Network Control Architecture Research Articles

Radial basis function network architecture for nonholonomic motion planning and control of free-flying manipulators

This paper considers a problem of nonholonomic motion planning. A practical paradigm for planning and stabilization of motion in a class of multivariate nonlinear (nonholonomic) systems is presented and applied to a planar free-floating manipulator system. The controller architecture designed in the paper is based on the radial basis function approximation of an optimal control program for any desired motion. This architecture also incorporates a sampled-data feedback stabilization algorithm. The proposed control technique overcomes certain problems associated with other control approaches available for nonholonomic systems. The presented simulation results reveal a promising potential of the proposed control paradigm. This paradigm can be extended to a broader class of nonlinear control problems.

Neural network architecture for process control based on the RTRL algorithm

AbstractNeural‐network‐based control schemes are generally designed by replacing standard elements of the classic control schemes by feedforward neural networks. The introduction of discrete time recurrent networks, which are inherently dynamic systems, into those schemes can simplify the design of neural controllers. The concept of applying recurrent networks in indirect adaptive control schemes is described. A combined network cluster consisting of the control network and the model network is constructed to allow the use of the real‐time recurrent learning algorithm. To demonstrate the feasibility of the method two simulation examples are presented.

Neural network architecture for trajectory generation and control of automated car parking

This paper describes the development of a control system to support an automated parking mode in driving passenger cars. By using recent advances in the artificial neural network technology and a combination of linear feedback and nonlinear feedforward control, we propose a novel architecture for the parking motion controller. The paper presents the results of the controller design and analysis, including parking problem analysis, stability analysis for the feedback controller, formulation and optimal solution of the parking trajectory planning problem, and design of a parking motion planning architecture based on a radial basis function network. Two general cases of backward parking considered in this work are emulated using the proposed controller. The emulation results reveal high efficiency of the presented approach and demonstrate that the proposed system can be implemented on a typical passenger car.

Taming Xunet III

An architecture for network management and control for emerging wide-area ATM networks is presented. The architecture was implemented on XUNET III, a nationwide ATM network deployed by AT&T. The Xunet network management system is based on the OSI standards and includes configuration, fault and performance management. An OSI agent resides at every switching node. Its capabilities include monitoring of cell level quality of service in real time and estimation of the schedulable region. The complexity and accuracy of real-time monitoring functionalities is investigated. To provide realistic traffic loads, distributed traffic generation systems both at the cell and call level have been implemented. In order to study the trade-off between the network transport and signalling system we have implemented a virtual path signalling capability. Our experiments show that the ability of a network to admit calls is limited by two distinct factors: the capacity of the network and the processing power of the signalling system. Depending on the bandwidth requirement of calls, the limit of one or the other will be first reached. This is a key observation, unique to broadband networks.

Trainable neural network for mechanically flexible systems based on nonlinear filtering

A trainable neural network controller architecture is investigated for motion control systems involving significant distributed mechanical flexibility. In general, this neural network based controller can be trained on-line to learn the behavior of another controller which serves as the teacher implementing algorithmic or nonalgorithmic control law. To address potential of such a scheme in real time, the weight adjustments of the network connection strengths and biases are based on a nonlinear filtering adaptation rule, extended Kalman filter, to reduce training time and achieve fast convergence rate. Computer simulations are performed to test the performance of this training algorithm.

A distributed control architecture of high-speed networks

A control architecture for a high-speed packet-switched network is described. The architecture was designed and implemented as part of the PARIS (subsequently plaNET and BBNS) networking project at IBM. This high bandwidth network for integrated communication (data, voice, video) is currently operational as a laboratory prototype. It will also be deployed within the AURORA Testbed that is part of the NSF/DARPA gigabit networking program. The high bandwidth dictates the need for specialized hardware to support faster packet handling for both point-to-point and multicast connections. A faster and more efficient network control is also required in order to support the increased number of connections and their changing requirements with time. The new network control architecture presented exploits specialized hardware, thereby enabling tasks to be performed faster and with less computation overhead. In particular, since control information can be distributed quickly using hardware packet handling mechanisms, decisions can be made based upon more complete and accurate information. In some respects, this has the effect of having the benefits of centralized control (e.g., easier bandwidth resource allocation to connections), while retaining the fault tolerance and scalability of a distributed architecture. >

