Zesi Pan

Intelligent decision-making is an essential sub-task for wireless electromagnetic countermeasures, with the objective of imposing effective electronic jamming based on the observation and estimation of rival equipment. With the proliferation of radar networks, the effectiveness of a single jammer is being questioned, raising research interest in jamming resource allocation (JRA) of cognitive jamming system. This paper presents a study on the multi-agent collaborative jamming power allocation (MACJPA) task to achieve the overall barrage jamming of a radar network using limited jamming resources. This task is first formulated as a decentralized partially observable Markov decision process (Dec-POMDP) to be oriented towards the fully cooperative setting. And we introduce an improved centralized-training-withdecentralized-execution (CTDE) paradigm, namely, designing an “Attention on Local Information” aided multi-agent proximal policy optimization (ALI-MAPPO) network. The core motivation of ALI is that the customized global information with weights is more conducive to learning effective individual value functions. Through adequate network training, the cognitive jamming system utilizes observation information and well-trained policies make optimal decisions individually. In addition, we incorporate radar anti-jamming signal processing features, such as frequency agility and uncertainty in parameter estimation, into the scenario design. Extensive experimental results demonstrate that the ALI-MAPPO enables the cognitive jamming system to generate robust policies and obtain the best jamming performance compared to other existing MARL methods.

Developing intelligent jamming methods to combat the multi-function radar (MFR) has become a vital task in electronic warfare, because the MFR can tune into different working modes according to surrounding environment information. In this paper, we present a solution to the limitations of conventional jamming methods, including strong prior knowledge dependence and inaccurate selection strategies.Specifically, we study a joint optimization of jamming type selection and power control task (JO-JTSPC) for a general radar countermeasure scenario. In particular, we first model the sequential decision-making task JO-JTSPC as a Markov Decision Process (MDP). Subsequently, considering the differences in the designed action space, we accordingly develop two algorithms, i.e., dueling double deep Q-learning (D3QN) and hybrid proximal policy optimization (HPPO), to solve the optimization problem. Taking into consideration the threat level of various MFR working modes and the corresponding required jamming effect, we elaborately design the reward function of MDP as a weighted summation of the mode switching factor, jamming performance factor and jamming power factor. Further, the learned polices of these algorithms are derived based on the designed reinforcement learning (RL) elements. Extensive simulation results demonstrate that the proposed algorithms can learn highly adaptive polices in the radar countermeasure scenarios and achieve good jamming performance.

Existing studies for automatic waveform recognition of overlapping signals have mostly been conducted in a supervised manner. Although demonstrating superior performance in recent years, supervised methods rely heavily on sufficient labeled samples, rely but the acquisition of annotated data is expensive, time-consuming, and sometimes rely infeasible. This shortage drives the need for semi-supervised learning methods, where the unlabeled samples can be fully exploited in the training stage. In addition, multi-instance multi-label (MIML) learning is essentially another weakly supervised learning protocol, and precisely fits the form of the time-frequency images TFIs obtained from the transformation of overlapping signals. In this paper, delving into the MIML learning problem, we leverage the advantage of adversarial training to formulate an effective algorithm MIML-GAN, which is tailored to the MIML problem of overlapping signal waveform recognition. After feeding the TFIs into the network, MIML-GAN approximates the distribution of the training data using the adversarial learning principle. Subsequently, the baglevel prediction can be derived from the instance-level prediction upon the MIML discriminator through adaptive threshold calibration. Specifically, we elaborately studied the global optimality of the MIML-GAN objective function, and extensive simulations are carried out with overlapping signal dataset, validating the ascendancy of the proposed method. Comparative experiments demonstrate that the proposed algorithm possesses promising feature representation capability, and outperforms the existing semi-supervised and supervised signal waveform recognition approaches.

In an ever-increasingly complex electromagnetic environment, multiple low probability of intercept (LPI) radar emitters may transmit their own signals simultaneously on similar bands, resulting in overlapping receiving signals in both time and frequency domain. In this letter, a novel Multi-Instance MultiLabel learning framework based on Deep Convolutional Neural Network (MIML-DCNN) is proposed to automatically recognize the overlapping LPI radar signals,which is trained by single type of signals only. The framework handles signals in an end-to-end manner that is integrated with a well-designed instance generation module, a sophisticated MIML classifier, and an adaptive threshold calibration. Through comprehensive experiments on simulated overlapping signals with four different modulation types, we prove that the proposed framework identifies each individual signal type precisely in the presence of overlapping signals, and is also robust to variation of the signal-to-noise ratio (SNR) and power ratio conditions.