Unmanned aerial vehicles (UAVs) are expected to be an integral part of wireless networks, and determining collision-free trajectory in multi-UAV non-cooperative scenarios while collecting data from distributed Internet of Things (IoT) nodes is a challenging task. In this paper, we consider a path planning optimization problem to maximize the collected data from multiple IoT nodes under realistic constraints. The considered multi-UAV non-cooperative scenarios involve random number of other UAVs in addition to the typical UAV, and UAVs do not communicate or share information among each other. We translate the problem into a Markov decision process (MDP) with parameterized states, permissible actions, and detailed reward functions. Dueling double deep Q-network (D3QN) is proposed to learn the decision making policy for the typical UAV, without any prior knowledge of the environment (e.g., channel propagation model and locations of the obstacles) and other UAVs (e.g., their missions, movements, and policies). The proposed algorithm can adapt to various missions in various scenarios, e.g., different numbers and positions of IoT nodes, different amount of data to be collected, and different numbers and positions of other UAVs. Numerical results demonstrate that real-time navigation can be efficiently performed with high success rate, high data collection rate, and low collision rate.

Unmanned aerial vehicles (UAVs) are expected to be an integral part of wireless networks, and determining collision-free trajectories for multiple UAVs while satisfying requirements of connectivity with ground base stations (GBSs) is a challenging task. In this paper, we consider non-cooperative multi-UAV scenarios, in which multiple UAVs need to fly from initial locations to destinations, while satisfying collision avoidance, wireless connectivity, and kinematic constraints. We aim to find trajectories for the UAVs with the goal to minimize their mission completion time. We first formulate the multi-UAV trajectory optimization problem as a sequential decision making problem. We, then, propose a decentralized deep reinforcement learning approach to solve the problem. More specifically, a value network is developed to obtain values given the agent’s joint state (including the agent’s information, the nearby agents’ observable information, and the locations of the nearby GBSs). A signal-to-interference-plus-noise ratio (SINR)-prediction neural network is also designed, using accumulated SINR measurements obtained when interacting with the cellular network, to map the GBSs’ locations into the SINR levels in order to predict the UAV’s SINR. Numerical results show that with the value network and SINR-prediction network, real-time navigation for multi-UAVs can be efficiently performed in various environments with high success rate.

A K-tier heterogeneous downlink millimeter wave (mmWave) cellular network with user-centric small cell deployments is studied in this paper. In particular, we consider a heterogeneous network model with user equipments (UEs) being distributed according to a Poisson cluster process (PCP). Distinguishing features of mmWave communications including directional beamforming and a sophisticated path loss model incorporating both line-of-sight (LOS) and non-line-of-sight (NLOS) transmissions, are taken into account. Initially, we determine general expressions for the association probabilities of different tier BSs. Using tools from stochastic geometry, we then characterize the Laplace transform of the interference and derive a general expression for the signal-to-interference-plus-noise ratio (SINR) coverage probability. While these expressions are applicable to any PCP, we later specialize the results to two popular PCPs, namely (i) Thomas cluster process, where the UEs are clustered around the base stations (BSs) according to a Gaussian distribution, and (ii) Mat\'ern cluster process, where the UEs are scattered according to a uniform distribution. Subsequently, upper and lower bounds for the coverage probability are provided. Special cases are also addressed, providing the insight that when the cluster size grows without bound, our PCP-based model specializes to a Poisson point process (PPP)-based model. Area spectral efficiency is investigated as well. Moreover, we have discussed extensions to cases in which more practical antenna gain patterns are taken into account and also the shadowing is considered. Via numerical and simulation results, we explore the effects of the beamwidth of the main lobe, the main lobe directivity gain, the biasing factor and the transmit power of the small-cell BSs to get insight on the performance in practical scenarios. Performance in dense networks is also analyzed numerically.

In this paper, we investigate jamming-resilient UAV path planning strategies for data collection in Internet of Things (IoT) networks, in which the typical UAV can learn the optimal trajectory to elude such jamming attacks. Specifically, the typical UAV is required to collect data from multiple distributed IoT nodes under collision avoidance, mission completion deadline, and kinematic constraints in the presence of jamming attacks. We first design a fixed ground jammer with continuous jamming attack and periodical jamming attack strategies to jam the link between the typical UAV and IoT nodes. Defensive strategies involving a reinforcement learning (RL) based virtual jammer and the adoption of higher SINR thresholds are proposed to counteract against such attacks. Secondly, we design an intelligent UAV jammer, which utilizes the RL algorithm to choose actions based on its observation. Then, an intelligent UAV anti-jamming strategy is constructed to deal with such attacks, and the optimal trajectory of the typical UAV is obtained via dueling double deep Q-network (D3QN). Simulation results show that both non-intelligent and intelligent jamming attacks have significant influence on the UAV’s performance, and the proposed defense strategies can recover the performance close to that in no-jammer scenarios.

In the detection of surface floating garbage, the existence of complex backgrounds and the small target sizes make the surface floating garbage easy to mis-detect. Existing approaches cannot yet provide a solution to the aforementioned problems and they are typically limited to addressing specific issues. This paper proposes a PC-Net algorithm for floating garbage detection. First, a pyramid anchor generation approach is proposed, which makes the anchor to be generated centrally near the target and reduces the interference of background information in the anchor generation. Then, in the RoI Pooling feature map import stage, the classification map is used as the feature map. This approach generates feature maps with a higher resolution and more distinct features, thereby enhancing the feature information of small targets and enhancing the classification accuracy. Experimental results on floating garbage dataset indicate that the average detection accuracy of the proposed approach is 86.4%. Compared with existing detection approaches, such as Faster R-CNN, YOLOv3, YOLOX, and Dynamic R-CNN, the average accuracy of detection is increased by 4.1%, 3.6%, and 2.8%, respectively.