Digital Systems RESEARCH PRESENTATION Digital Systems Research Presentation

Digital Systems RESEARCH PRESENTATION Digital Systems Research Presentation Mohammad Sharifkhani © 1

Lecture 3: Research Methodology II Digital Systems Research Presentation Mohammad Sharifkhani © 2

Research Methodology and Skills: Introduction q What does a professional research like a Ph. D mean? Ø Brought to you by Matt Might's page at http: //matt. might. net/articles/phd-school-in-pictures/

Research Methodology and Skills: Introduction

Research Methodology and Skills: Introduction

Research Methodology and Skills: Introduction

Research Methodology and Skills: Introduction

Research Methodology and Skills: Introduction

Research Methodology and Skills: Introduction

Research Methodology and Skills: Introduction

Research Methodology and Skills: Introduction

Research Methodology and Skills: Introduction

Research Methodology and Skills: Introduction

Research Methodology and Skills: Introduction

Research Methodology and Skills: Introduction Keep Pushing!

Outline Introduction q Research Methodology and Skills q Ø Introductory Concepts Ø Problem Definition Ø Literature Survey (have been discussed earlier) Ø Research Conduct (Simulation, Experiment) Ø Data Management Ø Reporting Digital Systems Research Presentation Mohammad Sharifkhani © 16

Introduction � In most research works you have: � A New: ○ Method, Model, Tool, Device, Architecture, Algorithm, Circuit, etc. � which is superior to the previous works in terms of ○ Criteria � You need to prove it through: � Analysis � Simulation � Experiment � Most of the times, you need two consistent evidences Digital Systems Research Presentation Mohammad Sharifkhani © 17

Outline Introduction q Research Methodology and Skills q Ø Introductory Concepts Ø Problem Definition Ø Literature Survey (have been discussed earlier) Ø Research Conduct (Simulation, Experiment) Ø Data Management Ø Reporting Digital Systems Research Presentation Mohammad Sharifkhani © 18

Research Methodology and Skills: Introduction

Research Methodology: What is Research? q Discovery of new things that have been independently verified by other professionals. q Research comprises Defining and redefining problems, Formulating hypothesis or suggested solutions; Collecting, organizing and evaluating data; Making deductions and reaching conclusions; and At last carefully testing the conclusions to determine whether they fit the formulating hypothesis. Ø Reporting the results. Ø Ø Ø

Research Methodology: What is Research?

Research Methodology: What is Research? Prior knowledge An idea The Outcome is Recognised Submit as a Major Contribution Report, Thesis, to the Field Journal Discovery Independent verification: literature, experiment, literature, numerical model, analytical model, etc or Conference Paper Assessors

Research Methodology: What is Research? q Research topic should address one or few major unanswered questions in the field or may address the existing challenge in the field. q If the answer is clear, then the selected topic is not a research topic!

Research Methodology: What is Research? q The purpose of the research should be clearly defined. q The research procedure used should be described in sufficient detail to permit another researcher to repeat the research for further advancement, keeping the continuity of what has already been attained. q The procedural design of the research should be carefully planned to yield results that are as objective as possible.

Research Methodology: What is Research? q Research Prerequisites: Ø Enough background on the field Ø Literature survey Ø Experiment/Simulations q Research outputs/products: Ø Papers Ø Books Ø Presentations Ø Real Stuff!

Research Methodology: Types of Research? q Fundamental or Practical q Analytical or Qualitative q Conceptual or Practical q Industrial

Good Researcher q Has four Ps: PATIENCE, PERSISTENCE, PERSEVERANCE, PERSPIRATION q Takes the critics and comments very serious q Attend various talks and defense sessions to get diversity q Try to give feedback and criticize other works q Does not lose his motivation when see a small failure (like a paper rejection or a bad criticism)

Good Researcher q Hamming: Take two people with the same potential and capability, if one of them works 10% more, he would produce twice as much! Ø The more you know, the more you learn Ø The more you learn, the more you can Ø The more you can, the more opportunities you get

Good Researcher q Comparison of your results with previous work always help q Take note from all papers you have read q Take a note of ideas that can be pursued q Take a note of potential solutions to the problem of interest q Write down your potential solution. On paper, gaps and flaws are easily detected q Start writing your expected paper in parallel to your research to have a big picture of where you are going q Review your notes quite often

Perspiration vs. Inspiration (Intelligence) q Perspiration ONLY does not work! q Inspiration/intelligence ONLY does not work! q Perspiration and intelligence together makes magic! Ø Pasteur: "Luck favors the prepared mind. " Ø Newton: "If others would think as hard as I did, then they would get similar results. " Ø Edison: "Genius is 99% perspiration and 1% inspiration. " q When you concentrate on a topic and sleep on a problem your inspiration would help you a lot.

