6 November 2024 in IEEE Transactions on Aerospace and Electronic Systems
The increasingly intelligence of electronic warfare jamming techniques and the threat of main-lobe jamming have significantly deteriorated radar detection capabilities. As a result, it is essential for radar systems to implement active and advanced countermeasures to effectively combat such interference. Current radar active countermeasures, although effective in anti-jamming, often compromise radar operations by consuming valuable and limited resources of radar. Inspired by the use of electronic warfare pods to evade radar detection, a collaborative framework is designed between radar and radar self-defense jammer (RSDJ) for active defense. This paper proposes RSDJ-aided strategies to enhance radar anti-jamming capability and survivability, while significantly reducing the demand on radar's operational resources. Considering the practical limitations on available frequency resources, we devise dual coordination strategies that optimize the collaboration between the radar and RSDJ. The coordination strategies unitizes different frequency resources constrained by the requirement of mutual interference reduction. Confronting the unpredictability of adversary jamming in practice, this work employs reinforcement learning (RL) to develop coordination strategies without the need for pre-training or prior information. The application of RL enables the RSDJ to effectively deceive the adversary jammer and enhance radar anti-jamming capabilities in unknown environments. Comparative simulations demonstrate the effectiveness of proposed strategies, showing a notable improvement in radar anti-jamming performance and interference mitigation of nearly 50%. This advancement significantly boosts the adaption of radar operations amidst complex electronic warfare environments.
1 January 2020 in Oncologie
Predicting the onset and metastasis of early tumor is the primary means of improving lung cancer prognosis. The purpose of this study was to identify the ability of plasma metabolomics combined with blood markers to establish benign lung disease versus lung cancer regression models. Blood samples were collected from 174 lung cancer patients, 350 benign lung disease patients and 100 healthy volunteers and the metabolites were analyzed by mass spectrometry. The target metabolites consisted of 23 amino acids, 26 acylcarnitines and 45 conventional blood markers. A regression analysis model was established based on 12 metabolites and five blood markers selected by elastic network analysis. Two-thirds of the data were used in a training set for modeling and signature construction, and the remaining one-third were used in a validation set to test the model. This model was identified to have good specificity and sensitivity in distinguishing between lung cancer and lung disease. The performance of the model was evaluated using the area under the receiver operating curve, which was 0.915 in training set and 0.875 in validation set. In conclusion, this study demonstrates that regression model established by plasma metabolomics in combination with conventional serum markers has potential for the diagnosis of lung cancer.
16 September 2025 in IEEE Transactions on Aerospace and Electronic Systems
Cooperative track spoofing (CTS) poses an unsolved conundrum to netted radar systems (NRS) by creating increasingly persistent and precise phantom tracks. Existing anti-deception methods in NRS struggle to counter this sophisticated joint range-angle deception. To address this pressing challenge, we propose a supervisory-review mechanism with its parameter optimization to identify CTS while minimizing its impact on tracking and searching tasks of NRS. The mechanism begins with a supervisory process that involves selecting the supervisory frame and radar node, along with scheduling resources during dwell time to check targets. Following this, binary hypothesis testing in the review process determines whether to repeat the supervisory process. These steps thus enable NRS with supervisory-review mechanism to detect CTS. To enhance effectiveness, the parameter optimization incorporates a hybrid policy for selecting supervisory radar nodes and employs reinforcement learning (RL)-based resource scheduling via a constructed Markov decision process (MDP). Simulations in complex multi-target scenarios validate the effectiveness of the proposed mechanism in recognizing CTS and demonstrate the advantages of RL-based resource scheduling.
1 January 2026 in IEEE Transactions on Aerospace and Electronic Systems