Designing Efficient and Accurate Behavior-Aware Mobile Systems

189 Pages·2015·2.63 MB·English

by  Abhinav Parate

University of Massachusetts - Amherst
ScholarWorks@UMass Amherst
Doctoral Dissertations 2014-current Dissertations and Theses
Summer 2014
Designing Efficient and Accurate Behavior-Aware
Mobile Systems
Abhinav Parate
University of Massachusetts - Amherst, [email protected]
Follow this and additional works at:http://scholarworks.umass.edu/dissertations_2
Recommended Citation
Parate, Abhinav, "Designing Efficient and Accurate Behavior-Aware Mobile Systems" (2014).Doctoral Dissertations 2014-current.Paper
224.
This Open Access Dissertation is brought to you for free and open access by the Dissertations and Theses at ScholarWorks@UMass Amherst. It has
been accepted for inclusion in Doctoral Dissertations 2014-current by an authorized administrator of ScholarWorks@UMass Amherst. For more
information, please [email protected].
DESIGNING EFFICIENT AND ACCURATE BEHAVIOR-AWARE
MOBILE SYSTEMS
ADissertationPresented
by
ABHINAVPARATE
SubmittedtotheGraduateSchoolofthe
UniversityofMassachusettsAmherstinpartialfulfillment
oftherequirementsforthedegreeof
DOCTOROFPHILOSOPHY
September2014
SchoolofComputerScience
(cid:13)c CopyrightbyAbhinavParate2014
AllRightsReserved
DESIGNING EFFICIENT AND ACCURATE BEHAVIOR-AWARE
MOBILE SYSTEMS
ADissertationPresented
by
ABHINAVPARATE
Approvedastostyleandcontentby:
DeepakGanesan,Chair
PrashantShenoy,Member
BenjaminM.Marlin,Member
EvangelosKalogerakis,Member
ChristopherSalthouse,Member
LoriA.Clarke,Chair
SchoolofComputerScience
||Om||
Idedicatethisthesistomyparents,
RekhaandShrawanParate,
fortheirlove,endlesssupportandencouragement.
Andtomywife,
Sujata,
forherlove,sacrifices,patienceandsupport
thatmadethisthesispossible.
ACKNOWLEDGEMENTS
This journey towards Ph.D. has been a long one and this would not have been possible
without the support of all those who accompanied me at various stages. I would like to
express my immense gratitude to all those who extended their support when I faced the
cold winters, and to all those whose company brought joy, cheer and the beautiful springs
inthisjourney.
First and foremost, I would like to thank my advisor, Deepak Ganesan. His perse-
verance, curiosity and willingness to dive into the unknown and exciting fields are some
of the traits that helped me explore research problems in the territory which neither of us
had experience with. His advices, guidance and support encouraged me to evolve into an
ambitious researcher from a struggling graduate student. He taught me to embrace the
uncertainty in research with optimism. I thank him for teaching me the valuable lessons
aboutthelifecycleofperformingresearch. Thankstomanyhoursofdiscussionswithhim,
Ilearnthowtoidentifyaninterestingresearchproblem,crystallizethenovelideasandfor-
mulaterobustexperiments. Hisaimforperfectionhelpedmeimprovemyresearchprocess
andpolishmyskill-set.
Iwouldliketoexpressmygratitudetotheesteemedmembersofmythesiscommittee-
PrashantShenoy,BenjaminMarlin,ChristopherSalthouseandEvangelosKalogerakis. My
special thanks to Prashant who helped me at various stages of this journey and without his
support and trust in my capabilities, I would not have reached here. I would like to thank
BenjaminMarlinwithwhomIcollaboratedonvariousprojects. Hiskeeninsightshadbeen
extremely helpful in addressing various challenges in my research. His advices helped me
broadenmyvisionasIlearnedtoincorporatemachinelearninginmyresearchinapractical
manner. I thank Christopher Salthouse for his valuable feedback that has been extremely
v
helpful in improving this thesis. His research work is truly inspiring and motivating. I
thank Evangelos Kalogerakiswhose advice has been very helpfulin developing a working
gesture recognition system for phones. Thanks to him, the research is now on its way to
explorenewterritories.
IthankfellowgraduatestudentsMeng-ChiehChiuandChanielChadowitzwhosecon-
tributions in my research projects were of immense help. I would like to thank David Chu
(Microsoft Research) and Matthias Bo¨hmer (DFKI Germany) with whom I collaborated
on a research project that resulted in a truly successful mobile application in the Android
market. Matthias’ advices and ideas on HCI continue to be of help when I design new
Android applications. I thank the team from Yale School of Medicine- Gustavo Angarita,
EdwardGaiser,RobertMallison. Ihadagreatlearningexperiencewhileworkingwiththis
medical team when we conducted user studies with human subjects; the experience that I
wasabletouselaterinmyownresearch.
Iwouldliketothankmylab-matesandcolleaguesatUMasswhowereonthesameboat
as me and it was great to have their company. Jeremy Gummeson, Pengyu Zhang, Rahul
