Ke He
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Ke He has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Haiyan Wang
School of Marine Science and Technology, Northwestern Polytechnical University, 127 West Youyi Road, Xi’an 710072, China
3 shared publications - Shilei Ma
School of Marine Science and Technology, Northwestern Polytechnical University, Xi’an 710072, China
2 shared publications - Xiaohong Shen
School of Marine Science and Technology, Northwestern Polytechnical University, Xi’an 710072, China
2 shared publications - Changcheng Qiao
The 27th Research Institude of China Electronic Technology Group Corporation, Zhengzhou 450047, China
1 shared publication - Gaoyue Ma
School of Marine Science and Technology, Northwestern Polytechnical University, Xi’an 710072, China
1 shared publication - Haitao Dong
Xi’an Key Laboratory of Intelligent Spectrum Sensing and Information Fusion, Xidian University, Xi’an 710071, China
1 shared publication
Latest Publications
- Open Access
- 10 April 2025
- Published by MDPI AG in Journal of Marine Science and Engineering
- Vol. 13 (4) , 756
Marine Multi-Physics-Based Hierarchical Fusion Recognition Method for Underwater Targets
With the rapid advancement of ocean monitoring technology, the types and quantities of underwater sensors have increased significantly. Traditional single-sensor approaches exhibit limitations in underwater target classification, resulting in low classification accuracy and poor robustness. This paper integrates deep learning and information fusion theory to propose a multi-level fusion perception method for underwater targets based on multi-physical-field sensing. We extract both conventional typical features and deep features derived from an autoencoder and perform feature-level fusion. Neural network-based classification models are constructed for each physical field subsystem. To address the class imbalance and difficulty imbalance issues in the collected physical field target samples, we design a C-Focal Loss function specifically for the three underwater target categories. Furthermore, based on the confusion matrix results from the subsystem’s validation set, we propose a neural network-based Dempster–Shafer evidence fusion method (NNDS). Experimental validation using real-world data demonstrates a 97.15% fusion classification accuracy, significantly outperforming both direct multi-physical-field network fusion and direct subsystem decision fusion. The proposed method also exhibits superior reliability and robustness.
- 10 April 2025
- Published by MDPI AG in Journal of Marine Science and Engineering
- Vol. 13 (4) , 756
Open AccessMarine Multi-Physics-Based Hierarchical Fusion Recognition Method for Underwater Targets
With the rapid advancement of ocean monitoring technology, the types and quantities of underwater sensors have increased significantly. Traditional single-sensor approaches exhibit limitations in underwater target classification, resulting in low classification accuracy and poor robustness. This paper integrates deep learning and information fusion theory to propose a multi-level fusion perception method for underwater targets based on multi-physical-field sensing. We extract both conventional typical features and deep features derived from an autoencoder and perform feature-level fusion. Neural network-based classification models are constructed for each physical field subsystem. To address the class imbalance and difficulty imbalance issues in the collected physical field target samples, we design a C-Focal Loss function specifically for the three underwater target categories. Furthermore, based on the confusion matrix results from the subsystem’s validation set, we propose a neural network-based Dempster–Shafer evidence fusion method (NNDS). Experimental validation using real-world data demonstrates a 97.15% fusion classification accuracy, significantly outperforming both direct multi-physical-field network fusion and direct subsystem decision fusion. The proposed method also exhibits superior reliability and robustness.
