基于网络演算的确定性以太网性能研究与分析
发布时间：2018-10-24 17:18
【摘要】：在网络演算概念提出之后,就吸引了大量的学者去研究,使它得以不断发展和丰富,形成了两个理论体系——确定性和随机性网络演算。以太网技术有很多优点,如成本低廉、使用稳定和共享资源丰富等,以以太网技术为代表的通信技术,发展迅速,很快垄断了办公领域的使用网络,而且慢慢应用到工业控制和航空航天等领域。但对以太网技术的需求不断更新,不仅需要其高速率,还需要有实时性和确定性,由此,确定性以太网应运而生。确定性以太网还处于研究阶段,网络演算理论为其性能研究和分析提供了很好的理论依据。网络演算理论和确定性以太网技术各自都有很大的发展,但采用网络演算进行确定性以太网研究分析的工作很少。本论文的工作就是采用网络演算理论工具对确定性以太网进行性能研究与分析。首先,建立仿真通信模型,建立基本模型:数据模型(时间触发、速率受限和尽力而为数据模型)、进程模型、节点模型,由简单到复杂构建网络拓扑结构,提出了简单星型、级联和双通道拓扑结构,级联结构是星型结构的延伸,双通道是为了解决确定性以太网的容错问题。其次,建立三种网络拓扑结构的网络演算数学模型,基于简单星型拓扑结构,复杂拓扑结构在考虑了多种因素(如不同流向的数据)的同时,还使用了反卷积理论建立数学模型,得到三种拓扑结构下的网络性能参数的理论结果。最后,从端到端时延、吞吐量和丢包率三个方面将仿真结果和理论结果进行对比研究和分析,得出确定性以太网的性能结论。本论文经过研究和分析,由简单到复杂的三种拓扑结构中,时间触发数据是保证实时服务的,无丢包;速率受限数据的优先级较时间触发数据次之,其时延值在理论计算范围之内,除非数据量很大时,有少量丢包,一般无丢包;尽力而为数据是确定性以太网兼容的传统以太网的数据,不能保证其服务,允许其在排队等待处理时的时延超出理论值,之后又恢复正常。双通道拓扑结构在保证确定性和实时性的基础上,很好的发挥了容错功能。总的来说,确定性以太网拥有良好的性能,可促使其应用到航空航天或者工业控制等领域进行推广。
[Abstract]:After the concept of network calculus was put forward, it attracted a large number of scholars to study it, which allowed it to develop and enrich constantly, and formed two theoretical systems-deterministic network calculus and stochastic network calculus. Ethernet technology has many advantages, such as low cost, stable use and rich shared resources, etc. The communication technology represented by Ethernet technology has developed rapidly and soon monopolized the use of network in the office field. And slowly applied to industrial control and aerospace and other fields. However, the demand for Ethernet technology is constantly updated, which not only requires its high speed, but also needs real-time and certainty. Therefore, deterministic Ethernet emerges as the times require. Deterministic Ethernet is still in the research stage, network calculus theory provides a good theoretical basis for its performance research and analysis. The theory of network calculus and deterministic Ethernet technology have developed greatly, but the research and analysis of deterministic Ethernet based on network calculus is rare. The work of this thesis is to study and analyze the performance of deterministic Ethernet with network calculus theory. First of all, the simulation communication model is established, and the basic model is established: data model (time trigger, rate limited and best effort data model), process model, node model, from simple to complex to construct the network topology structure. Cascade and double channel topology, cascade structure is the extension of star structure, dual channel is to solve the fault tolerance problem of deterministic Ethernet. Secondly, the mathematical models of three network topologies are established. Based on the simple star topology, the complex topology takes into account a variety of factors (such as different flow data) at the same time. The mathematical model is established by using deconvolution theory, and the theoretical results of network performance parameters under three topologies are obtained. Finally, the simulation results and theoretical results are compared and analyzed from three aspects of end-to-end delay, throughput and packet loss rate, and the performance conclusions of deterministic Ethernet are obtained. Through the research and analysis of this thesis, among the three topologies from simple to complex, the time trigger data is guaranteed to serve in real time and has no packet loss, and the priority of rate limited data is second to that of time trigger data. The delay value is within the range of theoretical calculation, unless there is a small amount of packet loss when the amount of data is very large, generally there is no packet loss, but the best effort data is the data of the traditional Ethernet, which is compatible with deterministic Ethernet, and cannot guarantee its service. It allows the delay to exceed the theoretical value while queuing for processing, and then returns to normal. Dual-channel topology works well on the basis of certainty and real-time. In general, deterministic Ethernet has good performance and can be applied to aerospace or industrial control.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.11
本文编号：2292033
[Abstract]:After the concept of network calculus was put forward, it attracted a large number of scholars to study it, which allowed it to develop and enrich constantly, and formed two theoretical systems-deterministic network calculus and stochastic network calculus. Ethernet technology has many advantages, such as low cost, stable use and rich shared resources, etc. The communication technology represented by Ethernet technology has developed rapidly and soon monopolized the use of network in the office field. And slowly applied to industrial control and aerospace and other fields. However, the demand for Ethernet technology is constantly updated, which not only requires its high speed, but also needs real-time and certainty. Therefore, deterministic Ethernet emerges as the times require. Deterministic Ethernet is still in the research stage, network calculus theory provides a good theoretical basis for its performance research and analysis. The theory of network calculus and deterministic Ethernet technology have developed greatly, but the research and analysis of deterministic Ethernet based on network calculus is rare. The work of this thesis is to study and analyze the performance of deterministic Ethernet with network calculus theory. First of all, the simulation communication model is established, and the basic model is established: data model (time trigger, rate limited and best effort data model), process model, node model, from simple to complex to construct the network topology structure. Cascade and double channel topology, cascade structure is the extension of star structure, dual channel is to solve the fault tolerance problem of deterministic Ethernet. Secondly, the mathematical models of three network topologies are established. Based on the simple star topology, the complex topology takes into account a variety of factors (such as different flow data) at the same time. The mathematical model is established by using deconvolution theory, and the theoretical results of network performance parameters under three topologies are obtained. Finally, the simulation results and theoretical results are compared and analyzed from three aspects of end-to-end delay, throughput and packet loss rate, and the performance conclusions of deterministic Ethernet are obtained. Through the research and analysis of this thesis, among the three topologies from simple to complex, the time trigger data is guaranteed to serve in real time and has no packet loss, and the priority of rate limited data is second to that of time trigger data. The delay value is within the range of theoretical calculation, unless there is a small amount of packet loss when the amount of data is very large, generally there is no packet loss, but the best effort data is the data of the traditional Ethernet, which is compatible with deterministic Ethernet, and cannot guarantee its service. It allows the delay to exceed the theoretical value while queuing for processing, and then returns to normal. Dual-channel topology works well on the basis of certainty and real-time. In general, deterministic Ethernet has good performance and can be applied to aerospace or industrial control.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.11
【参考文献】
相关期刊论文 前1条
1 胡晓娅;朱德森;汪秉文;;基于网络演算的交换式以太网实时特性分析[J];计算机工程与应用;2008年01期
，本文编号：2292033
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2292033.html
