In this paper, we investigate a resource allocation problem for a multi-unmanned aerial vehicle (UAV) assisted integrated sensing and communication system (ISAC), where a group of dual-functional UAVs perform simultaneous radar sensing of a target and data communication with multiple ground users. In particular, the trajectory of UAVs, user association, and beamforming design are jointly considered to maximize the sum weighted bit rate of all ground users while ensuring the sensing beampattern gain of the target. To cope with the above mixed-integer non-convex optimization problem, we propose an efficient strategy by decomposing the original problem into two subproblems under the alternating optimization framework. For the user association and beamforming design, we propose a novel algorithm to circumvent the coupling relationship among ground users and UAVs by leveraging matching theory and fractional programming theory. For the non-convex UAV trajectory subproblem, we apply the sequential quadratic programming to obtain a sub-optimal solution by solving a sequence of quadratic programming problems. The above two subproblems are iteratively solved and a stable solution is obtained upon convergence. Simulation results show that the proposed strategy outperforms various benchmark schemes that are based on the deferred acceptance algorithm, K-means algorithm, and a heuristic algorithm. It is demonstrated that the proposed strategy efficiently improve the sensing beampattern gain and communication rate.

In this paper, we investigate a resource allocation problem for a multi-unmanned aerial vehicle (UAV)-assisted full-duplex wireless-powered Internet-of-things (IoT) network, where the slot partition, power allocation, user association, and three dimensional (3D) UAV placement are jointly considered to maximize the sum bit rate of all IoT devices under the imperfect self-interference cancellation and generalized probabilistic air-ground channel model. To deal with the formulated mixed-integer non-convex problem, we propose a novel resource allocation strategy with three nested parts by integrating the model-based optimization theory with the data-based learning theory. Particularly, the data-based deep deterministic policy gradient algorithm is only explicitly used to train the 3D UAV placement policy, while the model-based Lagrange dual theory and matching theory are implicitly used to explore the hidden tractability of the rest two parts and design efficient algorithms, where the optimization results are passed onto the data-based part through reward values. Simulation results show that the proposed strategy greatly cuts down the execution time of the exhausting search-based genetic algorithm by 4 orders of magnitude at the cost of less than 5.1 percent performance loss.

In the emerging non-orthogonal multiple access-based unmanned aerial vehicle systems, it is of practical importance to maximize the number of users with satisfied quality-of-service experience by incorporating placement design, admission control, and power allocation. To overcome the combinatorial nature and non-convexity of the above problem, a low-complexity mechanism is proposed based on the penalty function method and the successive convex approximation, and it converges to a Karush-Kuhn-Tucker point of the original problem. Simulations are provided to show the necessity of conducting the above optimization and the efficiency of the proposed mechanism.

The detection of fine-grained objects in remote sensing images has been recognized as a challenging issue, which cannot be well addressed by the existing deep learning-based methods due to their inadaptability to multi-scale objects, slow convergence speed, and limitations to scarce datasets. To cope with the above issue, we propose a deep convolutional neural network (CNN)-based detection method for multi-scale fine-grained objects in complex remote sensing scenarios. Specifically, we adopt a deep CNN with a residual structure as the backbone network, to extract deep-level details from the image. Besides, we introduce a multi-scale region generation network to overcome the limitations of fixed receptive field convolution kernels and enable multi-scale object detection. Lastly, we replace fully connected layers in the fully convolutional region classification network with $1\times 1$ convolutional layer to enhance detection efficiency and detection speed. To overcome the limitation of scarce datasets, we conducted experiments on the FAIR1M dataset, which is currently the largest fine-grained object detection dataset in the remote sensing field. Simulation results show that the proposed detection method achieves the highest average precision (35.86%) among all benchmarks and outperforms the classic Faster R-CNN-based method by 3.44%. Furthermore, our method demonstrates significantly improved detection speed compared to the Faster R-CNN-based methods.

In order to relieve the traffic burden of the ground base station (BS) and improve system operation efficiency, we study the energy-efficient optimization problem in a non-orthogonal multiple access (NOMA)-based unmanned aerial vehicle (UAV)-assisted data collection system, where a UAV assists a ground BS for traffic offloading. Particularly, the joint optimization problem of the placement of the UAV and transmit power of the sensor nodes is formulated to maximize the energy efficiency of all sensor nodes under the quality-of-service constraints and the probabilistic ground-to-air channel model. To balance the optimized performance and online operation time, we propose an alternating-based offline optimization algorithm for obtaining the optimal online UAV trajectory policy. Under the given UAV placement, the formulated problem is first transformed into a convex power allocation subproblem. Based on the power-allocation solutions, the deep deterministic policy gradient algorithm is then leveraged to approach the optimal UAV placement. Simulations show that the proposed algorithm can obtain an efficient performance while consuming an average online operation time of fewer than 0.2 seconds.

To compensate for the transmission efficiency in a cooperative non-orthogonal multiple access (NOMA) system, we enable in-band full-duplex (IBFD) and two-way cognitive transmission at the mobile relay, which assists the NOMA user in weak channel condition while transmitting and receiving its own data by means of spectral sharing. To strike a balance between spectral efficiency and energy conservation, we consider the power allocation problem to maximize the energy efficiency of the hybrid system while maintaining the bit rate requirements of NOMA users. To overcome the non-convexity of the original problem, we apply the successive inner approximation technique and propose an efficient iterative scheme to obtain a Karush-Kuhn-Tucker point in polynomial time. The gain of IBFD-enabled two-way cognitive transmission and the efficiency of the proposed power allocation scheme are validated via simulation results.