Greedy Strategy Research Articles

Multi-agent reinforcement learning enabled link scheduling for next generation Internet of Things

In next generation Internet of Things (NG-IoT) networks, numerous pieces of information are aggregated from the user devices and sensor nodes to the local computing units for further computing to support high-level applications. Those multitudinous transmission demands have raised new challenges for current link scheduling protocols. The centralized link scheduling protocols are inappropriate in some large-scale NG-IoT scenarios. The previously distributed link scheduling uses the randomized transmission scheme to avoid interference, making it hard to utilize the bandwidth resources fully. The multi-agent machine learning (MAML) technique is a potential approach to finding the most optimal link scheduling strategy. At the same time, the over-large state space will take a long time for them to approach the optimal solution, which reduces the practicality of the MAML. To fully utilize the bandwidth resource and improve the efficiency of link scheduling, this paper studies a multi-agent reinforcement learning enabled link scheduling problem. Different from the conventional MAML techniques that randomly select a state to do their exploration, in our multi-agent reinforcement learning algorithm, a good state is firstly obtained within polynomial time steps by executing a distributed and randomized sub-algorithm. We say a state is good if it is not far from the optimal state. Then, our multi-agent reinforcement learning scheme starts from the good state and does its exploration with an ɛ greedy scheme, which significantly reduces the time steps to get close to the optimal link scheduling strategy. Extensive simulations are conducted to investigate the performance of our work.

Cost- and Delay-Efficient Backhaul Selection for Time-Sensitive Maritime Communications

Extending the existing near-shore terrestrial infrastructure with non-terrestrial network capabilities helps maritime operators alleviate the high costs of communication and meet the requirements imposed by time-sensitive applications. Recognizing that the deployment of terrestrial and non-terrestrial networks necessitates selecting from the available wireless backhaul solutions, which have dissimilar data transmission costs and communication link qualities, it is essential to propose an appropriate backhaul selection policy. Specifically, in this letter, we coin a backhaul selection policy that manages the inherent trade-off between data transmission expenses and timely throughput guarantees for maritime communications. We formulate the backhaul selection problem as a Markov decision process and show that the proposed solution is not only more cost-efficient, but also satisfies the timely throughput requirements in contrast to currently used greedy strategies.

Circulant Shift-Based Beamforming for Secure Communication With Low-Resolution Phased Arrays

Millimeter wave (mmWave) technology can achieve high-speed communication due to the large available spectrum. Furthermore, the use of directional beams in mmWave system provides a natural defense against physical layer security attacks. In practice, however, the beams are imperfect due to mmWave hardware limitations such as the low-resolution of the phase shifters. These imperfections in the beam pattern introduce an energy leakage that can be exploited by an eavesdropper. To defend against such eavesdropping attacks, we propose a directional modulation-based defense technique where the transmitter applies random circulant shifts of a beamformer. We show that the use of random circulant shifts together with appropriate phase adjustment induces artificial phase noise (APN) in the directions different from that of the target receiver. Our method corrupts the phase at the eavesdropper without affecting the communication link of the target receiver. We also experimentally verify the APN induced due to circulant shifts, using channel measurements from a 2-bit mmWave phased array testbed. Using simulations, we study the performance of the proposed defense technique against a greedy eavesdropping strategy in a vehicle-to-infrastructure scenario. The proposed technique achieves better defense than the antenna subset modulation, without compromising on the communication link with the target receiver.

A Bayesian Approach to Network Monitoring for Progressive Failure Localization

Boolean Network Tomography (BNT) aims at identifying failures of internal network components by means of end-to-end monitoring paths. However, when the number of failures is not known a priori, failure identification may require a huge number of monitoring paths. We address this problem by designing a Bayesian approach that progressively selects the next path to probe on the basis of its expected information utility, conditioned on prior observations. As the complexity of the computation of posterior probabilities of node failures is exponential in the number of failed paths, we propose a polynomial-time greedy strategy which approximates these values. To consider aging of information in dynamic failure scenarios where node states can change during a monitoring period, we propose a monitoring technique based on a sliding observation window of adaptive length. By means of numerical experiments conducted on real network topologies we demonstrate the practical applicability of our approach, and the superiority of our algorithms with respect to state of the art solutions based on classic BNT as well as sequential group testing.

