Content caching at the edge nodes is a promising technique to reduce data traffic in next-generation wireless networks. Inspired by success of Deep Reinforcement Learning (DRL) solving complicated control problems, this work presents DRL-based framework with Wolpertinger architecture for content base station. The proposed aimed maximizing long-term cache hit rate, and it requires no knowledge popularity distribution. To evaluate framework, we compare performance other algorithms, including...

With the purpose to offload data traffic in wireless networks, content caching techniques have recently been studied intensively. Using these and a portion of popular files at local servers, users can be served with less delay. Most replacement policies are based on popularity, that depends users' preferences. In practice, such information varies over time. Therefore, an approach determine file popularity patterns must incorporated into policies. this context, we study network edge using...

To make efficient use of limited spectral resources, we in this work propose a deep actor-critic reinforcement learning based framework for dynamic multichannel access. We consider both single-user case and scenario which multiple users attempt to access channels simultaneously. employ the proposed as single agent case, extend it decentralized multi-agent multi-user scenario. In cases, develop algorithms evaluate policies via experiments numerical results. model, order performance channel...

The growing demand on high-quality and low-latency multimedia services has led to much interest in edge caching techniques. Motivated by this, we this paper consider at the base stations with unknown content popularity distributions. To solve dynamic control problem of making decisions, propose a deep actor-critic reinforcement learning based multi-agent framework aim minimize overall average transmission delay. evaluate proposed framework, compare learning-based performance three other...

Anomaly detection is widely applied in a variety of domains, involving for instance, smart home systems, network traffic monitoring, IoT applications and sensor networks. In this paper, we study deep reinforcement learning based active sequential testing anomaly detection. We assume that there an unknown number abnormal processes at time the agent can only check with one each sampling step. To maximize confidence level decision minimize stopping concurrently, propose actor-critic framework...

As the applications of deep reinforcement learning (DRL) in wireless communication grow, sensitivity DRL based strategies against adversarial attacks has started to draw more attention. In order address such and alleviate resulting security concerns, we this paper study defense DRL-based jamming attacker on a dynamic multichannel access agent. To defend attacks, propose three diversified strategies: proportional-integral-derivative (PID) control, use an imitation development orthogonal...

Due to the scarcity in wireless spectrum and limited energy resources especially mobile applications, efficient resource allocation strategies are critical networks. Motivated by recent advances deep reinforcement learning (DRL), we address multi-agent DRL-based joint dynamic channel access power control a interference network. We first propose DRL algorithm with centralized training (DRL-CT) tackle problem. In this case, is performed at central unit (CU) after training, users make...

We consider the dynamic multichannel access problem, which can be formulated as a partially observable Markov decision process (POMDP). first propose model-free actor-critic deep reinforcement learning based framework to explore sensing policy. To evaluate performance of proposed policy and framework's tolerance against uncertainty, we test in scenarios with different channel switching patterns probabilities. Then, time-varying environment identify adaptive ability framework. Additionally,...

This study introduces a novel approach, Local Spatial Projection Convolution (LSPConv), for point cloud classification and semantic segmentation. Unlike conventional methods utilizing relative coordinates local geometric information, our motivation stems from the inadequacy of existing techniques representing intricate spatial organization unconsolidated irregular 3D clouds. To address this limitation, we propose Module vector projection strategy, designed to capture comprehensive...

Due to the fact that WorldView-2 (WV2) has a small time lag while acquiring images from panchromatic (PAN) and two multispectral (MS1 MS2) sensors, moving vehicle is located at different positions in three image bands. Consequently, such displacement can be utilized identify vehicles, information, as speed direction estimated. In this paper, we focus on detection according information present novel processing chain. The locations are extracted by an improved morphological detector based...