Integrated network management and control for interactive video dial tone networks

The emerging interactive video dial tone (VDT) services pose some unique requirements for real-time network monitoring and control. To support the roll-out of VDT-based networks and services, GTE is developing new network management applications on its TONICS (Telephone Operations Network Integrated Control System) platform. TONICS's applications are embedded in a high availability software platform providing distributed processing. The article describes the real-time management requirements of VDT networks, provides an overview of the TONICS integration platform, outlines a network management and control architecture for VDT networks, illustrates example VDT management applications, and discusses the deployment issues. >

Fast restoration of ATM networks

Asynchronous transfer mode (ATM) is now well recognized as the fundamental switching and multiplexing technique for future broadband ISDN. As these networks will be increasingly relied upon for providing a multitude of integrated voice, data, and video services, network reliability is a key concern. There are several intrinsic features of ATM networks that could potentially be exploited to provide improved restoration techniques, beyond those established for synchronous transfer mode (STM) networks, such as digital cross-connect restoration or self-healing rings. These features include ATM cell level error detection, inherent rate adaptation and nonhierarchical multiplexing. The authors explore the use of these features in developing fast restoration strategies for ATM networks. In particular, they address: (1) ATM error detection capabilities for enhanced failure detection, (2) network rerouting strategies, (3) spare capacity allocation, and (4) network control architecture and related implementation aspects. Their findings suggest that fast network span failure detection and bandwidth-efficient rerouting capabilities can be combined to develop restoration strategies for ATM networks with significantly greater performance-cost ratios when compared to existing STM network restoration strategies.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Two-stage neural network architecture for feedback control of dynamic systems

In this paper we investigate system identification and the design of a controller using neural networks. A two-stage neural network design for controllers using single-layer structures with functional enhancements is introduced. This neural network architecture allows the design of a controller with less a priori knowledge about the plant as well as allowing for nonlinear plants. The paper also addresses the special characteristics and problems concerning the use of neural networks in control and demonstrates their performance by showing the successful implementation of a nonlinear control example via simulation.

A CMAC neural-network-based algorithm for the kinematic control of a walking machine

In this paper a hierarchical, neural network control architecture of a walking machine is proposed. The neural network is based on the theory of the Cerebellum Model Articulation Controller (CMAC) which is a neuromuscular control system. Some preliminary studies of kinematic control and gait synthesis are presented to demonstrate the effectiveness of the CMAC neural network. After having been trained to learn the multivariable, nonlinear relationships of the leg kinematics and gaits, CMAC is utilized to perform feedforward kinematic control of a quadruped in straight-line walking and step climbing. Simulation examples are provided and discussed. This algorithm can be extended to control other highly nonlinear processes which are hierarchical in nature and cannot be modeled by mathematical equations.

Process control via artificial neural networks and reinforcement learning

In the training of artificial neural networks, reinforcement learning substitutes a qualitative binary target of “success” or “failure” for the quantitative error criterion of supervised learning. By this method of learning control can be established for the special case of operation in which no objective function exists. If no model whatsoever exists of the dynamics of a chemical process, it still may be possible to train an artificial neural network to control the process much as a human being would learn by trial and error. We describe a network architecture for process control, and explain quantitatively how the weights on the connections in the network can be adjusted to yield the desired control action. An example of a nonlinear CSTR is used to illustrate the performance of the proposed net and compare it with that of a tuned PID controller. The net can be trained to meet almost any criteria selected for control as long as the criteria can be expressed in the form of inequality constraints, but requires extensive, and perhaps excessively long, training on a serial computer to do so.

