How to Deliver a Successful Thesis in Cyber-Physical Systems

agreeablesocietyΤεχνίτη Νοημοσύνη και Ρομποτική

29 Οκτ 2013 (πριν από 3 χρόνια και 9 μήνες)

89 εμφανίσεις

How to Deliver a Successful
Thesis in Cyber
-
Physical Systems
(CPS) research

Dr.
Driss

Benhaddou


Fulbright Visiting Professor, Al
Akhawayn

University

Associate Professor, University
of
Houston, TX

dbenhaddou@uh.edu




June 27
-
28, 2013

Start with the end in mind


Imagine yourself defending your thesis in front
of a committee delivering an outstanding
Ph.D. research project!

Outline


Research cycle


Project and Research Management


Keys to success


Cyber
-
Physical Systems


Challenges and Opportunities


Questions and Answers


Project

Manager

Project

Worker

From Passive Learner to Active Researcher


Project Manager
: Question


Plan
-

Communicate


Define scope based on project goal


Develop project plan


Track progress toward completion


Seek necessary resources, training, and tools


Adapt to deliver project on time and within constraints


Deliver project: report, oral presentation, poster presentation



Project Worker
: Collect
-

Analyze


Complete project tasks


Conduct research responsibly


Document work


Analyze data

An active researcher plays two roles

Thesis Scope


The thesis objectives and what it takes, in
term of knowledge and skills, to accomplish
those objectives.


Consider
novelty and uniqueness, and
impact/benefits
.

Project Management: The triple constraints


Scope

(Specifications and objectives)

Scope: What the project is
trying to accomplish?


Time: How long should it take
to accomplish the project ?
What is the schedule?


Cost: What will it cost to
complete the project.

Setting SMART Objectives/Goals


Establish and refine objectives based on
Needs

and/or
Hypothesis


Establish SMART objectives/goals


S
pecific


M
easurable


A
greed


R
ealistic


T
ime
-
specific



A
Thesis

is an educated guess? You need to approve it or
disapprove it.

Steps to success


The search


Data collection and analysis


Draw conclusions and concept development


Step by step systematic process


Knowledge does not come from void!


Task


Task Components


WHAT


Objective: what is to be achieved by performing this task.


Deliverable: the results of the task; includes hardware, software,
documents, recommendations, etc.


WHEN


Start date and estimated duration


WHO


Responsibility


RELATION to other tasks


Predecessors


Successors

Gantt Chart


Display project scheduling information by listing
project activities and their corresponding start and
finish dates in a calendar format.



The activities in the Gantt chart should coincide with
the activities in the WBS

Using a Gantt Chart

http://www.vertex42.com/

Keys to Success


Disciplined people


Leadership


Work hard and follow the schedule


Disciplined thoughts


Follow the research steps and maintain unwavering faith that you can
and have the discipline to confront the most brutal facts.


Know your technical strength and capitalize on it. Learn if necessary!


Disciplined Action


When you have disciplined people you don’t need hierarchy. When
you have disciplined thought, you don’t need bureaucracy. When you
have disciplined action, you don’t need excessive controls.


Utilized the tools to help develop/test your ideas. The tools will not do
them for you!

Disciplined People

Disciplined Thought

Disciplined Action

Build Up

F L Y W H E E L

The Flywheel and the Doom Loop

Steps Forward, Consistent
with Hedgehog Concept

Accumulation of Visible
Results

People Line Up,

Energized by Results

Flywheel Builds
Momentum

Weekly research
-
team debates


Debate


Disagree


Pound on tables


Raise voices with respect of course!


Pause and reflect


Debate some more


Pause and think,


Discuss


Resolve


Question


Debate yet again


The research life goes on…

Be aware


Simple solutions are powerful


Look for the “dogs that did not bark”


Data, analyze, debate, conclusions (be aware of
second bullet), probably build a framework.


Loop back an forth between data, analysis and ideas?


Develop the ability to take a look at organized data
and extract order and conclusion
-
> data
representation is very important. It is why you need
to take a look at data from different point of views…

Perseverance


Good thesis never happen in one fell swoop.


There was no one killer innovation, no solitary
lucky break, no miracle moment.


Rather, the process resembled relentlessly
pushing a giant heavy flywheel building
momentum until a point of breakthrough and
beyond.