ISOCD: Overlapping community detection algorithm based on local influence and similarity

Many nodes of a complex network in the real world belong to multiple communities. Research on the detection of overlapping communities with common nodes can help us understand the topology of network and extract information hidden in the network. Considering the problems of many existing overlapping community detection methods, such as high time complexity or poor division results, we propose an overlapping community detection method based on local influence and similarity (ISOCD). The ISOCD method selects seed nodes based on local influence, follows a greedy strategy, takes the fitness function of similarity as the optimization goal, and finally optimizes and merges the detected communities. In experiments, the method we propose has good community division effects both on the real-world and synthetic networks, and has low time complexity.

A Power Allocation Scheme for MIMO-NOMA and D2D Vehicular Edge Computing Based on Decentralized DRL

In vehicular edge computing (VEC), some tasks can be processed either locally or on the mobile edge computing (MEC) server at a base station (BS) or a nearby vehicle. In fact, tasks are offloaded or not, based on the status of vehicle-to-infrastructure (V2I) and vehicle-to-vehicle (V2V) communication. In this paper, device-to-device (D2D)-based V2V communication and multiple-input multiple-output and nonorthogonal multiple access (MIMO-NOMA)-based V2I communication are considered. In actual communication scenarios, the channel conditions for MIMO-NOMA-based V2I communication are uncertain, and the task arrival is random, leading to a highly complex environment for VEC systems. To solve this problem, we propose a power allocation scheme based on decentralized deep reinforcement learning (DRL). Since the action space is continuous, we employ the deep deterministic policy gradient (DDPG) algorithm to obtain the optimal policy. Extensive experiments demonstrate that our proposed approach with DRL and DDPG outperforms existing greedy strategies in terms of power consumption and reward.

On the Vulnerability of Graph Learning-based Collaborative Filtering

Graph learning-based collaborative filtering (GLCF), which is built upon the message-passing mechanism of graph neural networks (GNNs), has received great recent attention and exhibited superior performance in recommender systems. However, although GNNs can be easily compromised by adversarial attacks as shown by the prior work, little attention has been paid to the vulnerability of GLCF. Questions like can GLCF models be just as easily fooled as GNNs remain largely unexplored. In this article, we propose to study the vulnerability of GLCF. Specifically, we first propose an adversarial attack against CLCF. Considering the unique challenges of attacking GLCF, we propose to adopt the greedy strategy in searching for the local optimal perturbations and design a reasonable attacking utility function to handle the non-differentiable ranking-oriented metrics. Next, we propose a defense to robustify GCLF. The defense is based on the observation that attacks usually introduce suspicious interactions into the graph to manipulate the message-passing process. We then propose to measure the suspicious score of each interaction and further reduce the message weight of suspicious interactions. We also give a theoretical guarantee of its robustness. Experimental results on three benchmark datasets show the effectiveness of both our attack and defense.

Strengthening evolution-based differential evolution with prediction strategy for multimodal optimization and its application in multi-robot task allocation

Many real-world problems can be considered multimodal optimization problems (MMOPs), which require locating as many global optima as possible and refining the accuracy of the found optima as high as possible. However, there are some issues with existing algorithms for solving MMOPs. For instance, most of the existing methods adopt the greedy selection strategy to select offspring, which may lead some individuals to fall into local optima and the repetitive evaluations for these local optima will exhaust many fitness evaluations (FEs). Moreover, many MMOPs tend to be expensive to evaluate, and the rational allocation of evaluation resources to better deal with MMOPs is a critical challenge within a limited number of FEs. How to allocate FEs reasonably in a whole evolution and how to avoid individuals becoming trapped in local optima are two key problems in solving MMOPs. Therefore, this paper proposes a strengthening evolution-based differential evolution with prediction strategy (SEDE-PS) for solving MMOPs and verifies its performance in a multirobot task allocation (MRTA) problem, which has the following three contributions. First, a neighbour-based evolution prediction (NEP) strategy is proposed to predict the position of individuals in the next generation by using the historical information of individuals as much as possible. Second, a prediction-based mutation (PM) strategy is introduced to accelerate convergence by combining it with the NEP strategy. Third, a strengthening evolution (SE) strategy is proposed to select inferior individuals to evolve them unconditionally several times and make them approach global optima or jump out of local optima. We compare the SEDE-PS with state-of-the-art multimodal optimization algorithms on the widely used CEC’2013 benchmark. The experimental results show that SEDE-PS performs better than, or is competitive with these compared algorithms. Moreover, SEDE-PS is applied to a real-world MRTA problem to further verify the effectiveness of SEDE-PS.

