UAV remote sensing image registration has a wide range of applications in the fields of fine geographic information extraction, environmental change monitoring, agriculture and forestry, serving as a fundamental and critical step in supporting various applications. However, existing registration methods often extract fewer feature points, uneven distribution and poor stability when faced with significant changes in land features. Accordingly, in this paper, we propose a UAV remote sensing image registration method based on an improved SuperPoint+SuperGlue deep learning network. We add a feature point extraction branch to the shallow feature map in the SuperPoint network, and add the newly extracted feature points to the existing feature points, and then perform non-maximum suppression (NMS) for further refinement. This enhances its capability to extract image feature points, resulting in more accurate and evenly distributed positions of these points. The experiments carried on two sets of real UAV datasets indicate that our method outperforms several typical registration approaches in terms of registration accuracy and detection efficiency.

Microscopic imaging plays an important role in the biomedical field. Existing deep learning based methods rely on high-quality data. However, there is a lot of noise (such as bubbles and impurities) in the microscopic images of biological samples collected outdoors, which may lead to significant interference in the microscopic objects identification task. To solve this problem, this paper proposes a deep learning based method for microscopic object localization and classification. Firstly, the whole slide image is preprocessed to obtain the microscopic images after preliminary filtering bubbles and impurities. Then, the sensitized pollen grains are located based on the deep learning model to remove the interference of remaining impurities, and the microscopic images of sensitized pollen grains are classified. This method can effectively suppress the interference of noise in microscopic images on object classification and improve the accuracy and reliability of model. The proposed method is verified by experiments based on real data and the results show that the proposed method achieves the highest accuracy compared with other deep learning methods.

In this letter, a model of wide-area distribution multistatic$\text{T}^{M}$-$\text{R}^{N}$($M$independent transmitters and$N$independent receivers) sky-wave over-the-horizon radar (WDMS-OTHR) is investigated. A localization algorithm that utilizes the hill-climbing algorithm based on the weighted least squares (HC-WLS) method for this model is also proposed. The localization algorithm mainly includes three stages. First, the geodetic coordinate of station position is transformed into the spatial rectangular coordinate. Second, the initial position is estimated by WLS that forms a pseudolinear equation from the measurements of the time of arrival (TOA) and introduces nuisance variables as additional unknowns. Third, the spatial rectangular coordinate of the initial position is transformed into the geodetic coordinate, and the initial position is optimized by HC that performs eight neighborhood searches around the initial position and adopts multiple iteration searches until the optimal position is obtained. Numerical simulations demonstrate that the positioning accuracy of the proposed algorithm is better than WLS.

Cache randomization has recently been revived as a promising defense against conflict-based cache side-channel attacks. As two of the latest implementations, CEASER-S and ScatterCache both claim to thwart conflict-based cache side-channel attacks using randomized skewed caches. Unfortunately, our experiments show that an attacker can easily find a usable eviction set within the chosen remap period of CEASER-S and increasing the number of partitions without dynamic remapping, such as ScatterCache, cannot eliminate the threat. By quantitatively analyzing the access patterns left by various attacks in the LLC, we have newly discovered several problems with the hypotheses and implementations of randomized caches, which are also overlooked by the research on conflict-based cache side-channel attacks.However, cache randomization is not a false hope and it is an effective defense that should be widely adopted in future processors. The newly discovered problems are corresponding to flaws associated with the existing implementation of cache randomization and are fixable. Several new defense ideas are proposed in this paper. Our experiments show that all the newly discovered problems are fixed within the current performance budget. We also argue that randomized set-associative caches can be sufficiently strengthened and possess a better chance to be actually adopted in commercial processors than their skewed counterparts because they introduce less overhaul to the existing cache structure.

In this paper, we consider the simultaneous information and power transfer (SIPT) in cognitive radio networks (CRNs) where the secondary receiver (SR) can harvest energy from both the primary transmitter (PT) and the secondary transmitter (ST). The SIPT-enabled SR turns the harmful interference from the PT into energy and uses it to prolong battery life. We assume that the ST and SR adopt orthogonal frequency division multiplexing (OFDM) modulation and our goal is to find the optimal power allocation and power splitting ratio that maximize the energy efficiency of the secondary system. To settle this non-convex problem, we utilize the fractional programming to transform the objective function into the counterpart subtractive form and then take advantage of an approximation to turn the problem convex. After that, the equivalent convex problem is solved by an efficient iterative algorithm. Numerical results illustrate that the proposed method performs well and achieves a good tradeoff between energy efficiency and capacity of the secondary system, especially in high signal interference-plus-noise-ratio (SINR) regions.

In this paper, we investigate resource allocation in OFDM based CRNs with cyclic prefix power transfer (CPPT). In this system, the secondary receiver (SR) can extract power from the cyclic prefix (CP) of the received signal and use the harvested energy for its own energy supply. Our objective is to find the optimal power allocation and CP size that maximize the energy efficiency (EE) of the secondary system, under the maximum primary user (PU) interference and the minimum harvested energy constraints. As the CP size will impact the interference constraints, the relationship between optimal power allocation and CP size is hard to be expressed by a function. In this paper, we solve the problem by two steps. First, we propose an efficient iterative algorithm to achieve the optimal power allocation for the EE maximization problem with a certain CP size, and then find the optimal CP size through a full-search method. The influence caused by the CP size on power allocation is illustrated by simulations and numerical results prove CPPT can improve the system utilization greatly compared with traditional methods.

Unlicensed spectrum, around 5 GHz, will be introduced to Long Term Evolution (LTE) systems, referred to as LTE-Unlicensed (LTE-U), to combat the explosive growth of traffic volume in next 10 years. In this paper, the interference-controlled power allocation problem is studied for LTE-U systems, which can be inherently classified as orthogonal frequency division multiplexing (OFDM)-based cognitive radio (CR) systems, where optimal power allocation algorithms are currently available by resorting to computation-intensively numerical iterations with a moderate risk of divergence. In order to satisfy the rigorous algorithm requirements, i.e., convergent outputs of power allocation and running time of milliseconds, for practical LTE-U deployments, a cap-limited waterfilling method is proposed to regulate the interference to primary users one by one successively, by which not only a near-optimal solution can be obtained, but also the intractable iteration divergence and computation complexity issues can be excluded completely. Simulation results indicate the capacity performance of the proposed low-complexity method approaches to the optimal solution with a slight loss less than 5%, and is remarkably superior to the existing suboptimal methods.