Searcharxiv⌕ Search

arXiv subjects

Guangbo Chen

Publications and source records attributed to Guangbo Chen.

2 recordsLinked to original sources

HERO-D3QN: Inter-Cluster Routing with Energy-Aware Soft Constraints in Wireless Sensor Networks

Inter-cluster forwarding in wireless sensor networks can concentrate energy consumption on frequently selected relays, while hard energy thresholds restrict forwarding choices as residual energy declines. We propose Hierarchical Energy-balanced Routing Optimization based on a dueling double deep Q-network (D3QN), termed HERO-D3QN. Given the cluster structure, the method combines direction- and cost-based relay screening with energy-aware soft constraints. Relays below an adaptive energy threshold remain eligible if the other selection criteria are met. An enhanced state representation captures relay energy, progress toward the sink, potential forwarding load, and hotspot risk, and a multi-objective reward guides learning. In the comparison with D3QN using the same node locations, HERO-D3QN delayed first-node death (FND) from time step 629 to 1269 and reduced peak residual energy variance by 55.5%. In the ablation study, the full method achieved the latest FND and half-node death (HND) and the largest cumulative transmitted volume (170.52 Mbit) among the tested variants. Across five seed settings in which node locations and the random processes used in training and evaluation varied jointly, the coefficients of variation of FND, HND, and transmitted volume were 8.26%, 2.75%, and 3.88%, respectively. These results demonstrate the effectiveness of HERO-D3QN in delaying early node failure and reducing peak energy dispersion.

cs.NI↗

Alloyed Re$_x$Mo$_{1-x}$S$_2$ Nanoflakes with Enlarged Interlayer Distances for Hydrogen Evolution

Molybdenum sulfide (MoS$_2$) has attracted significant attention due to its great potential as a low-cost and efficient catalyst for the hydrogen evolution reaction. Developing a facile, easily upscalable, and inexpensive approach to produce catalytically active nanostructured MoS$_2$ with a high yield would significantly advance its practical application. Colloidal synthesis offers several advantages over other preparation techniques to overcome the low reaction yield of exfoliation and drawbacks of expensive equipment and processes used in chemical vapor deposition. In this work, we report an efficient synthesis of alloyed Re$_x$Mo$_{1-x}$S$_2$ nanoflakes with an enlarged interlayer distance, among which the composition Re$_{0.55}$Mo$_{0.45}$S$_2$ exhibits excellent catalytic performance with overpotentials as low as 79 mV at 10 mA/cm2 and a small Tafel slope of 42 mV/dec. Density functional theory calculations prove that enlarging the distance between layers in the Re$_x$Mo$_{1-x}$S$_2$alloy can greatly improve its catalytic performance due to a significantly reduced free energy of hydrogen adsorption. The developed approach paves the way to design advanced transition metal dichalcogenide-based catalysts for hydrogen evolution and to promote their large-scale practical application.

cond-mat.mtrl-sci↗