1.School of Information Science and Engineering, Hebei University of Science and Technology, Shijiazhuang 050018, China 2.Science and Technology on Communication Networks Laboratory, Shijiazhuang 050081, China
Fund Project:Project supported by the Science and Technology Research Project of Hebei Higher Education Institutions, China (Grant No. QN2019081), the Science and Technology Research and Development Plan Project of Shijiazhuang, China (Grant No. 195790055A), and the Research Projects on Social Science Development in Hebei Province, China (Grant No. 2019041201007)
Received Date:05 February 2020
Accepted Date:23 April 2020
Available Online:12 June 2020
Published Online:20 September 2020
Abstract:The two-layer network model offers us a new viewpoint to observe the traffic dynamics of multilayer network systems. An efficient coupling mechanism is of great importance for alleviating the traffic congestion on two-layer networks. In order to reduce the network congestion and improve network transmission performance, the coupling mechanism between two layers of network and three coupling methods, which are random coupling, disassortative coupling and assortative coupling, are studied based on degree correlation. The packet transmission process is analyzed with both the shortest path routing strategy and degree-based weight routing strategy. The influences of the coupling mode and its corresponding routing strategy on the traffic capacity of the two-layer network are studied. In this paper, two scale-free networks are used to construct the two-layer network for simulation experiments. The network scale is in a range from 200 to 2400 with the value of average degree being 8. We focus on the traffic dynamics of two-layer network, and analyze the relationship between the traffic capacity and the three coupling modes, which are random coupling, disassortative coupling and assortative coupling, under the constraints of the shortest path routing strategy and the weight-based routing strategy. According to the characteristics of the coupling connection between the two layers of network, the best coupling method which is suitable for a certain routing strategy should be investigated. The suitable coupling connection between the two layers can effectively increase the traffic capacity. Both numerical result and analytical result show that the packet generation rate, average transmission time, and average throughput can be obviously improved under the shortest path routing strategy with the disassortative coupling method. When the degree-based static weight routing strategy is used, the traffic performance parameters such as packet generation rate, average transmission time, and average throughput can reach the optimal values with the assortative coupling method. It makes the traffic flow uniform that the routing strategy is chosen with the most suitable coupling method on the two-layer network, and the network traffic capacity may be effectively enhanced. More generally, the results indicate that the coupling modes can give rise to traffic behavior that relies subtly on the routing strategy on the two-layer network. Our work may shed some light on the design and optimization of some real traffic or communication networks. Keywords:coupling network/ traffic capacity/ coupling strength/ average path length
其中$ n $是在指定时间内到达目的节点的数据包个数; $ \left\langle T_i \right\rangle $是数据包$ i $的传输时间, 包括数据包网络中的传送时间和在拥塞节点处的等待时间. 当$ R < R_{\rm c} $时, 网络处于没有发生拥塞的自由态, 数据包的传输时间只依赖于在网络中的传递时间, 而等待时间几乎为零. 由于数据包的传递时间相对较小, 所以此时$ \left\langle T \right\rangle $数值较小. 当$ R > R_{\rm c} $时, 系统开始进入拥塞状态, 数据包的传输时间主要由传递时间和等待时间之和决定, 其中传递时间较小, 主要由等待时间决定, 所以进入拥塞状态时$ \left\langle T \right\rangle $数值会突然增大, 发生相变. 如图9所示, 分别采用SPR 策略和DWR策略, 网络规模$ N = 400 $, 平均度$ \left\langle k \right\rangle = 8 $, 研究在RC, DC 和AC耦合方式时$ \left\langle T \right\rangle $ 随数据包产生率$ R $增长的变化趋势. 从图9(a)可以看出, 采用SPR策略, 比较RC, DC和AC这三种耦合方式下的$ \left\langle T \right\rangle $的数值大小, 在DC方式下, $ \left\langle T \right\rangle $发生的相变值最大, 网络对数据包的处理能力相对较强, DC 方式为最佳耦合方式. 从图9(b) 可以看出, 采用DWR策略, 对比RC, DC和AC方式下的变换曲线, 可得AC方式下的$ \left\langle T \right\rangle $的相变点数值最大, 在一定程度上提高了传输容量, AC方式为最佳耦合方式. 图 9 采用两种路由策略不同耦合方式平均传输时间$ \left\langle T \right\rangle $与$ R $的关系 (a) SPR策略; (b) DWR策略, Figure9. Relationship between average transmission time $ \left\langle T \right\rangle $ and $ R $ under two routing strategies with different coupling: (a) SPR; (b) DWR
为了分析RC, DC和AC这三种耦合方式对不同网络规模传输效率的影响, 研究了数据包产生率临界值$ R_{\rm c} $随网络规模$ N $变化的趋势, 如图10所示. 图10(a)是采用SPR策略, 比较RC, DC和AC三种耦合方式$ R_{\rm c} $随网络规模N从200增加到2400变化的趋势, 其中网络平均度$ \left\langle k \right\rangle = 8 $. 可看出采用SPR策略AC方式的$ R_{\rm c} $值显示的数据包产生率十分稳定, 不随网络规模变化而变化. 采用DC方式的$ R_{\rm c} $值随着网络规模而变化, 变化幅度较大, 最终趋于稳定, 采用RC方式的$ R_{\rm c} $值随着网络规模变换较小. 可见, 随着网络规模变换DC方式很大程度上提高了网络的传输容量, 效果最佳. 图10(b)是采用DWR策略, 可看出RC, DC和AC三种耦合方式的$ R_{\rm c} $都随着网络规模的增加而变大, 但是采用AC方式的$ R_{\rm c} $值比DC和RC两种耦合方式的$ R_{\rm c} $值变化幅度更大, 效果更加明显, 能够更好地改善网络性能, 提高传输容量. 图 10 两种路由策略三种耦合方式$ R_{\rm c} $随网络规模N的变化 (a) SPR策略; (b) DWR策略 Figure10. Relationship between network size N and $ R_{\rm c} $ under two routing strategies with different coupling: (a) SPR; (b) DWR
衡量网络传输性能的参数还有平均路径长度$ \left\langle L \right\rangle $, 其定义为