Perspiration vs. Inspiration (Intelligence) q Your research topic should be with you everywhere, i. e. , in bus, taxi, when watching a movie, walking, eating, … q Sometimes when it is running on the background, potential solutions come to foreground!

Effective Research: Time Management q Time management is the key. q You have limited number of weeks till your defense count them q Research takes much more time than you normally expect q Tempting activities that are irrelevant to your research (check mail, news, …) should be delayed or be done in a pre-planned and limited period of time. q Give limited time to yourself to surf the web for leisure. Break your final goal into small sub-goals feasible to be done in a week or month to enhance your efficiency

The Research Community q All use the same scientific method. q All follow the same ethical principles. q All use the same language and terms. q All provide information to the world-wide community reported in a full and open manner. q All acknowledge the previous work of others.

Research Failure q What happens if I do not reach my targeted goals in the research? Ø This is intrinsic to the nature of the research Ø A comprehensive justification of a failure in a research trend is a valuable research itself (can also be published) Ø This would prevent others to take the same direction Ø It should be grounded on rational and well-established analysis and verifications

Research Failure q Failure is part of the game q You should be careful not to define a high-risk topic for your twoyear master’s topic q You should always have an alternative or fall-back position q It is important to recognize the reasons behind this failure to first resolve it and second do not repeat it in future

Research Failure: Similar Work q Paper search continuously to prevent missing a similar work q After a while you may find a similar work already published Ø Don’t worry, seat back and relax! Ø In most cases that work is not 100% the same so find the differences Ø Read it once more with the objective of finding the disparities Ø Give it to another expert like your supervisor to comment Ø You can even contact the author Ø Even if it is 100% the same you can take different approach to continue it

Research Failure: Active vs. Passive Learning

Outline Introduction q Research Methodology and Skills q Ø Introductory Concepts Ø Problem Definition Ø Literature Survey (have been discussed earlier) Ø Research Conduct (Simulation, Experiment) Ø Data Management Ø Reporting Digital Systems Research Presentation Mohammad Sharifkhani © 38

Problem Definition Problem definition is a key step at the beginning q Proper topic definition is part of the solution! q Good Definition requires: q Ø Sufficient background on the field Ø Be aware of the current status of the filed (literature survey) Ø Recognition of the existing challenges Ø Insight about the next step

Outline Introduction q Research Methodology and Skills q Ø Introductory Concepts Ø Problem Definition Ø Literature Survey (have been discussed earlier) Ø Research Conduct (Analysis, Simulation, Experiment) Ø Data Management Ø Reporting Digital Systems Research Presentation Mohammad Sharifkhani © 40

Research Conduct q In most research works you have: Ø A New Method, Model, Tool, Device, Architecture, Algorithm, Circuit, …, which is superior to the previous works q You need to prove it through: Ø Analysis Ø Simulation Ø Experiment q Most of the times, you need two consistent evidences 41

Research Conduct You need enough simulations and analysis to first well define the topic and then conduct the simulation q Sometime your great idea is not appreciated by the reviewers or your supervisor because there is a gap between the idea and the results q This gap includes intermediate simulations results and/or justifications showing the correctness and accuracy of your final results q You should justify your approach not just the final result q 42

Analysis � Be clear about your analytical procedure � Define what is your ultimate objective �Is it a parameter? �Is it a design guideline? �Is it a criteria? A>B �A closed-form equation is always favorable �People like math! Digital Systems Research Presentation Mohammad Sharifkhani © 43

Analysis � Be clear about your presumptions �What are the factors that you are going to ignore? �What are the dominant factors? �Why would you think you can make such assumptions? ○ Simulations ○ Reports Digital Systems Research Presentation Mohammad Sharifkhani © 44

Analysis � Be clear about the variable names, indexes, etc. �Try to make use of common variable names in the literature �Avoid using unnecessary variables when you do not refer to it �Be consistent throughout your entire research work Digital Systems Research Presentation Mohammad Sharifkhani © 45