- Open Access
Adaptive Intrawell Matched Stochastic Resonance with a Potential Constraint Aided Line Enhancer for Passive Sonars
Remote passive sonar detection and classification are challenging problems that require the user to extract signatures under low signal-to-noise (SNR) ratio conditions. Adaptive line enhancers (ALEs) have been widely utilized in passive sonars for enhancing narrowband discrete components, but the performance is limited. In this paper, we propose an adaptive intrawell matched stochastic resonance (AIMSR) method, aiming to break through the limitation of the conventional ALE by nonlinear filtering effects. To make it practically applicable, we addressed two problems: (1) the parameterized implementation of stochastic resonance (SR) under the low sampling rate condition and (2) the feasibility of realization in an embedded system with low computational complexity. For the first problem, the framework of intrawell matched stochastic resonance with potential constraint is implemented with three distinct merits: (a) it can ease the insufficient time-scale matching constraint so as to weaken the uncertain affect on potential parameter tuning; (b) the inaccurate noise intensity estimation can be eased; (c) it can release the limitation on system response which allows a higher input frequency in breaking through the large sampling rate limitation. For the second problem, we assumed a particular case to ease the potential parameter . As a result, the computation complexity is greatly reduced, and the extremely large parameter limitation is relaxed simultaneously. Simulation analyses are conducted with a discrete line signature and harmonic related line signature that reflect the superior filtering performance with limited sampling rate conditions; without loss of generality of detection, we considered two circumstances corresponding to (periodic signal with noise) and (pure noise) hypotheses, respectively, which indicates the detection performance fairly well. Application verification was experimentally conducted in a reservoir with an autonomous underwater vehicle (AUV) to validate the feasibility and efficiency of the proposed method. The results indicate that the proposed method surpasses the conventional ALE method in lower frequency contexts, where there is about 10 dB improvement for the fundamental frequency in the sense of power spectrum density (PSD).Open AccessAdaptive Intrawell Matched Stochastic Resonance with a Potential Constraint Aided Line Enhancer for Passive Sonars
Remote passive sonar detection and classification are challenging problems that require the user to extract signatures under low signal-to-noise (SNR) ratio conditions. Adaptive line enhancers (ALEs) have been widely utilized in passive sonars for enhancing narrowband discrete components, but the performance is limited. In this paper, we propose an adaptive intrawell matched stochastic resonance (AIMSR) method, aiming to break through the limitation of the conventional ALE by nonlinear filtering effects. To make it practically applicable, we addressed two problems: (1) the parameterized implementation of stochastic resonance (SR) under the low sampling rate condition and (2) the feasibility of realization in an embedded system with low computational complexity. For the first problem, the framework of intrawell matched stochastic resonance with potential constraint is implemented with three distinct merits: (a) it can ease the insufficient time-scale matching constraint so as to weaken the uncertain affect on potential parameter tuning; (b) the inaccurate noise intensity estimation can be eased; (c) it can release the limitation on system response which allows a higher input frequency in breaking through the large sampling rate limitation. For the second problem, we assumed a particular case to ease the potential parameter . As a result, the computation complexity is greatly reduced, and the extremely large parameter limitation is relaxed simultaneously. Simulation analyses are conducted with a discrete line signature and harmonic related line signature that reflect the superior filtering performance with limited sampling rate conditions; without loss of generality of detection, we considered two circumstances corresponding to (periodic signal with noise) and (pure noise) hypotheses, respectively, which indicates the detection performance fairly well. Application verification was experimentally conducted in a reservoir with an autonomous underwater vehicle (AUV) to validate the feasibility and efficiency of the proposed method. The results indicate that the proposed method surpasses the conventional ALE method in lower frequency contexts, where there is about 10 dB improvement for the fundamental frequency in the sense of power spectrum density (PSD). - Open Access
Path Diversity Improved Opportunistic Routing for Underwater Sensor Networks
The packets carried along a pre-defined route in underwater sensor networks are very vulnerble. Node mobility or intermittent channel availability easily leads to unreachable routing. Opportunistic routing has been proven to be a promising paradigm to design routing protocols for underwater sensor networks. It takes advantage of the broadcast nature of the wireless medium to combat packet losses and selects potential paths on the fly. Finding an appropriate forwarding candidate set is a key issue in opportunistic routing. Many existing solutions ignore the impact of candidates location distribution on packet forwarding. In this paper, a path diversity improved candidate selection strategy is applied in opportunistic routing to improve packet forwarding efficiency. It not only maximizes the packet forwarding advancements but also takes the candidate’s location distribution into account. Based on this strategy, we propose two effective routing protocols: position improved candidates selection (PICS) and position random candidates selection (PRCS). PICS employs two-hop neighbor information to make routing decisions. PRCS only uses one-hop neighbor information. Simulation results show that both PICS and PRCS can significantly improve network performance when compared with the previous solutions, in terms of packet delivery ratio, average energy consumption and end-to-end delay.
Open AccessPath Diversity Improved Opportunistic Routing for Underwater Sensor Networks
The packets carried along a pre-defined route in underwater sensor networks are very vulnerble. Node mobility or intermittent channel availability easily leads to unreachable routing. Opportunistic routing has been proven to be a promising paradigm to design routing protocols for underwater sensor networks. It takes advantage of the broadcast nature of the wireless medium to combat packet losses and selects potential paths on the fly. Finding an appropriate forwarding candidate set is a key issue in opportunistic routing. Many existing solutions ignore the impact of candidates location distribution on packet forwarding. In this paper, a path diversity improved candidate selection strategy is applied in opportunistic routing to improve packet forwarding efficiency. It not only maximizes the packet forwarding advancements but also takes the candidate’s location distribution into account. Based on this strategy, we propose two effective routing protocols: position improved candidates selection (PICS) and position random candidates selection (PRCS). PICS employs two-hop neighbor information to make routing decisions. PRCS only uses one-hop neighbor information. Simulation results show that both PICS and PRCS can significantly improve network performance when compared with the previous solutions, in terms of packet delivery ratio, average energy consumption and end-to-end delay.