A



Prof. Vincent Lau’s Paper Review Checklist

Headers/Footers

← Title is clear and very concise. Hopefully, reader can see this work is different from the rest from the title (the best case). The worse case is the title give reader an impression that this is just another paper in xxx area.

← Date must be included either below authors or in footer

← Journal footer must include information of prior conference submissions (conference and date) if closely related.

← Journal footer should include contact info, funding info (if relevant) – this saves space vs. in main header

← Page numbers on every page

← Do not include other “junk” on every page like “DRAFT”, title, author names, etc.

Abstract

← As concise as possible. Every sentence is essential and cannot be compressed.

← Avoid acronyms and abbreviations and references

← Math used only if essential to the result, for example a scaling law

← First sentence (or 2) motivates the problem

← Last sentence (or 2) summarize the key takeaways/contribution of the paper

Introduction

← MUST TELL A SIMPLE and ATTRACTIVE STORY with a STRONG SKELETON of story flow. {When you tell a story, you do it top down… introduce the big picture first then details later… in a hierarchical manner}

← Starts with motivation and context (1 paragraphs) – [Why this area is important??]

← Includes discussion of key prior works (at least one paragraph) properly organized and grouped (classified) – e.g. this group of papers considered this case, the other groups of papers considered that case. [What people have done in such an important area?]

← Highlight (e.g. in bullet forms) What are the remaining important issues / challenges not yet addressed by the others? {List in descending order of importance – only list the first order challenges and write it in simple language that is easy to be comprehended even by a laymen}. {Imagine you are asking for funding support from an agency to do research in this area…. You have to convince the manager (who is not exactly in this area) why they have to “waste $$” working in this area given there are a lot of works out there already…}

← Contributions of the paper (summarized) (1 paragraphs for journal papers)

← Paper organization provided (not required for conference papers).

← If complicated notation is used, it should be summarized at the end. For example order notation (i.e. big O, little o), matrix notation.

← Introduction litmus tests: Can someone understand your contribution, why it is important, and how it fits in with the field just from reading the introduction? Would a reviewer be able to accept your paper based just on the introduction? (Assuming no errors in the main body of course).

System Model and Preliminaries

← Sets up the system model used in the paper and provide key results you may use repeatedly

← Use only the most general assumptions needed for the paper to be technically correct. For example, do not assume Gaussian channel or noise, Poisson node distribution, etc unless you need those in all parts of paper.

← If you have a very large number of variables, I suggest putting a table in this section with all of them defined (and their units, if relevant)

Notation

← Notation is the KEY BACKBONE of the paper. A good notation system can substantially simplify the presentation and the derivation of results.

← A GOOD NOTATION is something that reader can understand / guess its meaning even without referring back to the definition all the time.

o E.g. Make consistent use of index (e.g. m always refer to the node #), (k always refer to the user index)..e.tc. and K (related to k) denote the total number of users….etc.

o Do not have more than 3 levels of subscripts… e.g. X_{m,n,p}… this will be quite nasty… think of other ways to represent it.

o Always use STANDARD notations.. e.g. H for fast fading channel gain, X for transmit symbol, bold for matrix or vector, \mathbb{E} for expectation…etc.

o

←

Body of Paper

← WRITE THE MAIN BODY in hierarchical manner too… (Top down).

o A good paper will allow the reviewer to understand the flow at different level. E.g. if a reader is in a hurry, he can still follow the main skeleton WITHOUT having to understand EVERYTHING first in order to comprehend your statements.

o In other words, a GOOD PAPER allows the flexibility of the reader to CHOOSE to understand your paper at DIFFERENT LEVELS of depth.

o A BAD paper is something that you DON”T HAVE A CHOICE… either you have to understand it to the LOWEST LEVEL to get your logical flow OR cannot understand anything at all. i.e. ALL or NOTHING.

← Before formulating a problem, make sure you define

o Your optimization variables: e.g. what is your control policies… and the associated constraints…. Explain the physical meaning of these…

o Your optimization objective function. E.g. derive your system average throughput and express it in terms of your “variables” (control policy) to be optimized.

o Problem formulation: Formally write out a core problem formulation as your “STARTING POINT”….

o

← Theorems are clearly stated when appropriate. Long proofs are put in appendices at the end.

