Introduce the overall situation(technology advances, theoretical reasons, industry needs, etc.) leading to the problem to be solved. A few paragraphs are sufficient.
State the problem clearly and concisely. List the criteria for solutions and key goals to be achieved in solving the problem. Use a numbered list format for easy reference in subsequent sections.
Review of recent (within the last 5 years) works on solving this problem. Referring to the list in section 2, discuss how these previous techniques work and what are their technical strengths and shortcomings. Do not disparage other people's work just to make your work look good: your work is good only when it solves the problem. Be fair in criticism and support your views with technical reasons, not personal opinions.
Present an in-depth discussion of the theoretical aspects of your proposed approach. Use fundamental theory (EE, Math, Physics, etc.), circuit design, circuit analysis, algorithms, etc. Analyze the efficiency of your approach after the presentation, again using theoretical analysis. Theoretical results may be plotted or illustrated by figures and / or tables for later comparison with simulated and experimental results. Analyze the complexity, overhead, cost, etc. of your approach. This is the most important part of the paper. All approaches must be theoretically solid and justifiable. Trial-and-error and lucky good guesses are not acceptable.
Simulate your approach using CAD tools such as Matlab (behavioral), Spice (circuit level), Verilog (logic level and timing analysis), etc. Generate figures and / or tables of results and compare with the theoretical results in section 4.1 to support the theoretical analysis. Discuss agreements and discrepancies. Choose realistic case studies, not just those that your approach happens to work well with. Do not choose "toy" problems as case studies. For example, an 8-bit ADC is a valid case study, a 3-bit ADC is "toy." Some conferences accept papers with simulation results but others demand more (see section 4.3).
Use subsections if necessary if you have a few case studies.
Present experimental results from real ICs, boards, circuits, etc. In analog and mixed-signal test and design-for-test, "experimental data" refers to real circuit and board data, not to "simulated" data. Simulation alone is not sufficient to support your theory since simulation is not accurate and does not include many real-life effects. High-quality conferences and journals demand experimental results before publications. Plan the experiment well in advance since it takes at least one year to get real test data from ICs and board from the time the design begins.
* All Ph.D. theses and journal papers must include experimental data *
Compare experimental results in figures / tables with simulated results (section 4.2) and theoretical predictions (section 4.1). Discuss agreements and discrepancies.
Use subsections if necessary if you have a few case studies or if you want to present the design and test procedure for the IC/ board case study in a systematic manner.
Compare honestly your work with previous works reviewed in section 3, using the list of goals / criteria established in section 2. Discuss where your solution is better, where it is weaker, etc. Use a table with criteria as column headings and solution methods (others and yours) as row headings to show tradeoffs. Use figures to illustrate differences in solutions. Check to make sure that your approach actually solves the problem posed in section 2.
No work is perfect. Discuss honestly the limitations of your proposed approach. Limitations may be theoretical or experimental or practical (e.g. manufacturing cost increasing too fast with circuit complexity, etc.). Is it possible to overcome some limitations while others are totally unavoidable?
Summarize the problem and your proposed approach, with its strong and weak points. A short paragraph is sufficient.
Acknowledge sponsorship (NSF, SRC, specific companies for chip fabrication, etc.) and technical discussions with other colleagues, if any.
List appropriate works referred to in the paper. Use recent works (in the last 5 years).