Recently, wireless caching techniques have been studied to satisfy lower delay requirements and offload traffic from peak periods. By storing parts of the popular files at mobile users, users can locate some their requested in own caches or neighbors. In latter case, when a user receives its neighbors, device-to-device (D2D) communication is performed. The D2D underlaid with cellular networks also new paradigm for upcoming systems. allowing pair adjacent communicate directly, achieve higher...

The existing relative radiometric normalization methods are insufficient to define the invariant pixels automatically, and conventional do not perform well when multitemporal images contain a lot of changes. Two types changes should be particularly considered: one is caused by significant spectral differences due change ground objects, other in regions misalignment displacement acquisition view angles geometrical distortions. To automatically extract reduce influence changes, hierarchical...

The high spatial resolution remote sensing image provide more details such as color, size, shape, context and texture. traditional pixel-based classifier cannot satisfying results may reduce the classification accuracy. So object-oriented processing for extraction of information from data become interest. first step is segmentation. object derived by means multi-scale segmentation in this paper. hierarchical region-merging are implemented. procedure does not stop until average size all...

As a successful metaheuristic, Ant Colony Optimization (ACO) performs excellently in solving most combinatorial optimization problems. However, the ACO algorithm needs considerable computational time and resources when complexity of problem increases. Parallel implementing is good ideal to speedup it. A new parallel ant colony (PACO) presented, which has characteristics coarse-granularity interacting multi colonies, partially asynchronous implementation information exchange strategy. The...

To address the anomaly detection problem in presence of noisy observations and to tackle tuning efficient exploration challenges that arise deep reinforcement learning algorithms, we this paper propose a soft actor-critic framework. evaluate proposed framework, measure its performance terms accuracy, stopping time, total number samples needed for detection. Via simulation results, demonstrate when algorithms are employed, identify impact key parameters, such as sensing cost, on performance....

Extracting buildings from remote sensing images is a significant task with many applications such as map drawing, city planning, population estimation, etc. However, traditional methods that rely on artificially designed features struggle to perform well due the diverse appearance and complicated background. In this paper, we design an end-to-end convolutional neural network combines semantic segmentation edge detection for building extraction. addition, propose residual unit combined...

We address the problem of sequentially selecting and observing processes from a given set to find anomalies among them. The decision-maker observes subset at any time instant obtains noisy binary indicator whether or not corresponding process is anomalous. develop an anomaly detection algorithm that chooses be observed instant, decides when stop taking observations, declares decision on anomalous processes. objective identify with accuracy exceeding desired value while minimizing delay in...

As the applications of deep reinforcement learning (DRL) in wireless communications grow, sensitivity DRL based communication strategies against adversarial attacks has started to draw increasing attention. In order address such and alleviate resulting security concerns, we this paper consider a victim user that performs DRL-based dynamic channel access, an attacker executes DRLbased jamming disrupt victim. Hence, both are agents can interact with each other, retrain their models, adapt...

Time domain gating in vector network analyzers (VNAs) is used widely EMC and other RF measurement applications. Conceptually, it straightforward - the analyzer first converts frequency data to time via inverse Fourier transform, a gate applied include or exclude certain periods. Because provided as built-in function commercial analyzers, user typically does not concern himself with details of mathematical process. However, actual implementations are different VNAs, especially designs band...

The precise capture and identification of movement features are important for numerous scientific endeavors. In this work, we present a novel multimodal sensor, called the resistance/capacitance dual-mode (RCDM) which effectively differentiates between compression stretchable strains during tennis motion; meanwhile, it can also accurately identify various joint movements. proposed wearable device seamless design, comprising two separate components: resistive part capacitive part. components...

Unmanned Aircraft Systems (UAS) operations are changing the way aviation and commerce conducted today. Until recently, for civil commercial operations, nearly all UAS within visual line of sight (VLOS). However, this severely limits economic benefits that can be realized by use these unmanned, someday, autonomous systems.Beyond (BVLOS) require much more capabilities operator to rely on general public condone comfortable with. BVLOS ground platform technologies with varying states maturity....