Singh, Tingxin Yan, Upendra Sharma, Akshat Kumar, Aditya Mishra, Anand Seetharam,
Annamalai Natarajan, Moaj Musthag, Vijay Pasikanti, Siddharth Gupta - brainstorming,
havinglunchestogether,discussingandlaughingwithyouallmadethisjourneyaloteasier.
During these years of Ph.D. journey, I made a few wonderful friends. Himanshu
Agrawal and Pranshu Sharma - thanks for all the fun-filled cooking sessions and the dis-
astrous skiing sessions we had. Yash Sanghai and Anil Jain, thanks for all the wonderful
memoriesinPuftonVillageandforthecampingexperiencesinthewild. VarunSrinivasan,
an unexpected friend who happens to be full of excitement, and has always been there to
helpinanysituation.
I would also like to thank the staff of School of Computer Science, especially Leeanne
Leclercforherhelpintrackinggraduateschoolrequirements,andKarrenSaccoforallher
vi
administrative help in handling non-research problems that I faced while conducting user
studiesrequiredformyresearch.
Icannotexpressenoughgratitudetowardsmyfamily. Myparents,RekhaandShrawan
Parate, and my sister Sweety Parate, thanks for supporting me and my journey in a far
awaycountry. IthankmyparentsfortheimmeasurableloveandencouragementIreceived
from them during the tough times. The sweet phone conversations with my sister are
among the sweetest memories I made here. My acknowledgements can never be complete
without expressing my heartfelt appreciation for the person who completes me, my better
half, Sujata. Sujata accompaniedme with herlove and unlimitedpatience when Iwas pre-
occupiedwithmypaperdeadlines. Itwashersupportthathelpedmedefendmydissertation
proposal less than 2 weeks before our wedding ceremony. Without her support, I would
neverhavebeenabletoaccomplishthisdissertation.
vii
ABSTRACT
DESIGNING EFFICIENT AND ACCURATE BEHAVIOR-AWARE
MOBILE SYSTEMS
SEPTEMBER2014
ABHINAVPARATE
B.Tech.,INDIANINSTITUTEOFTECHNOLOGYKANPUR
M.S.,UNIVERSITYOFMASSACHUSETTSAMHERST
Ph.D.,UNIVERSITYOFMASSACHUSETTSAMHERST
Directedby: ProfessorDeepakGanesan
Theproliferationofsensorsonsmartphones,tabletsandwearableshasledtoaplethora
of behavior classification algorithms designed to sense various aspects of individual user’s
behavior such as daily habits, activity, physiology, mobility, sleep, emotional and social
contexts. This ability to sense and understand behaviors of mobile users will drive the
next generation of mobile applications providing services based on the users’ behavioral
patterns. In this thesis, we investigate ways in which we can enhance and utilize the un-
derstanding of user behaviors in such applications. In particular, we focus on identifying
the key challenges in the following three aspects of behavior-aware applications: detec-
tion, understanding, and prediction of user behaviors; and present systems and techniques
developedtoaddressthesechallenges.
In the first part of this thesis, we demonstrate the utility of wristbands equipped with
inertial sensors in real-time detection of health-related behaviors such as smoking and eat-
viii
ing. Our approach detects these behaviors in a passive manner without any explicit user
interaction and does not require use of any cumbersome device. Our results show that
we can detect smoking with 95% accuracy, 91% precision and 81% recall in the natural
environment.
In the second part of the thesis, we design a context-query engine for sensing multiple
user contexts continuously, accurately and efficiently on mobile devices; the key necessity
for understanding and analyzing behaviors. Our context-query engine performs informa-
tionfusionofcontextsforanindividualusertoenableoptimizationslikei)energy-efficient
sensing, and ii) accurate context inference while minimizing the privacy risks associated
withtheleakageofsensitivecontextstothirdpartyapplications. Weshowthatourcontext-
query engine can improve accuracy of a context classifier by up to 42% and reduce the
numberofclassifiersrequiredtoobservetheuserstateby33%.
Finally, in the third part of the thesis, we demonstrate the utility of predicting app
usage behavior, in improving the freshness of mobile apps such as Facebook that present
users with the latest content fetched from remote servers. We present an app prediction
algorithmthatutilizesusercontextstopredicttheappauserislikelytouseandpre-fetches
the data over the network for the predicted app. We show that our proposed algorithm
delivers application content to the user that is on an average fresh within 3 minutes. An
implementation of our algorithm is available as a widget in the Google Play Store that
shows shortcuts for the predicted apps; and has now been downloaded and installed on
morethan50,000devices.
ix