Fixed Network Topology Research Articles

Neural quantile optimization for edge–cloud networking

We seek the best traffic allocation scheme for the edge–cloud networking subject to SD-WAN architecture and burstable billing. First, we formulate a family of quantile-based integer programming problems for a fixed network topology with random parameters describing the traffic demands. Then, to overcome the difficulty caused by the discrete feature, we generalize the Gumbel-softmax reparameterization method to induce an unconstrained continuous optimization problem as a regularized continuation of the discrete problem. Finally, we introduce the Gumbel-softmax sampling neural network to solve optimization problems via unsupervised learning. The neural network structure reflects the edge–cloud networking topology and is trained to minimize the expectation of the cost function for unconstrained continuous optimization problems. The trained network works as an efficient traffic allocation scheme sampler, outperforming the random strategy in feasibility and cost value. Besides testing the quality of the output allocation scheme, we examine the generalization property of the network by increasing the time steps and the number of users. We also feed the solution to existing integer optimization solvers as initial conditions and verify the warm-starts can accelerate the short-time iteration process. The framework is general, and the decoupled feature of the random neural networks is adequate for practical implementations.

Optimal Coordination Of Multiple Setting Group Type Overcurrent Relays

Relay coordination optimization issues are multi-objective, multi-constrained problems that aim to reduce each relay's operation time while satisfying all relevant constraints. Fulfillment of the coordination time interval (CTI) and time limitations for relay operations are the coordination constraints. To solve coordination optimization issues, decision-making variables are defined such that the objective function is minimized with due satisfaction of necessary constraints. The majority of the most recent research that is currently available defines relay coordination for fixed network topologies, but in practice, it varies for a variety of reasons, including maintenance and element failures. Over current relays now have the capability of storing multiple relay settings but the number of relay settings that can be stored in relays is far less than the possible number of network configurations, which is why a suitable clustering technique such as K-Means can be used to cluster out various network topologies using a proper clustering index. Because both the objective function and the constraints are time dependent, time dependent attributes such as average relay operating time are used as the clustering index instead of the conventional index fault current deviation.

Physical-assisted multi-agent graph reinforcement learning enabled fast voltage regulation for PV-rich active distribution network

Active distribution network is encountering serious voltage violations associated with the proliferation of distributed photovoltaic. Cutting-edge research has confirmed that voltage regulation techniques based on deep reinforcement learning manifest superior performance in addressing this issue. However, such techniques are typically applied to the specifically fixed network topologies and have insufficient learning efficiency. To address these challenges, a novel edge intelligence, featured by a multi-agent deep reinforcement learning algorithm with graph attention network and physical-assisted mechanism, is proposed. This novel method is unique in that it includes the graph attention network into reinforcement learning to capture spatial correlations and topological linkages among nodes, allowing agents to be “aware” of topology variations caused by reconfiguration real time. Furthermore, employing a relatively exact physical model to generate reference experiences and storing them in a replay buffer enables agents to identify effective actions faster during training and thus, greatly enhances the efficiency of learning voltage regulation laws. All agents are trained centralized to learn a coordinated voltage regulation strategy, which is then executed decentralized based solely on local observation for fast response. The proposed methodology is evaluated on the IEEE 33-node and 136-node systems, and it outperforms the previously implemented approaches in convergence and control performance.

Distributed adaptive fuzzy fault-tolerant control for multi-agent systems with node faults and denial-of-service attacks

This paper studies the distributed adaptive fuzzy fault-tolerant control for multi-agent systems (MASs) with node faults and denial-of-service (DoS) attacks. On account of DoS attacks, the fixed network topology is changed into switching topologies, which brings about many difficulties in designing fault-tolerant controllers. In addition, node faults and nonlinearity terms make it more challenging to analyze the stability of MASs with DoS attacks. To cope with the nonlinear terms in MASs, a fuzzy logic systems method is provided. Also, based on the switching mechanism, the average dwell time technique, and the Lyapunov stability theory, a distributed adaptive fuzzy fault-tolerant controller is proposed, and the bounded synchronization of the MASs is achieved. A simulation example is used to validate the efficiency of the proposed approach.

