BUPT researcher spanning LLM-driven LEO satellite routing and task-oriented semantic communication for remote sensing.
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-02) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Not yet scored: the visible record contains only two 2026 arXiv preprints, with no verified publication count, venue history, or authorship volume. This axis would measure total research output, venue tier, and first/corresponding-author share.
Impact—
Not yet scored: no citation counts, h-index, or downstream-adoption signals exist in the available data. This axis would measure citation impact and influence on follow-up work in satellite networking and semantic communications.
Focus0/100
Scores 0/100: the field distribution is an even 50/50 split between computer networks (LEO satellite communications) and satellite remote sensing / semantic communications, with one 2026 preprint in each bucket. A low concentration score flags breadth rather than depth in any single field, though both lines share a common satellite-systems theme, suggesting a coherent application-driven agenda rather than scattered interests.
Collaboration39/100
Scores 39/100 with 6 distinct coauthors on record. Recurring collaborators are Mugen Peng, Hongyu Wang, Bowen Lu, Yaohua Sun, Kerui Guo, and Shuoyuan Sun, forming a compact cluster around both 2026 preprints. Xu appears as last author — not corresponding author — on each paper, a placement often associated with a supervisory role; the small coauthor footprint indicates a stable single-team pattern rather than broad external collaboration.
Activity—
Not yet scored: activity cadence cannot be computed from a single publication year (2026). This axis would measure publication frequency, year-over-year consistency, and recent engagement in conferences and preprint servers.
Reproducibility55/100
Scores 55/100 across 2 reviewed papers, each rated medium reproducibility readiness. Evolve's reviews temper confidence: STR-Agent's headline numbers (intent satisfaction 45%→92%, ~60% delay reduction) come from a synthetic evaluation with weak baselines, and the semantic-transmission comparison uses baselines without task-aware training. Releasing code, trained agents or models, and evaluation scenarios would be the clearest path to stronger verification.
Research summary
Wenjia Xu is a researcher at Beijing University of Posts and Telecommunications (BUPT) whose cataloged work centers on satellite communication systems at the intersection of networking, large language models, and semantic communications. This is an early-stage profile built from two 2026 arXiv preprints in the Evolve catalog, on both of which Xu holds the last author position. The first, STR-Agent, presents an LLM-driven agent that translates natural-language operator requests into QoS-adaptive routing in LEO satellite networks, built on supervised fine-tuning over a Dijkstra-based routing substrate; Evolve's review credits large reported gains — intent satisfaction rising from 45% to 92% and roughly 60% delay reduction — while flagging a synthetic evaluation and weak baselines. The second, on task-oriented semantic feature transmission, sends frozen-backbone features for multi-task satellite remote sensing through a small convolutional channel adaptation module, reporting large low-SNR gains over reconstruction-based joint source-channel coding, again with baseline caveats. The field distribution is evenly split (50/50) between LEO satellite communications and satellite remote sensing / semantic communications, yielding a low focus score (0/100), and the collaboration footprint is compact: six distinct coauthors, anchored by Mugen Peng, Hongyu Wang, Bowen Lu, Yaohua Sun, Kerui Guo, and Shuoyuan Sun. Reproducibility is scored 55/100 across two papers, both rated medium readiness, while output, impact, and activity dimensions remain unscored for lack of data.
Collaboration pattern
All cataloged output is co-authored within one compact cluster: Mugen Peng, Hongyu Wang, Bowen Lu, Yaohua Sun, Kerui Guo, and Shuoyuan Sun are the six recorded collaborators. Xu is last author on both 2026 preprints without corresponding-author status, a placement typically tied to a supervisory or senior-contributor role. The modest 6-coauthor footprint and 39/100 collaboration score point to stable intra-group collaboration rather than wide external partnership — a pattern worth watching as the record grows.
Research keywords
Highlights
STR-Agent: LLM-driven QoS-aware routing for LEO satellite networks
Task-oriented semantic feature transmission over low-SNR satellite channels
Timeline
- 2026STR-Agent arXiv preprint: an LLM-driven agent for QoS-aware routing in LEO satellite networks, with Xu as last author.
- 2026Task-oriented semantic feature transmission preprint appears on arXiv: multi-task satellite remote sensing over low-SNR channels, again with Xu as last author.
- 2026Both cataloged papers land in the same six-person coauthor cluster, spanning LEO satellite networking and semantic communications for remote sensing — the full extent of the current record.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometric dimensions are computed from the public bibliometric data behind the Evolve radar; output, impact, and activity are unscored because the underlying data is not yet available.
- Coverage is limited to two 2026 arXiv preprints in the Evolve catalog — an early-stage profile from which year-by-year output cadence cannot be established.
- Both papers carry Evolve confidence caveats: synthetic evaluation with weak baselines (STR-Agent) and baselines lacking task-aware training (semantic feature transmission).
- Country metadata is not provided; identity is name-resolved from public bibliographic data tied to ORCID 0000-0002-1425-4162.