← Name the Theorem (with some meaning) if possible. e.g. “Theorem 1 (Asymptotic Throughput)” …balabala…..

← Every theorem has some discussion explaining what it means. Do not end a section with a theorem.

← Sections are not overlength, i.e. break a long section (more than 5 pages or so) into 2 shorter sections

← Sections with a subsection A also have a subsection B

Conclusion

← Summarizes the main results and take aways of the paper. Future work can be suggested.

← Not simply a restatement of the abstract. Distinct from the abstract in the sense that usually motivation, key assumptions, and context is not given here.

References

← All relevant work is cited. Journal papers should have around 20 references, conference papers around 6-10 (depending on the page limit).

← Cite current and recent work when possible, including unpublished work that is of interest (this shows you know the state of the art).

← When there are both journal and conference versions of the same paper, cite the journal version.

← All unpublished work should include a web URL to facilitate the reviewing process, particularly those of your own papers. ( is preferred, but your personal webpage is acceptable).

← All citations are properly formatted. Authors’ first names are abbreviated. Reference is complete including page and volume numbers. All journals are abbreviated consistently.

← Strongly recommend using Bibtex to organize your references. This saves massive amounts of time in the long run.

Figures

← Each Figure must have one or two messages to be delivered. These messages MUST BE CLEAR (first order) message. No two figures should deliver similar messages (and if so, merge them).

← All figures are clearly legible

← Figures have legends, with the legend order matching the order of the curves, i.e. top curve appears on the top of the legend, and so on. (use arrow to point legends to the curve)

← Axes are labeled consistent with terminology in paper and clear in their meaning

← Avoid grids which clutter and make figures hard to read when published

← Do not rely on color to distinguish different curves; Use appropriate font sizes

← Figures should be saved in .eps or .pdf format (also save it in .fig format in case later on, you need to revise the figures).

← Figures, Plots, Block diagrams are never copied from another paper not by you. Even self-copying should be avoided – at least change something on the figure (this is a copyright issue).

← Remember to write a concise figure caption. Indicate the parameters setting in the caption as well so that when people read your figure, they can know what it is talking about by just looking at the caption.

Common Writing and Formatting Suggestions

← All acronyms are defined

← Equations are treated like parts of a sentence. This means that punctuation is used appropriately. If an equation is at the end of a sentence then it should be followed by a period. Do not use colons before equations unless that would be part of a normal sentence.

← All variables defined before or after equation. If it’s been several pages since it’s last use, remind the reader what it means.

← Manuscript conforms to all guidelines for the journal or conference in question. Use one inch margins and 8.5” x 11” paper size unless otherwise noted. Do not play any games with the font or spacing unless explicitly asked to by Prof. Andrews.

← Avoid superscripts in variables – they get confused with transpose, exponents, and Hermitian.

← Section titles: All consistent with uppercase initials.

← No sentence starts with And, But, an equation number, or a reference

← Equations referenced like (6). Figure abbreviated like Fig. 6. Section abbreviated like Section 6 (use roman numbering depending on IEEE format).

Other Guidelines

← I prefer to get complete papers (journal versions) and then cut down to conference versions. Journal versions should always be single column, 1.5-2 lines of spacing. Conference versions can be turned in in “camera ready form” (2 columns, smaller font), etc.

← When you send me the complete journal paper, you must feel that it is CAMERA-READY (you have the confidence to submit it even without my comments).

o Ie. You should complete all the things that can be handled by you.. e.g. spell check, consistent notations, …etc.

o Depending on the target journal, you should conform to the journal’s requiement. E.g. 20-21 pages of text for IEEE Transactions on Wireless Com. Virtually no page limit for IEEE Transactions on Information Theory, Transactions on Networking, Transactions on Signal Processing.

o

← When I pass you comments, I expect to only do some minor touch up on the pdf. If I have to type in directly on .tex, then you will be in trouble!!

← Based on the cleaned Journal version, you could reduce it to a conference version.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download