Agent Skills
› brycewang-stanford/Awesome-Journal-Skills
› mobicom-writing-style
mobicom-writing-style
GitHub指导 MobiCom 论文写作，强调首段明确机制与测量证据、声明运行环境、用百分位替代最高级，并适配双栏12页格式。
Trigger Scenarios
起草 MobiCom 论文
修改 MobiCom 论文以符合风格
Install
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill mobicom-writing-style -g -y
SKILL.md
Frontmatter
{
"name": "mobicom-writing-style",
"description": "Use when drafting or revising a MobiCom paper for house style — putting the wireless\/mobile mechanism and its operating regime on the first page, pairing every claim with over-the-air measurement, replacing superlatives with percentiles and intervals, and fitting the argument into the double-column 12-page body."
}
MobiCom Writing Style
A MobiCom paper reads as a mechanism justified by measurement. The first page states a wireless or mobile-networking problem, the mechanism that addresses it, and the measured evidence — before any implementation detail. The double-column 12-page body is for the core argument; derivations, protocols, and extra results go to references and appendices.
The first-page arc
Put this in the abstract and first page, in order:
- The wireless/mobile pain — the concrete condition that breaks existing designs (a fade under mobility, interference, a spectrum constraint, an energy limit).
- Why existing mechanisms fall short — each prior approach gets a specific, named failure tied to the condition, not "prior work is limited."
- The mechanism — what is new, stated as a networking/PHY-MAC/routing/sensing mechanism, with its operating regime named (distance, mobility, channel, power).
- The measured evidence — the headline result with its condition and its uncertainty, not a bare "up to N×."
- Why it matters for mobile networking — the community-level payoff.
See ../../resources/worked-examples/01-introduction.md
for a before→after that executes this arc.
Name the operating regime
The single most common wireless-writing failure is a claim with an unstated envelope. Every headline claim carries its regime:
| Vague | MobiCom-calibrated |
|---|---|
| "achieves high throughput" | "sustains X at 1-4 m NLOS under pedestrian mobility" |
| "low power" | "energy-per-bit within the harvested budget at Y% duty cycle" |
| "robust to interference" | "holds delivery above threshold at Z dB co-channel interference" |
| "works in real settings" | "measured over the air in a furnished room with walkers" |
A reviewer cannot fault — or reward — a claim whose conditions are hidden. Stating the regime is not hedging; it is the claim.
Claims are paired with evidence
Every major claim maps to a proof, a figure, a measured condition, or a reproducibility location. Build the map before submitting:
Claim -> evidence pairing (fill for each headline sentence):
"decodable through fades" -> Fig. 3 (delivery vs walker path) + channel traces
"within energy budget" -> Fig. 4 (energy-per-bit) + instrument note (App. B)
"no dedicated carrier" -> system description Sec. 3 + power measurement
An unpaired claim is either cut or demoted to future work.
Percentiles, not superlatives
- Report distributions and confidence intervals, not single best runs or unqualified
maxima. "Up to N×" without the spread behind it is a review risk (
mobicom-experiments). - Prefer the median and a tail over the mean alone when the metric is delivery or latency; wireless distributions are skewed.
- Avoid "significantly" as prose; show the interval and let it speak.
Fitting the double-column 12 pages
- Figures and tables count against the 12; a radio-testbed diagram and three CDFs is
real page budget. Triage content, never geometry (
mobicom-submission). - The body carries the mechanism and its headline evidence; channel-model derivations,
full protocols, extended sweeps, and instrument details go to appendices after the
references, outside the cap (
mobicom-supplementary). - Keep the introduction's roadmap to one or two sentences; over-signposting substitutes for argument.
- Do not put the strongest result only in an appendix — reviewers should not reconstruct the contribution from supplementary material.
Prose discipline
- Lead sections with the finding, then the method — not a chronological lab diary.
- Name the abstraction or mechanism once and use the name consistently.
- Separate engineering effort from research contribution: building a testbed is work, but the contribution is the mechanism it demonstrates.
Revision checklist
- Mechanism and operating regime on the first page.
- Each prior-work line has a specific, named failure.
- Every headline claim paired to a figure, condition, or appendix.
- Distributions/intervals replace superlatives throughout.
- Figures/tables budgeted against the 12 double-column pages.
- Strongest result in the body, not only the appendix.
Output format
[First page] mechanism + regime present? y/n — what is missing
[Regime] which headline claims lack a stated envelope
[Claim map] unpaired claims listed for cut-or-support
[Superlatives] flagged phrases -> distribution/interval replacement
[Page budget] over/under 12; what to move to appendix
Version History
- 9f86f09 Current 2026-07-19 17:04