Analysis � Come up with real life examples in your analysis �Helps the reader to grasp what you mean � Flow of analysis might be useful �Use appendix if necessary �Not every equation is useful all the time ○ Too sophisticated equations can not be used in hand calculations and numerical analysis can provide more accurate results � Use analysis to find the trends Digital Systems Research Presentation Mohammad Sharifkhani © 46

Simulation � Be clear about your objective for the simulation �It influences how fast, accurate you would simulate your idea � Choose the right model for your case �Be clear about how it is described ○ How detailed, accurate it is Digital Systems Research Presentation Mohammad Sharifkhani © 47

Simulation � Simulation Tool �Is it the right tool for the abstraction layer? ○ Time vs. Accuracy �How does it work? ○ You need to know what you are doing! �Is it an standard tool? (License? ) ○ Should we report the tool? �Right Device Models for the Tool is available? Digital Systems Research Presentation Mohammad Sharifkhani © 48

Simulation � Simulation Test Bench �Type of test-bench/benchmarks �Standard test-benches/test vectors � Simulation Test Setting �Temperature, corner cases, etc. � Well-know simulation methods �Time domain, Frequency domain, Monte carlo, etc. Digital Systems Research Presentation Mohammad Sharifkhani © 49

Simulation � Sim. Result Analysis �Realtime analysis �Post sim (off-line) analysis �Calibrate your computational analysis ○ How the criteria, FOM, etc. is calculated in your post sim analysis Digital Systems Research Presentation Mohammad Sharifkhani © 50

Experiment � Make up your mind on what you are going to measure � Choose your measurement setup before hand �Ports, Voltage Levels, Equipment, Connections, … �Example: current measurement for power analysis, etc. Digital Systems Research Presentation Mohammad Sharifkhani © 51

Experiment Digital Systems Research Presentation Mohammad Sharifkhani © 52

Experiment � Stimulating equipments � Data generator, Vector signal generator, pulse generator, etc. � Observation equipments � Logic analyzer, Vector signal analyzer, oscilloscope, meters, … � Is there standard equipment for a particular testbench/standard? � Refer to other works � Repeatability � Yield, etc. � Record your results, etc. � Analyze and justify your results Digital Systems Research Presentation Mohammad Sharifkhani © 53

Research Conduct q Keep in mind the final goal of the research Ø Low-Power Ø High-Throughput Ø Low-Area Ø Low-Latency Ø A mix of these… q Make sure to survey related papers and think of the solution towards this goal. 54

Research Conduct: Gantt Chart You need to have a Gantt chart q A Gantt chart is a type of bar chart, developed by Henry Gantt in the 1910 s, that illustrates: q Ø A project schedule and timeline Ø The start and finish dates of the terminal elements Ø Work breakdown structure of the project Ø All intermediate stages and tasks Ø The dependency between tasks Ø The people assigned to each task Ø The required HW/SW needed for each task Normally done by Microsoft Project 55

Research Conduct: Gantt Chart Sample 56

Outline Introduction q Research Methodology and Skills q Ø Introductory Concepts Ø Problem Definition Ø Literature Survey (have been discussed earlier) Ø Research Conduct (Simulation, Experiment) Ø Data Management Ø Reporting Digital Systems Research Presentation Mohammad Sharifkhani © 57

Outline Introduction q Research Methodology and Skills q Ø Introductory Concepts Ø Problem Definition Ø Literature Survey (have been discussed earlier) Ø Research Conduct (Simulation, Experiment) Ø Data Management Digital Systems Research Presentation Mohammad Sharifkhani © 58

Data Management The output of each research include some of the files and technical reports. q The files created in the intermediate stages of the research process should be efficiently archived and managed. q Efficient data management would enhance the research efficiently significantly. q Failure to do so may result in time/money wastes q Ø May even result in failure in the whole research. 59

Data Management Create a logical folder structure for your files q The folder name should be self-descriptive representing q Ø the content Ø the date it created Make sure to archive the files in related folders q If a project have several files say the code, its testbench and the output file, archive them in one folder q Have a readme file in each folder describing the folder files q Ø If a test has passed mention it in the readme file for future reference q Archiving will help the mutual comparison of files over time 60

Digital Systems Research Presentation Thank you! Digital Systems Research Presentation Mohammad Sharifkhani © 61