On fast greedy block Kaczmarz methods for solving large consistent linear systems

A fast greedy block Kaczmarz method combined with general greedy strategy and average technique are proposed for solving large consistent linear systems. Theoretical analysis of the convergence of the proposed method is given in detail. Numerical experiments show that the proposed methods are efficient and faster than the existing methods.

Point–Line-Aware Heterogeneous Graph Attention Network for Visual SLAM System

Simultaneous localization and mapping (SLAM), as an important research topic in robotics, is useful but challenging to estimate robot pose and reconstruct a 3-D map of the surrounding environment. Despite recent success of several deep neural networks for visual SLAM, those methods cannot achieve robust results in complex industrial scenarios for constructing accurate and real-time maps due to the weak texture and complex geometric structure. This paper presents a novel and efficient visual SLAM system based on point–line-aware heterogeneous graph attention network, which combines points and line segments to solve the problem of the insufficient number of reliable features in traditional approaches. Firstly, a simultaneous feature extraction network is constructed based on the geometric relationships between points and points and points and lines. To further improve the efficiency and accuracy of the geometric association features of key regions, we design the point–line-aware attention module to guide the network to pay attention to the trivial features of both points and lines in images. Moreover, the network model is optimized by a transfer-aware knowledge distillation strategy to further improve the system’s real-time performance. Secondly, to improve the accuracy of the point–line matching, we design a point–line heterogeneous graph attention network, which combines an edge aggregation graph attention module and a cross-heterogeneous graph iteration module to conduct learning on the intragraph and intergraph. Finally, the point–line matching process is transformed into an optimal transport problem, and a near-iterative method based on a greedy strategy is presented to solve the optimization problem. The experiments on the KITTI dataset and a self-made dataset demonstrate the better effectiveness, accuracy, and adaptability of our method than those of the state of the art in visual SLAM.

Jerk-Optimal Piecewise Planning of Tool Orientation for 5-Axis Ball-End Machining With Linearized Kinematic Constraints

Abstract When planning tool orientations for 5-axis machining from the perspective of kinematics, existing works often tend to optimize the angular velocity or acceleration of the rotary axes rather than the angular jerk due to its calculation sensitivity to discretization variations, though the jerk is fundamentally tied to the tracking errors and residual vibrations of the actuators. In this paper, a study is reported on how to optimize the angular jerk of the rotary axes while comprehensively considering the kinematic constraints, thus achieving a jerk-optimal tool orientation along the tool path, with driving capacity of the rotary axes respected. In this method, the displacements of the rotary axes are continuously represented by two quintic B-spline curves, and then the angular velocity, acceleration, and jerk of the rotary axes, which are the derivatives of the displacements, can be succinctly represented as a B-spline curve. Taking advantage of the convex hull property of B-spline curve, the linear analytical representations of the kinematic constraints of the rotary axes can be successfully derived in form of control coefficient combinations. To prevent the machining interference at the same time, a greedy strategy that incorporates a process of alternately smoothing tool orientation and checking machining interference is employed. Then, the smooth displacement splines of the rotary axes can be obtained by solving a constructed quadratic programming (QP) model that minimizes the angular jerk along the tool path, while satisfying kinematic constraints and without machining interference. Moreover, to generate efficiently tool orientations for long tool paths, a piecewise planning strategy that optimizes the tool orientation from coarse to fine is developed. Finally, the conducted experiments validate the proposed method.

Auction-Based Dependent Task Offloading for IoT Users in Edge Clouds

The rapid proliferation of latency-sensitive Internet of Things (IoT) applications boosts the frequency of offloading compute-intensive tasks from IoT users to mobile edge computing (MEC) due to the limitation resources of IoT devices. It is inevitably for IoT users to compete for the computing resources of the MEC, especially when the computation tasks are dependent and have hard deadline constraints. However, most existing dependent task offloading schemes may not well consider the resource competition issues among IoT users, and possibly lead to limited system performance in multiuser scenario. To address this issue, we intend to design an auction-based dependent task-offloading mechanism to improve the efficiency of task offloading for multiple IoT users. First, we formulate the dependent task offloading as a valuation maximization problem in the trade of computing resources satisfying users’ latency requirements, which has been proved to be NP-hard. Then, by jointly considering the task graph structure and the current status of the MEC, we propose a truthful auction mechanism, named greedy winner selection strategy, in which a heuristic dependent task assignment for winners is designed to improve the efficiency of the task offloading. By conducting extensive simulations, we validate that the performance of the proposed dependent task offloading strategy is superior to existing competition algorithms, in terms of total valuations, average makespans, and success rates.

