Nonlinear Neural Network Congestion Control Based on Genetic Algorithm for TCP/IP Networks

This paper is concerned with the design of robust active queue management (AQM) controllers for a class of TCP networks. In TCP/IP networks, the packet-dropping probability function is limited between 0 and 1. Therefore, a TCP network with AQM controller was modeled as a rate-based non-linear system with a saturated input. The nonlinear output feedback AQM controller is proposed for achieving the desired queue size and guaranteeing asymptotic stability of the operating point. The performance and effectiveness of the proposed control laws are validated for serial network scenarios through numerical simulations by Network Simulator-2 (NS-2). Simulation results confirm that the proposed scheme outperforms other AQM schemes.

In this paper, we use a previously developed nonlinear dynamic model of TCP to design Active Queue Management (AQM) controllers based on different structures of Neural Networks. We will introduce two types of Neural Networks controllers which are single neuron adaptive neural network (SANN) PID controller and the back propagation neural network (BPNN) PID controller. We illustrate the structures and the learning rules of both controllers. Then we give and compare their performances based on NS-2 simulation platform to support our designs. In the end, we prefer to use the BPNN controller as its better performance at dynamic network load.

<p><span>In this work, we used a new approach as active queue management (AQM) to avoid data congestion in TCP/IP networks. The new approach is PSO-PI controller which use the proportional-integral controller as a control unit and particle swarm optimization (PSO) algorithm as an optimization technique to improve the performance of the PI controller and therefore improving the performance of TCP/IP networks as a required goal. The optimization control (PSO-PI) is characterized by access to design and choosing the optimal parameters of </span>(K_1 and K_p) <span>to reach optimal solutions in a short way (fewer iterations). The implementation of the PSO algorithm is achieving by using the mathematical system model and M-file and SIMULINK in Mathlab program. Simulation results show good congestion management performance with PSO-PI controller better than the PI controller as AQM in TCP networks, and the proposed method was very fast and required few iterations.</span></p>

Anti-windup compensator for active queue management in TCP networks

The variable structure control algorithm has been proposed for active queue management (AQM) in TCP networks. In this paper we have considered the saturation constrain, i.e. marking probability p(t) /spl isin/ [0,1], which was omitted by many papers. By introducing the sliding model plane, this AQM algorithm can maintain an input rate around the target rate, while attempting to regulate the queue length. In this paper, we analyze the existence of the sliding mode plane in a nonlinear model. We intend to implement this algorithm in a per-flow queuing framework and we also guarantee that this algorithm can make the system converge to max-min fairness. We show the effectiveness of our algorithm compared to widely used AQM scheme RED through ns-2 simulations.

This paper proposes a new ns-3 model and presents the evaluation results for Proportional Integral controller Enhanced (PIE), a recently designed Active Queue Management (AQM) mechanism to address the problem of bufferbloat. The problem of bufferbloat arises due to the presence of large unmanaged buffers in routers. This leads to high queuing latency and significantly degrades the performance of timesensitive and interactive traffic. AQM mechanisms that aim to address the problem of bufferbloat try to achieve an optimal trade-off between high link utilization and low mean queue length. PIE is a lightweight AQM mechanism that tries to achieve the same. To our knowledge, ns-3 network simulator does not have a model for simulating PIE. Hence, in this paper, we implement a ns-3 model for PIE, and show that the results obtained from it are in line with those obtained from the ns-2 model of PIE, implemented by its authors.

This paper presents the application of a neural network controller to the problem of active drag reduction in a fully turbulent 3D fluid flow regime. Based on a successful yet infeasible previous active control scheme, we trained a neural network to mimic the control law using only surface spanwise shear stress measurements. We then demonstrate the ability of a neural controller implemented in an adaptive inverse model scheme to maintain a drag-reduced flow in a fully turbulent fluid simulation. By observing the weights of the on-line controller, a simple control law that predicts actuations proportional to the spanwise derivative of the spanwise shear stress is derived. Finally we examine the amount of parameter variation that may be required for a physical implementation of linear and nonlinear neural controllers.