Toward Attack-Resistant Route Mutation for VANETs: An Online and Adaptive Multiagent Reinforcement Learning Approach

Vehicular Ad hoc Networks (VANETs) are prone to packet drop attacks because of their inherent distributed architecture and dynamic topology. Existing security schemes mainly focus on multi-path and trust-based routing. Unfortunately, the former causes high energy consumption and the latter requires trust assessment, which is not easy to implement in practice. Route mutation (RM) is emerging as an active defense technology that changes routes periodically. Traditional RM is conceived for fixed network topologies, and needs a centralized controller, so that it cannot be applied to VANETs. Therefore, the present contribution investigates RM in VANETs by proposing a Grid-based extended Joint Action Learning approach (Grid-eJAL). To the best of our knowledge, this is the first contribution that designs an online and adaptive multi-agent reinforcement learning (MARL) for RM to mitigate attacks in VANETs. Differently from existing MARL schemes, Grid-eJAL allows vehicles to share parameters to accelerate the convergence speed of learning. In Grid-eJAL, the area of interest is split in equally sized grids and, when a vehicle transmits packets, the next hop with the minimum angle of mobility is selected within the grid considered as optimal by the learning policy. The convergence of Grid-eJAL is proved theoretically. Finally, extensive simulation results highlight the effectiveness of Grid-eJAL compared to representative state-of-the-art solutions.

Straight Forward Constructive Deep Learning Neural Network (SFC-DLNN) Algorithm

Straight Forward Constructive Deep Learning Neural Network (SFC-DLNN) algorithm is a new architecturebased algorithm for artificial neural networks. Rather than simply adjusting the weights in a fixed topology network, SFC-DLNN starts with a minimal topology (perceptron), then builds up their network by gradually trains and adds new nodes one by one, creating multiple layers’ network. Once a unit has been added to the network, the weights of the new architecture are generated. This unit then stands as a permanent detector of features in the network, and a more complex feature space is then created where the data is likely to be linearly separable. The SFC-DLNN algorithm has many advantages over existing ones: it has good learning speed, the network determines its topology size, and the structures it has built is retained after the training stage. We obtain from our built model (SFC-DLNN) an accuracy and specificity of 83:5% from a simulated data set using the uniform distribution. This is not the best but is enough to approve the model prediction capacity

Decentralized control and state estimation of linear time‐periodic systems

AbstractThe main contributions of this article are the design of a decentralized controller and state estimator for linear time‐periodic systems with fixed network topologies. The proposed method to tackle both problems consists of reformulating the linear periodic dynamics as a linear time‐invariant system by applying a time‐lifting technique and designing a discrete‐time decentralized controller and state estimator for the time‐lifted formulation. The problem of designing the decentralized estimator is formulated as a discrete‐time Kalman filter subject to sparsity constraints on the gains. Two different algorithms for the computation of steady‐state observer gains are tested and compared. The control problem is posed as a state feedback gain optimization problem over an infinite‐horizon quadratic cost, subject to a sparsity constraint on the gains. An equivalent formulation that consists in the optimization of the steady‐state solution of a matrix difference equation is presented and an algorithm for the computation of the decentralized gain is detailed. Simulation results for the practical cases of the quadruple‐tank process and an extended 40‐tank process are presented that illustrate the performance of the proposed solutions, complemented with numerical simulations using the Monte Carlo method.

Enabling Scalable Routing in Software-Defined Networks With Deep Reinforcement Learning on Critical Nodes