Leadership

Good is the enemy of great!


Thing about how to do a great thesis not just a good
thesis.
However avoid the
curse of perfection
.


Welcome to the intellectual adventure


Question and challenge what you learn


“the best students are those who never quite believe
their professors” Jim Collins’ professor!


“one ought not to reject the data merely because
one does not like what the data implies” Jim Collins’
professor!

A Culture of Discipline


Freedom and Responsibility within a
framework


A Culture of Discipline


Adherence to the hedgehog Concept

Disciplined Action

Meeting


Always bring your lab book


Take note in the meeting if an action item is
needed


Follow up on issues before the meeting

Research phases

Define the problem
-
Research question

Solve the problem
-
plan and implement

Data Collection


Analytical performance evaluation


Numerical simulation


Practical implementation

Analysis
-
Evaluate your solution

Communicate

Define research problem

The most important and tedious part

It requires:


-

Literature research


-

Critical reading


-

Discussion with advisor


-

Re
-
reading


-
Define the research topic

Solve the problem

A problem well defined is a problem solved

Think about a solution

Use simulation tools

Use analytical tools

Implement the solution

Evaluate and Analyze your solution

Develop the model and generate results.

Analyze the results

Modify the model for possible mistakes

Finalize the results



Communicate

How long it will take you to write a thesis
document?


How long it will take to write an conference or
journal article?



Research Sample Topic: Social and Wireless Sensor
Network Driven Intelligent Building in Smart Grid


What is a WSN


Each node as an embedded system


In
-
network processing possible


Manage per node resource


Programmable!


Generally a multi
-
hop wireless
network


Data gathered at sensors are relayed
through radio towards the sink


Manage network resource


The

Internet

Cyber Physical System








*Adapted From Dr. Yuan presentation

31

What are Cyber
-
Physical Systems?


CPS deeply integrate
computation, communication,
and control

into
physical

systems


CPS exploit pervasive, networked computation,
sensing, and control, i.e., “Internet of [
controlled]
things”


“CPS will transform how we interact with the physical
world just like the Internet transformed how we
interact with one another.” *


* CPS Summit Website:
http://varma.ece.cmu.edu/summit/index.html

CPS Challenges


New
Abstraction Challenges


Network Challenges


In Network Processing


Layered Architecture for CPS


Information centric protocol stack


Computing Challenge


Middleware Paradigm


Programming Language Design Challenge


Privacy and Security Challenge


Integration of Analysis Theories and Tools


*
Adapted From Dr. Yuan
presentation

32

Network Challenges


Present network architecture


Goal : communication


Connecting people (nodes)


Protocols geared primarily for point
-
to
-
point
communication


Data consumption applications: email, VoIP, Web,
P2P,..




*
Adapted From Dr. Yuan
presentation

Future Network


Goal : (autonomous) information distillation


Reducing data from sensing inputs to actionable
information


Protocols geared for data filtering, aggregation, mining


Data reduction applications: emergency
detection/response, statistical community analysis


Goal : action
-
capable network


Example: Detect the current gas (engine oil) of the driving
car, and find/reserve a most convenient gas
station/mechanic

Information
-
Centric Protocol Stack


Future networks will focus around elevating information
abstractions along communication pathways (from
sensors to consumers)


Scalable support will be needed for network storage and
in
-
network processing











*
Adapted From Dr. Yuan
presentati on

Network

raw data

actionable

application
-

specific

information

Future CPS Middleware Paradigm


Abstracts away distributed, domain
-
specific process of object
identification and its state tracking


Represents the external environment conveniently for the
application developer


Applications



Network / OS
Hardware



Domain
-
Specific
Knowledge &
Skill



CPS Middleware



Distributed
Protocols


CPS and Smart Grid: Home Scenario

Research Question


How human behavior can be modeled and
implemented in the system?


What “smart” algorithms can be implemented
so that the building is autonomous and adapt
to the human behavior?


What methods should we use to verify the
functionality of the system?

Research Outcome


Autonomous building that can act on its own
with minimum intervention of the manager.

Acknowledgement


Fulbright program for funding this Project


Al
Akhawayn

University, School of Science and
Engineering for hosting me


University of Houston for paying half of the bill
and for Seed Funds through GEAR program


Faculty and students

I would like to thank

Thank You!