The lack of study among the economic forecasting literature that can empirically proves the hypothesis of being more powerfulness of dynamic neural networks in comparison with the static neural networks models for forecasting, is the most important motivation of this study. In this paper, the utilization of NNARX as a nonlinear dynamic neural network model, ANN as a nonlinear static neural network model and ARIMA as a linear model were compared to forecast poultry retail price. As a case study on Iranian poultry retail price, we compare forecast performance of these models for three forecasts (1, 2 and 4 week ahead). Results show that NNARX and ANN models outperform ARIMA model, and also NNARX model outperforms ANN model for all three forecasts.

Design of a fuzzy controller for active queue management

DC-DC converters are often used in electrical systems to keep the output voltage constant. Boost converters are utilized for a variety of purposes, including regenerative braking of direct current motors, portable device applications, and regulated power supplies. Buck converters, on the other hand, are used in sophisticated communications, data communication, and self-regulating power supplies. One of the most crucial elements in the power conversion process is managing the DC-DC converters. This study aims to find out which nonlinear controller, fuzzy or neural network, works best when the output load or boost converter characteristics change. So, in this, we create a boost converter controller using a fuzzy and neural network. The fuzzy controller for this study's converter employs a standard set of rules, whereas the neural network controller employs two hidden layer networks. MATLAB software is then used to reconstruct both controllers. The simulation results show that the fuzzy logic controller has a very long transient and settling period with no steady-state error in both transient and steady-state situations. On the other hand, the neural network controller has a short transient and settling period with a steady-state error.

Augmenting effectiveness of control loops of a PMSG (permanent magnet synchronous generator) based wind energy conversion system by a virtually adaptive PI (proportional integral) controller

In this study, the disturbance and uncertainty on nonlinear and time varying systems as Active Queue Management (AQM) is analyzed. Many of AQM schemes have been proposed to regulate a queue size close to a reference level with the least variance. We apply a normal range of disturbances and uncertainty such as variable user numbers, variable link capacity, noise, and unresponsive flows, to the three AQM methods: Random Early Detection (RED), Proportional-Integral (PI) and Improved Neural Network (INN) AQM. Then we examine some important factors for TCP network congestion control such as queue size, drop probability, variance and throughput in NS-2 simulator, and then compare three AQM algorithms with these factors on congestion conditions. We present the performance of the INN controller in desired queue tracking and disturbance rejection is high.

An Active Queue Management (AQM) is a proactive scheme that controls the network congestion by avoiding the congestion before it happened. When implementing AQM in wireless networks several contemporary issues must be considered, such as interference, collisions, multipath-fading, propagation distance, shadowing effects and route failure, and whether the wireless networks is WLAN or other type. Therefore, the needs for AQM algorithm that can perform in WLAN network as good and efficient as in wired network become so crucial. This paper proposes a new AQM algorithm called Hybrid Fair AQM (HF-AQM) that can achieve better fairness and higher utilization in WLAN environment by hybridizing queue delay and input rate to measure network congestion. The simulation results show that HF-AQM manages to gain a higher fairness level in comparison to several types of AQM schemes with higher link utilization due to the adaptation and hybridization that been used in its congestion indicator, control function and feedback function.

This paper demonstrates that fuzzy logic systems and feedforward neural networks are equivalent in essence. First, we introduce the concept of interpolation representations of fuzzy logic systems and several important conclusions. We then define mathematical model for rectangular wave neural networks and nonlinear neural networks. With this definition, we prove that nonlinear neural networks can be represented by rectangular wave neural networks. Based on this result, we prove the equivalence between fuzzy logic systems and feedforward neural networks. This result provides us a very useful guideline when we perform theoretical research and applications on fuzzy logic systems, neural networks, or neuro-fuzzy systems.

The virtual rate control (VRC) algorithm has been proposed for active queue management (AQM) in TCP networks. This algorithm uses an adaptive rate control instead of queue length control in order to respond quickly to traffic change with high utilization and small loss. By introducing the notion of virtual target rate, the VRC algorithm can maintain an input rate around the target rate, while attempting to regulate the queue length. In this paper, we analyze the stability of the VRC algorithm in a linearized model. From the results of our analysis, we provide a design guideline for the system to remain stable. We show the validity of our analysis and the effectiveness of the VRC algorithm compared to RED, PI, REM and AVQ algorithms through ns-2 simulations.