Traditional routing schemes usually use fixed models for routing policies and thus are not good at handling complicated and dynamic traffic, leading to performance degradation (e.g., poor quality of service). Emerging Deep Reinforcement Learning (DRL) coupled with Software-Defined Networking (SDN) provides new opportunities to improve network performance with automatic traffic analysis and policy generation. However, existing DRL-based routing solutions usually rely on all node information to make routing decisions for the network and hence are both hard to converge in large networks and vulnerable to topology changes. In this paper, we propose ScaleDeep, a scalable DRL-based routing scheme for SDN, which improves the routing performance and is resilient to topology changes. Essentially, ScaleDeep takes advantage of partial control on network nodes and DRL. We select a set of critical nodes from a network as driver nodes, which can simulate the entire network operation, based on the control theory. By observing the traffic variation on the driver nodes, DRL dynamically adjusts some link weights for a weighted shortest path algorithm to change the routing paths and improve the routing performance. Limiting the control on driver nodes improves the convergence ability of DRL and reduces the dependency of the DRL agent on the fixed network topology. To validate the performance of ScaleDeep, we conduct packet-level simulations on different topologies. The results show that ScaleDeep outperforms existing DRL-based schemes by reducing the average flow completion time by up to 36% and exhibiting better robustness against minor topology changes.

MDS Coded Task Offloading in Stochastic Wireless Edge Computing Networks

Coded computation has attracted great interests as a promising technique to cope with straggling computing nodes in mobile edge computing (MEC) networks. Contrary to the existing coded computation schemes developed with a fixed network topology, this paper studies a MDS coded computation for random networks. Specifically, we put forth maximum distance separable (MDS) coded task offloading and investigate its MDS coded computing gain by deriving the average successful retrieval probability with stochastic geometry in random wireless edge computing networks, where it encodes the original task into multiple equal and small sized MDS coded sub-tasks and offloads their subset to edge computing nodes for computation. We also identify a tradeoff between the latency in processing a sub-task at an edge node and the minimal number of edge nodes required to retrieve the original task output, according to the size of MDS coded sub-tasks. To efficiently control the tradeoff, we determine the desirable size of MDS coded sub-tasks in a semi-closed form to maximize the average successful retrieval probability for regime 1 and regime 2 networks, which correspond to the cases that communication latency is negligible compared to computation latency and that computation latency is negligible compared to communication latency, respectively, and develop an efficient algorithm with low search complexity for a general environment. Our numerical results reveal that the proposed scheme outperforms the other conventional task offloading schemes such as partial task offloading and replication task offloading in terms of average successful retrieval probability.

Joint apron-runway assignment for airport surface operations

Airport surface operations, such as off-block control, taxi routing, and runway sequencing, are typically confined to a fixed network topology with given origin–destination (O-D) pairs for departures and arrivals. Reconfiguring the O-D distribution of flights by actively assigning their aprons and runways is a potentially effective measure to maximize the utilization of network capacity. To date this has not been investigated in the literature due to its complexity involving operational constraints, assignment rules, and different stakeholders. This paper demonstrates the significant potential of O-D reconfiguration in improving surface network efficiency by proposing a joint apron-runway assignment framework for pre-tactical operations. This is underpinned by a comprehensive review of apron and runway assignment rules, including constraints and preferences, and an elaborated optimization scheme that encompasses lexicographic and iterative approaches along with temporal buffers to absorb uncertainties in pre-tactical operations. The proposed apron-runway assignment is implemented and assessed in a case study at Beijing Capital International Airport. An airport cellular automata simulator is employed for quantitative evaluation, and qualitative assessment is based on interviews with subject matter experts. Compared to the current operations, the proposed apron-runway assignment is very promising, with reductions in total taxiing distance, average taxiing time, taxiing conflicts, runway queuing time, and fuel consumption by 15.5% (Table 3), 6.2%, 19.8%, 17.6% (Table 4) and 6.6% (Table 5) respectively; gated assignment is increased by 11.8% (Table 4). Importantly, these benefits remain robust to unforeseen flight delays, as demonstrated in a sensitivity analysis.

