In modulation classification domain, handcrafted feature based method can fit well from a few labeled samples, while deep learning based method require a large amount of samples to achieve the superior classification performance. In order to improve the modulation classification accuracy under the constraint of limited labeled samples, this paper proposes a few-shot modulation classification method based on feature dimension reduction and pseudo-label training (FDRPLT), which combines handcrafted feature based method with deep learning based method. First, an optimal low-dimensional feature subset is created by the combination of the handcrafted features and autoencoder-extracted features post-processed by a feature selection algorithm. Then, a fully connected network (FCN), trained on a small number of labeled signals, is designed for the automated annotating, where unlabeled samples can be annotated and used for the later convolution neural network (CNN) training. The simulation results show that the classification accuracy of eight kinds of modulation types can reach to 98.3% when the SNR is 20dB.
In order to solve the optimization of the interference resource allocation in communication network countermeasures, an interference resource allocation method based on the maximum policy entropy deep reinforcement learning (MPEDRL) was proposed. The method introduced the idea of deep reinforcement learning into the communication countermeasures resource allocation, it could enhance the exploration of the policy and accelerate the convergence to the global optimum with adding the maximum policy entropy criterion and adaptively adjusting the entropy coefficient. The method modeled interference resource allocation as Markov decision process, then established the interference strategy network to output allocation scheme, constructing the interference effect evaluation network of the clipped twin structure for efficiency evaluation, and trained the policy network and the evaluation network with the goal of maximizing the strategy entropy and the cumulative interference efficacy, then decided the optimal interference resource allocation scheme. The simulation results show that the algorithm can effectively solve the resource allocation problem in communication network confrontation, comparing with the existing deep reinforcement learning methods, it has faster learning speed and less fluctuation in the training process, and achieved 15% higher jamming efficacy than DDPG-based method.
Specific emitter identification (SEI) aims to distinguish different emitter individuals based on the subtle differences in the signals. The technology can be widely applied to various wireless communication systems. Existing individual identification methods often ignore the radio frequency (RF) fingerprint information carried by the waveforms and thus can be interfered by unreliable RF features. To alleviate these issues, we devise the deep bidirectional long short-term memory (DBi-LSTM) and the one-dimensional residual convolution network with dilated convolution and squeeze-and-excitation block (Conv-OrdsNet). We exploit the combination of DBi-LSTM and Conv-OrdsNet (CoBONet) to extract temporal structure features directly from baseband in-phase and quadrature (I/Q) samples. The proposed network is able to capture the fine-grained details of signals and combine different information extracted by two networks. Moreover, we propose a data augmentation method to solve the interference of unreliable features by randomly changing the values of noise, power, frequency offset (FO), and phase offset (PO). It is worth noting that our method only needs a short slice of a steady-state signal without complex preprocessing, which reduces the cost of acquisition and calculation. Extensive experiments show that our method can effectively extract reliable RF fingerprinting features from I/Q samples and the classification results are better than most existing methods.
In order to solve the problem of insufficient labeled samples in modulation recognition, this paper proposes a few-shot modulation recognition algorithm based on pseudo-label semi-supervised learning (pseudo-label algorithm). First of all, high quality artificial feature, excellent classifier and data-labeling method are used to build efficient pseudo label system, and then the pseudo label system is combined with signal classification method based on the deep learning to realize the modulation classification under the condition of a small number of labeled samples and a large number of unlabeled samples. The simulation results show that the pseudo-label algorithm can improve the model recognition performance by 5%-10% when the six kinds of digital signals are classified and identified and its SNR is greater than 5 dB. At the same time, the algorithm has a simple network design and is of great application value.
Planarians, the first kind of animal to have evolved a brain structure yet has not evolved vision, were demonstrated to have a capability of spatial learning in the last several decades, but what does the navigation of planarians depends on is still unknown. Here, we provide an objective, strictly variable-controlled planarian training method using 3D printing techniques fabricated mazes. Then we use modifications of the mazes to first demonstrate a learning paradigm that worms can memorize the location of a darkened surrounding through training. However, a memory formation failure was found that in the situation of providing identical shapes in a maze, planarians cannot memorize the location of the darkened surrounding. Thus, this result shows the planarians associated darkness with the crude shape of the objects they’ve crawled, which is a kind of spatial learning. This finding not only provides a key insight into spatial learning information that planarians are processing, but also an interpretation of the origin of memory formation where higher grades of memory formation might originate from.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.