Time-Aware Data Partition Optimization and Heterogeneous Task Scheduling Strategies in Spark Clusters

Abstract The Spark computing framework provides an efficient solution to address the major requirements of big data processing, but data partitioning and job scheduling in the Spark framework are the two major bottlenecks that limit Spark’s performance. In the Spark Shuffle phase, the data skewing problem caused by unbalanced data partitioning leads to the problem of increased job completion time. In response to the above problems, a balanced partitioning strategy for intermediate data is proposed in this article, which considers the characteristics of intermediate data, establishes a data skewing model and proposes a dynamic partitioning algorithm. In Spark heterogeneous clusters, because of the differences in node performance and task requirements, the default task scheduling algorithm cannot complete scheduling efficiently, which leads to low system task processing efficiency. In order to deal with the above problems, an efficient job scheduling strategy is proposed in this article, which integrates node performance and task requirements, and proposes a task scheduling algorithm using greedy strategy. The experimental results prove that the dynamic partitioning algorithm for intermediate data proposed in this article effectively alleviates the problem that data skew leads to the decrease of system task processing efficiency and shortens the overall task completion time. The efficient job scheduling strategy proposed in this article can efficiently complete the job scheduling tasks under heterogeneous clusters, allocate jobs to nodes in a balanced manner, decrease the overall job completion time and increase the system resource utilization.

MPPT Control Research of Improved Gray Wolf Algorithm According to Levy Flight and Greedy Strategy

MPPT control study of improved Gray Wolf algorithm (GWO) according to Levy flight and greedy strategy: Solving the problems of maximum power point tracking (MPPT) with traditional Gray Wolf algorithm (GWO) under local shadows of photovoltaic arrays and sudden environmental changes, which is easily trapped in local optimum and has a slow convergence rate, and poor solution accuracy, a improved gray wolf algorithm (LGWO) according to Levy flight and greedy strategy is presented. It is used for the first time for maximum power point tracking under partial shadow and dynamic shadow changes of photovoltaic array. It is based on the traditional Grey Wolf algorithm (GWO), it introduces Levy’s flight search strategy, improve the algorithm of global search ability, expands search range, and filters optimal range through greedy strategy to further accelerate the convergence speed. MPPT simulation model according to Boost circuit is built using MATLAB/Simulink to verify. Experiments in the presence of partial shadows and shadow mutations. The results of simulation experiments show that the ameliorated Gray Wolf algorithm (LGWO) improves the tracking accuracy of MPPT by 0.03%, improves the convergence speed by 1.1 times and is more stable after reaching the maximum number of iterations. This verifies the feasibility and superiority of the ameliorated Grey Wolf algorithm in maximum power point tracking control.

Sensor Deployment Strategies for Target Coverage Problems in Underwater Acoustic Sensor Networks

Owing to the dynamic characteristics of the underwater environment, the set of underwater targets needing to be covered and monitored by sensor nodes often exhibits a variety of irregular distributions, bringing a certain burden to node deployment. In this letter, an improved greedy strategy with a depth-first search mechanism was proposed to address the coverage problem of underwater targets. With the optimal selection criteria of deployment location at each step and a flexible search process, the global solution problem was converted into multiple local optimal solution problems to improve solution efficiency. Thus the performance of coverage and connectivity can be effectively guaranteed in underwater scenarios with irregular target distributions. The simulation results show that the proposed algorithm has superiority in solving the target coverage problem in underwater acoustic sensor networks (UASNs), which is manifested in node deployment cost and coverage efficiency.

A Toolpath Optimization Algorithm for Layered 3D Printings based on Solving the TSP

In order to optimize the tool path of 3D printing such that the efficiency is improved, by summerizing the pros and cons of existing methods, we proposed a noval tool path optimization algorithm for layered 3D printings based on solving the Traveling Salesman Problem. Our algorithm first adjusts the major printing direction using Principal Components Analysis, and then applies the greedy strategy and generates multiple printing paths by interleavingly appending the filling segments and contour segments along the major printing direction. Thereafter, by considering the multiple printing paths as “cities”, and elaborately defining the distances between them, we successfully model the problem of minimizing the flying distances as a kind of Traveling Salesman Problem. Af-ter converting the accurate solution of TSP to our problem, we are able to get the global optimized tool path with the given multiple printing paths. Experiments demonstrate that the proposed algorithm not only miminizes the flying distances, but also reduces the switches between printing status and flying status, and boosts the percentage of long segments in the whole tool path as well, thus the printing efficiency is significently imporved without sac-rifying the printing quality.