A Multi-Scale Method for Distributed Convex Optimization with Constraints

This paper proposes a multi-scale method to design a continuous-time distributed algorithm for constrained convex optimization problems by using multi-agents with Markov switched network dynamics and noisy inter-agent communications. Unlike most previous work which mainly puts emphasis on dealing with fixed network topology, this paper tackles the challenging problem of investigating the joint effects of stochastic networks and the inter-agent communication noises on the distributed optimization dynamics, which has not been systemically studied in the past literature. Also, in sharp contrast to previous work in constrained optimization, we depart from the use of projected gradient flow which is non-smooth and hard to analyze; instead, we design a smooth optimization dynamics which leads to easier convergence analysis and more efficient numerical simulations. Moreover, the multi-scale method presented in this paper generalizes previously known distributed convex optimization algorithms from the fixed network topology to the switching case and the stochastic averaging obtained in this paper is a generalization of the existing deterministic averaging.

Matrix-weighted consensus for multi-agent systems with sampled-data

This paper researches the consensus issue for multi-agent systems on matrix-weighted directed fixed and undirected switching network topologies by sampled data control method which saves resources and is more practical. Using the sampled information, the distributed control laws are designed under two network topologies, respectively. Under directed fixed network topology, the consensus conditions based on the sampling period and the eigenvalues of Laplacian matrix are deduced by matrix theory and analysis theory. Under undirected switching network topology, by using Lyapunov stability theory, the consensus conditions based on the sampling period and switched network topologies are built. Lastly, two simulation examples are offered to verify the validity of the obtained results.

Greedy initialization for distributed persistent monitoring in network systems

This paper considers the optimal multi-agent persistent monitoring problem defined for a team of agents on a set of nodes (targets) interconnected according to a fixed network topology. The aim is to control this team so as to minimize a measure of overall node state uncertainty evaluated over a finite time interval. A class of distributed threshold-based parametric controllers has been proposed in prior work to control agent dwell times at nodes and next-node destinations by enforcing thresholds on the respective node states. Under such a Threshold Control Policy (TCP), an on-line gradient technique was used to determine optimal threshold values. However, due to the non-convexity of the problem, this approach often leads to a poor local optima highly dependent on the initial thresholds used. To overcome this initialization challenge, we develop a computationally efficient off-line greedy technique based on the asymptotic analysis of the network system. This analysis is then used to generate a high-performing set of initial thresholds. Extensive numerical results show that such initial thresholds are almost immediately (locally) optimal or quickly lead to optimal values. In all cases, they perform significantly better than the locally optimal solutions known to date.

Theory and Experiments on Enclosing Control of Multi-Agent Systems

This paper proposes a control strategy called enclosing control. This strategy can be described as follows: the followers design their control inputs based on the state information of neighbor agents and move to specified positions. The convex hull formed by these followers contains the leaders. We use the single-integrator model to describe the dynamics of the agents and proposes a continuous-time control protocol and a sampled-data based protocol for multi-agent systems with stationary leaders with fixed network topology. Then the state differential equations are analyzed to obtain the parameter requirements for the system to achieve convergence. Moreover, the conditions achieving enclosing control are established for both protocols. A special enclosing control with no leader located on the convex hull boundary under the protocols is studied, which can effectively prevent enclosing control failures caused by errors in the system. Moreover, several simulations are proposed to validate theoretical results and compare the differences between the three control protocols. Finally, experimental results on the multi-robot platform are provided to verify the feasibility of the protocol in the physical system.

X-NEST: A Scalable, Flexible, and High-Performance Network Architecture for Distributed Machine Learning

In a large-scale distributed machine learning system, the interconnection network between computing devices has an important impact on performance in the training of neural network models. The current expansion of training data and model size has led to a rapid increase in the number of computing devices used in distributed machine learning systems, which places higher demands on network scalability. In addition, the synchronization algorithms used for data exchange between computing devices have different communication topologies, and traditional electrical networks have difficulty matching them due to their fixed network topology. Neural network models and model partitioning methods can also affect the amount of communication between devices, but the overprovisioned bandwidth of traditional electric networks incurs unnecessary costs. To address these issues, we propose a scalable, flexible, and high-performance network architecture called X-NEST. The flexibility of optical switching devices allows X-NEST to dynamically change its topology and the number of links between devices according to traffic pattern variations, thereby improving network performance and resource utilization. Although changes in the connection relationships between devices depend on the controller, the simple and flexible control plane of X-NEST can quickly respond to network communication requirements. Extensive analytical simulations using different traffic patterns demonstrate that X-NEST copes well with the communication characteristics of various synchronization algorithms.