Neural network architecture for crossbar switch control

A Hopfield neural network architecture for the real-time control of a crossbar switch for switching pockets at maximum throughput is proposed. The network performance and processing time are derived from a numerical simulation of the transitions of the neural network. A method is proposed to optimize electronic component parameters and synaptic connections, and it is fully illustrated by the computer simulation of a VLSI implementation of 4*4 neural net controller. The extension to larger size crossbars is demonstrated through the simulation of an 8*8 crossbar switch controller, where the performance of the neural computation is discussed in relation to electronic noise and inhomogeneities of network components. >

Neural network architecture for adaptive system modeling and control

A computing architecture for adaptive control and system modeling based on computational features of nonlinear discrete neural networks is proposed. These features are massively parallel and distributed structures for signal processing, with the potential for ever-improving performance through dynamical learning. In this paper the proposed delayed-input delayed-state network architecture and the general training scheme are described. A solution for the problem of analog signals representation by a binary neural network is suggested. Illustrative simulation results are promising and show good and robust performance in various cases.

Bandwidth management: A congestion control strategy for broadband packet networks — characterizing the throughput-burstiness filter

A congestion/flow/error control architecture for broadband packet networks, based on a “core” network congestion control strategy called Bandwidth Management (BWM), is described. Associated with this control architecture is a set of traffic/performance analysis modules for quantitatively assessing the overall control performance. A particular module, that quantitatively characterizes a “throughput-burstiness filter” associated with BWM, is studied through an analytically tractable traffic model, and results of analysis of this model are used to illustrate the operation of the BWM controls.

The SPS and LEP control network architecture

This paper describes the local-area network for the control of SPS and LEP. It presents the network architecture, its topology and shows the criteria that contributed to the selection of the transmission medium and its implementation. It also describes different aspects of network management, the software utilities for the detection of faults and the reconfiguration process.

Control architecture for next-generation communication networks based on distributed databases

Intelligent routing control is defined as the process in which the network interrogates the databases containing the relationships between logical numbers, such as personal or information identifiers, and physical addresses in the transport network to find the terminal having the information required to process a user request. The routing control system presented uses distributed databases, each of which manages a switching system and all of which are connected through high-speed signalling networks separate from the transport network. If the requested physical address cannot be found in one database, search requests are distributed at the same time to all other databases. For up to 100 million subscribers, the routing control system can find a physical address within 1 s when each database uses ten memories accessed at 200 ns with an interdatabase linkage speed of 14 Mb/s. >

Neural network architecture for robot hand control

A robot hand control system called GeSAM, which is under development at the University of Southern California, is described. The goal of the GeSAM architecture is to provide a generic robot hand controller that is based on a model of human prehensile function. It focuses on the relationship between geometric object primitives and the ways a hand can perform prehensile behaviors. It is shown how the relationship between object primitives and a useful set of grasp modes can be learned by an adaptive neural network. By adding training points as necessary, system performance can be improved, avoiding the tedious job of computing every relationship individually.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Neural network architecture for control

Two important computational features of neural networks are associative storage and retrieval of knowledge, and uniform rate of convergence of network dynamics independent of network dimension. It is indicated how these properties can be used for adaptive control through the use of neural network computation algorithms, and resulting computational advantages are outlined. The neuromorphic control approach is compared to model reference adaptive control on a specific example. It is shown that the utilization of neural networks for adaptive control offers definite speed advantages over traditional approaches for very-large-scale systems.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Design of the Distributed Control System for Beam Lines at the Photon Factory

A distributed control system has been designed to control all beam lines around the 2.5 GeV storage ring at the Photon Factory. It is a local area network composed of outlying nodes and a central node which are interconnected in a star configuration via optical fiber links, with a communication rate of 9600 baud. Each beam line is controlled by its own outlying node, independently of adjacent beam lines. The system is flexible and expandable enough to accommodate itself to variations in beam line features, according to experimental circumstances. In addition, when failure of a beam line occurs in the network, the system does not disrupt the rest of the beam line controls by disconnecting the defective line. The control scheme and network architecture are described, as well as the hardware configuration of the system.

"Local Control Network and Internetwork ISO-OSI Refernece Model"

The paper describes a new local control network architecture. The new control network is totally distributed and redundantly hardware and software structured, based on a bus configuration and on CSMA/CD media access control. The architecture of the control structure and of the data communications structure for both Local Network and Internetwork is discussed in comparison with ISO-OSI and Local Area Network IEEE Standard 802 (Draft) Reference Models. A previous paper [4] dealt with the physical implementation of this concept. The present paper is more software structure oriented.