China
Africa
Explore chapters and articles related to this topic
FAST TCP and Extensions
Published in Liansheng Tan, Resource Allocation and Performance Optimization in Communication Networks and the Internet, 2017
We now present simulation results based on a dumbbell network topology with five FAST TCP sender–receiver flows. For each flow i, we set the parameters γi = 0.01, αi = 50 packets, and di = 0.1 s. The capacity of the common link is set to 10 Mbps, and every packet size is set to 1 kbyte. We consider the random early detection (RED) [178] algorithm as the AQM in the router. The basic RED parameters are set to: minth = 150 packets, maxth = 350 packets, maxp = 0.1, and the target queue length qT = 300 packets. The simulation results for the sending rate of one of the five flows are shown in Figures 4.24 and 4.25. The other four flows exhibited similar behavior. We can observe that the modified flow-level model of FAST TCP reduces the oscillations of the sending rates.
An adaptive PI active queue management algorithm based on queue length
Published in Amir Hussain, Mirjana Ivanovic, Electronics, Communications and Networks IV, 2015
Hongcheng Huang*, Fan Yang, Shiwei Wang, Gaofei Xue
The AQM (Active Queue Management) techniques (Briscoe & Manner 2014) which adopt the queuing algorithm and pack discard strategy to manage the router buffer, so that it can adjust the data transmission rate to avoid a more serious congestion and improve network performance. Misra, V. established a TCP/ AQM cybernetic model (Misra et al. 2000), which allows people to study the AQM by using control theory. By using the instantaneous queue length, which is based on the TCP/AQM model as a measure of the congestion status, Hollot, C. designed the Hollot-PI algorithm (Hollot et al. 2001), which can eliminate the steady-state error effectively and keep the queue length in a remain stable. But as a result of the PI controller with fixed configuration parameters, the system response is slower, queues jitter is larger, and network throughput and bandwidth utilization are low.
Quality of Service in Switch/Routers
Published in James Aweya, Designing Switch/Routers, 2023
The use of techniques such as RED is commonly referred to as active queue management (AQM) [RFC7567]. AQM is the process of intentionally dropping packets randomly at a network node to signal TCP sources to slow down their data transmission rates. TCP is designed with congestion control mechanisms that allow a TCP source to adapt its data sending rate when packets are dropped or lost along the connection to the TCP destination [RFC5681]. Thus, network nodes (i.e., intermediate switches or routers) can exploit this TCP congestion control feature to intentionally drop packets “intelligently” when they experience local congestion or overutilization of local resources (e.g., link utilization, buffer oversubscription, processor overload, etc.).
Fractional-order PID control of tipping in network congestion
Published in International Journal of Systems Science, 2023
Jiajin He, Min Xiao, Yunxiang Lu, Zhen Wang, Wei Xing Zheng
A proliferation of progress have been made in the communication network, whether wired or wireless, but communication network always encounters performance bottlenecks due to the limited network load. Then, network congestion occurs when resource demands exceed the capacity (Welzl, 2005), which results in the decline of network performance, and even the collapse of the communication network. Therefore, novel network congestion control algorithms were developed to efficiently use all available capacity, and acquire higher bandwidth, higher reliability, and lower latency, e.g. dynamic explicit congestion notification (ECN) marking threshold algorithm (Lu et al., 2018) and active queue management (AQM) methods (Shen et al., 2022). Dual congestion algorithm models are commonly and widely used in congestion control (Srikant & Başar, 2004), which principally considers the dynamic nature of the link and the static nature of the source. Representative dual congestion models possessing one link and one delay were obtained, and the effect of delay on the model was investigated (Raina, 2005). However, many aspects of the simplified model are still worthy of further study to improve its efficiency and applicability, and so it is with the research of other network congestion models as well.
Frequency conditions for stable networked controllers with time-delay
Published in International Journal of Control, 2019
Johannes Nygren, Torbjörn Wigren, Kristiaan Pelckmans
The receiving node, or the base station, stores the data in a queue. The overall purpose of the control system is to manage the data volume in the queue, denoted as y(t), to avoid congestion and an empty queue. This is managed both by implementing the aforementioned bit-rate controller at the RNC site, and by implementing a simple active queue management (AQM) algorithm which actively imposes a packet-drop rate to the queue. Specifically, the AQM imposes a packet-drop rate which is proportional to the data volume, i.e. a drop rate of ϵy(t) where ϵ > 0 is some small number. The base station also sends queue level data y(t) to the RNC node through an uplink channel feedback control. The uplink has a time delay specified by Tul. The bit-rate emptying the queue is denoted wout(t).
Ant Colony Optimization based on Pareto optimality: application to a congested router controlled by PID regulation
Published in Systems Science & Control Engineering, 2018
Samira Chebli, Ahmed Elakkary, Nacer Sefiani
The main property of TCP is to guarantee reliable communication on the Internet. Congestion occurs when a link or node is carrying so much data that its quality of service deteriorates. To correct the problems with tail-drop buffer management, Floyd and Jacobsen introduced the concept of AQM, in which the routing nodes use a more sophisticated algorithm called RED to manage their queues (Chiu & Jain, 1989). Different algorithms were proposed and introduced into the Internet routers such as BLUE, REM or AVQ (Low et al., 2002; Ohsaki et al., 2009).