The optimal synchronizability of complex networks with saturated coupling strength

For networks with fixed network topology, when the total coupling strength between nodes is limited and the coupling strength between nodes is saturated, the global optimization algorithms including genetic algorithm (GA) and particle swarm optimization (PSO) algorithm are used to adjust the coupling strength between nodes to improve the synchronizability of the network, respectively. Simulation results show that in WS small-world network, when the edge betweenness centrality of the edge is large, the coupling strength of the edge after optimization is greater. Furthermore, compared with GA, PSO has better performance.

Bipartite Tracking Consensus of Generic Linear Agents With Discrete-Time Dynamics Over Cooperation-Competition Networks.

This article addresses the bipartite tracking consensus for a set of mobile autonomous agents over directed cooperation-competition networks. Here, cooperative and competitive interactions among the agents are described by positive and negative edges of the directed network topology, respectively. Both fixed and switching network topologies are considered. For the case with fixed network topology, the matrix product technique is utilized to derive the convergence result. For the case with switching network topologies, some key results related to the composition of binary relations are the main technical tools of analyzing the error system. In addition, the upper bound for the spectral radius of the system matrix is given to ensure the convergence of the system even if the single uncoupled system is strictly unstable. The applicability of the derived results is verified through two simulation experiments.

Disconnection-aware Attack Detection in Networked Control Systems

This study deals with security issues in dynamical networked control systems. The goal is to establish a unified framework of the attack detection stage, which includes the four processes of monitoring the system state, making a decision based on the monitored signal, disconnecting the corrupted subsystem, and operating the remaining system during restoration. This paper, in particular, considers a disconnection-aware attack detector design problem. Traditionally, observer-based attack detectors are designed based on the system model with a fixed network topology and cannot cope with a change of the topology caused by disconnection. The disconnection-aware design problem is mathematically formulated and a solution is proposed in this paper. A numerical example demonstrates the effectiveness of the proposed detector through an inverter-based voltage control system in a benchmark model.

Novel Distributed Dynamic Backbone-based Flooding in Unstructured Networks

Resource discovery on different unstructured and dynamic networks such as grid, peer-to-peer, and cloud networks is an inevitable challenging issue. The primary method for resource discovery on the unstructured networks is flooding a query on the network. All existing flooding algorithms for unstructured networks generate almost high additional duplicated queries. This high duplication of the unstructured networks causes a lot of network traffic. This paper, therefore, proposes a novel flexible Distributed Dynamic backbone-based Flooding (DDBF) algorithm for distributed unstructured networks. This paper explores Grid middleware, Peer-to-Peer (P2P) paradigm, and cloud networks resource discovery requirements and it proposes flooding algorithm based on the P2P networks using simulation. To evaluate and prove DDBF algorithm we, first, evaluated it on four fixed network topologies along with two different query flooder distributions, then we evaluated it with one dynamic network topology. The performance of the proposed DDBF algorithm was assessed with five different metrics. The result showed a dramatic decrease in the number of engaged flooder nodes, the number of duplicated queries and consequently, network delay compared with the state-of-the-art algorithms.

Flexible and Scalable Optical Interconnects for Data Centers: Trends and Challenges

Emerging applications such as machine learning place higher demands on the flexibility and scalability of data center networks. On one hand, it is difficult for the fixed network topology to match the traffic patterns of different neural networks. On the other hand, the rapid growth of training data, model size, and computing requirements has led to an ever-increasing scale of data center networks. Traditional electrical interconnects lack the ability to adjust the network topology and communication bandwidth, resulting in significant resource waste and high expansion costs. Optical interconnects are introduced to solve the above problems because of its high bandwidth, low power consumption, and reconfigurability. In this article, the optical interconnects in recent years are analyzed in detail, and the flexibility and scalability of different networks are compared qualitatively. The future trends and challenges of optical interconnects in data centers are discussed from the aspects of optical transmission technology, optical switching technology, and optical network control and management.