%%,最多降低80.0%。     

软件定义网络中基于优先级的在线流调度策略

为了保证服务质量(quality of service, QoS),进入网络中的数据流可被赋予不同的优先级。在未来到达流请求信息未到达的情况下,进行在线优先级流调度,以最大化网络吞吐量是一项挑战。基于网络带宽资源、请求流的优先级和带宽需求的约束研究了软件定义网络(software defined networking, SDN)中的在线流请求调度策略。首先,提出了流路由成本和利润的概念,并创新性地设计了一个考虑边际成本的模型来描述链路资源和路由路径的使用成本。然后,将优先级流请求调度问题刻画为混合整数线性规划模型(mixed integer linear programming, MILP),提出在线优先级流调度算法(online priority traffic scheduling algorithm, OPTSA)来求解,最后分析了OPTSA的竞争比。仿真结果显示,与基准算法相比,所提出的算法可以确保网络负载均衡,同时有效提高网络的累积带宽和吞吐量。     

Adaptive bit and power loading algorithm with low complexity in MIMO-OFDM systems

Two adaptive power and bit loading algorithms to maximize the throughput of MIMO-OFDM systems in frequency selective fading environment are proposed. The two algorithms allocate bit based on maximizing the overall throughput. One algorithm allocates power based on guaranteeing that the bit error rate （BER） of each sub-carrier and the total allocated power remain below a target BER threshold and a power threshold, respectively; another one allocates power based on guaranteeing that the mean BER of sub-carriers and the total allocated power remain below a target BER threshold and a power threshold, respectively. The simulation results show that the proposed algorithms can achieve faster throughput with lower computational complexity, which indicates that the proposed algorithms are effective when compared to some existing algorithms.     

IEEE802.11a WLAN中DCF性能分析

引入二维马尔科夫模型,对DCF的两种传输方式:基本接入方式和RTS/CTS方式,在IEEE802.11a WLAN中的饱和吞吐量和传输时延性能进行了理论分析和仿真,理论分析和仿真结果表明,与低速IEEE802.11WLAN中的情况相反,在IEEE802.11a WLAN中,基本接入方式在多数情况下比RTS/CTS方式更有效.     

用数学法研究总线网性能

网络中的数据流是不确定的，数据包的到达时间、长度和处理时间都是随机变量。通过数学方法研究总线网的性能就是在一定的假设基础上，计算出总线网的参数之间的数学关系，比如吞吐量与负载之间的关系、延迟随负载的变化情况等，其分析结果可以为计算机模拟提供理论支持。对于网络设计具有重要作用。     

802.11 DCF中优化吞吐率的RTS门限调整算法

针对IEEE 802.11 DCF网络,提出了优化吞吐率的RTS门限分布式调整算法.通过分析一次成功发送和一次碰撞持续的平均时间,找到网络饱和吞吐率与RTS门限的关系.分析显示,存在一个最佳RTS门限,使得网络的饱和吞吐率最大.从一次发送过程平均时间最小化出发给出了求解最佳RTS门限的数值计算方法,然后,设计出一个分布式算法,以使终端能自动调整其RTS门限到最佳值,从而优化网络的饱和吞吐率.计算机仿真证明了该分布式算法的有效性.