A note on the network coloring game: A randomized distributed (Δ + 1)-coloring algorithm

The network coloring game has been proposed in the literature of social sciences as a model for conflict-resolution circumstances. The players of the game are the vertices of a graph with n vertices and maximum degree Δ. The game is played over rounds, and in each round all players simultaneously choose a color from a set of available colors. Players have local information of the graph: they only observe the colors chosen by their neighbors and do not communicate or cooperate with one another. A player is happy when she has chosen a color that is different from the colors chosen by her neighbors, otherwise she is unhappy, and a configuration of colors for which all players are happy is a proper coloring of the graph. It has been shown in the literature that, when the players adopt a particular greedy randomized strategy, the game reaches a proper coloring of the graph within O(log⁡(n)) rounds, with high probability, provided the number of colors available to each player is at least Δ+2. In this note we show that a modification of the aforementioned greedy strategy yields likewise a proper coloring of the graph, provided the number of colors available to each player is at least Δ+1.

DDRCN: Deep Deterministic Policy Gradient Recommendation Framework Fused with Deep Cross Networks

As an essential branch of artificial intelligence, recommendation systems have gradually penetrated people’s daily lives. It is the active recommendation of goods or services of potential interest to users based on their preferences. Many recommendation methods have been proposed in both industry and academia. However, there are some limitations of previous recommendation methods: (1) Most of them do not consider the cross-correlation between data. (2) Many treat the recommendation process as a one-time act and do not consider the continuity of the recommendation system. To overcome these limitations, we propose a recommendation framework based on deep reinforcement learning techniques, known as DDRCN: a deep deterministic policy gradient recommendation framework incorporating deep cross networks. We use a Deep network and a Cross network to fit the cross relationships between the data, to obtain a representation of the user interaction data. The Actor-Critic network is designed to simulate the continuous interaction behavior of users through a greedy strategy. A deep deterministic policy gradient network is also used to train the recommendation model. Finally, we conduct experiments with two publicly available datasets and find that our proposed recommendation framework outperforms the baseline approach in the recall and ranking phases of recommendations.

Electric bus scheduling with the self-adaptive charging strategy based on a course selection idea

This paper investigates the electric bus scheduling problem, aiming to design collaborative service and charging schedules. A self-adaptive charging strategy is introduced, allowing electric buses to be recharged partially and flexibly. Then, a mixed-integer programme model is built for the problem, based on a course selection idea that considers continuous time. Two models for scenarios considering prior greedy and full charging strategies are also established for comparison. Next, a greed-based branch-and-price algorithm is developed for model solving, which incorporates a greedy attachment heuristic for the initialisation and a customised branching scheme based on the course flow. The results of a case study in Beijing show the efficiency of the proposed algorithm and the superiority of the self-adaptive charging strategy. Compared with the two prior charging strategies, the average energy replenishment of electric buses is reduced by 22.47% and 35.99%, and the system cost is reduced by 5.47% and 5.60%.

Cooperative driver module identification based on single cell data

Identifying cooperative driver modules from large-scale cancer omics data is one of the key topics in bioinformatics. Current methods usually only focus on discovering driver modules from batch omics data and ignore the cancer heterogeneity on the cell level, which are venerable to batch-level noises. To overcome these limitations, we propose a cooperative driver module identification method (CDMFinder) based on single-cell data and prior knowledge. CDMFinder first utilizes the gene co-expression specificity of different cancer subtypes and normal cell expression data to construct expression association networks and fuses these networks with a gene interaction network to obtain a gene functional association network. Thus, it effectively reduces network complexity while capturing in-depth functional associations between genes. It then adopts an overlapping Markov clustering on this functional network to mine functional clusters, along with a greedy strategy with a hybrid weight function to identify driver modules from the clusters. Finally, it introduces an interaction and mutation co-occurrence-based distance function on driver module sets to identify cooperative driver modules. CDMFinder fully integrates a variety of genetic factors (i.e., expression, mutation, and subtype specificity) and manifests a prominent performance. Experimental results on the multi-omics data of breast cancer and glioblastoma show that the number of driver genes identified by CDMFinder is 1.35 times larger than that of competitive methods. The identified cooperative driver modules are enriched at the pathway/functional level and are over 1.5 times more than compared methods.