Unobtrusive, Pervasive, and Cost-Effective Communications with Mobile Devices Dissertation Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University By Adam C. Champion, B.S., M.S. Graduate Program in Computer Science and Engineering The Ohio State University 2017 Dissertation Committee: Professor Dong Xuan, Advisor Professor Feng Qin Professor Ten H. Lai
135
Embed
Unobtrusive, Pervasive, and Cost-Effective Communications ...champion.17/Thesis_ChampionAda… · Unobtrusive,Pervasive,andCost-EffectiveCommunications withMobileDevices Dissertation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Unobtrusive, Pervasive, and Cost-Effective Communications
with Mobile Devices
Dissertation
Presented in Partial Fulfillment of the Requirements for the DegreeDoctor of Philosophy in the Graduate School of The Ohio State
University
By
Adam C. Champion, B.S., M.S.
Graduate Program in Computer Science and Engineering
The Ohio State University
2017
Dissertation Committee:
Professor Dong Xuan, Advisor
Professor Feng Qin
Professor Ten H. Lai
cCopyrightby
AdamC.Champion
2017
Abstract
Mobile devices such as smartphones are ubiquitous in society. According to Cisco
Systems, there were eight billion mobile devices worldwide in 2016 [1], which surpassed
the human population [2]. Mobile devices and wireless network infrastructure form an
“electronic world” of signals that is part of daily life. However, navigating this world
with users’ devices is challenging. The volume of signals may confuse users, wireless
communications often require manual connection establishment, and latency may be
large (such as Bluetooth device discovery). Pervasive mobile device communications
offer large-scale measurement opportunities when many devices connect to wireless
networks. For example, base stations to which devices connect can indicate human
mobility patterns. But existing work only studies coarse-grained cellular call records
and datasets used in wireless local area network (WLAN) studies typically consist of
laptops. Besides, existing vehicular communications technologies tend to be expensive
and available for new vehicles only.
This dissertation studies three topics that arise in the electronic world: unob-
trusive communications among mobile devices without manual connection establish-
ment; pervasive mobile device communications measurement; and cost-effective ve-
hicular communication among mobile devices. First, we design Enclave that enables
unobtrusive communication among mobile devices without network connections or
configurations. Unobtrusive communication is efficient wireless communication that
ii
does not require user interruption such as manual device connection or network con-
figuration. Enclave consists of a delegate mobile device (such as an unused phone)
that interposes between a user’s “master” device (such as her smartphone) and the
electronic world. Enclave communicates between the master and delegate devices us-
ing wireless name communication and picture communication. Second, we study a
new dataset with over 41 million anonymized (dis)association logs with WLAN ac-
cess points (APs) at The Ohio State University (via the osuwireless network) over
139 days from January to May 2015. The dataset includes more than 5,000 university
students with their birthdays, genders, and majors, which are made available after
anonymization. Using mobility entropy as our metric, we find that entropy increases
with age (for 19–21-year-olds), students’ entropic rates of change vary with majors,
and all students’ long-term entropy follows a bimodal distribution that has not been
previously reported. We design a mobile application that localizes users indoors for 73
campus buildings using WLAN site survey information. In general, our app achieves
room-level accuracy. Finally, we propose SquawkComm for cost-effective vehicular
communication using mobile devices and FM signals. SquawkComm encodes data as
audio, which is sent via inexpensive FM transmitters that plug into vehicles’ cigarette
lighters and received via vehicle stereos. SquawkComm uses a new physical-layer cod-
ing scheme and link-layer mechanisms for channel access. Our experimental evaluation
illustrates the promise of our work in the electronic world. We conclude with directions
for future work.
iii
This is dedicated to my family, my advisor, and my friends.
iv
Acknowledgments
Numerous people have provided me vital assistance during this long, challenging
“research journey.” Here, I express my gratitude to all of them.
First, I thank my advisor, Prof. Dong Xuan, for his insight and support during
my Ph.D. study. He taught me the importance of clear problem definition, solving
problems among multiple levels of abstraction, and writing with precision and con-
cision. His passion for impactful, high-quality work with practical utility guided my
research efforts and it will guide my future work.
Second, I thank several faculty members at The Ohio State University (OSU)
for providing me a strong education. Profs. Feng Qin and Srinivasan Parthasarathy
taught me a great deal about computer systems and data mining. Prof. Ness Shroff
illuminated queueing theory and stochastic processes in computer networks. Prof.
Yuan F. Zheng provided useful insights about computer vision, robotics, and machine
learning. I am grateful for Profs. Feng Qin, Mikhail Belkin, and Raghu Machiraju’s
incisive comments during my candidacy exam. I appreciate Profs. Feng Qin and Ten
H. Lai serving on my final dissertation committee.
Third, I thank the Department of Computer Science and Engineering at OSU
for my position as a full-time lecturer and the opportunity to “pay forward” to my
students.
v
Fourth, I thank my collaborators for their insight and discussions, including Prof.
Paul Y. Cao (from University of California, San Diego), Prof. Linghe Kong (from
Shanghai Jiao Tong University), Dr. Sihao Ding, Dr. Jin Teng, Dr. Xinfeng Li, Dr.
Boying Zhang, Dr. Zhezhe Chen, Dr. Boxuan Gu, Dr. Sriram Chellappan, Dr. Zhimin
Yang, Dr. Wenjun Gu, Dr. Xiaole Bai, Dr. Xun Wang, Gang Li, Ying Li, Qiang
Zhai, Guoxing Chen, Cheng Zhang, Steve Romig, and Austin Gilliam. Working with
them has taught me a great deal about computer science, electrical engineering, and
applied mathematics. I thank Prof. Cao for his insights on large-scale data analysis
and mobility entropy. Prof. Kong’s constructive criticism assisted my development of
SquawkComm and writing. Dr. Ding and Ying taught me about computer vision and
machine learning. Dr. Teng and Guoxing described electrical engineering and subtle
mathematical concepts. Drs. Zhezhe Chen and Boxuan Gu sharpened my knowledge
of high-performance computing and malicious code detection, respectively. Drs. Teng
and Li as well as Gang and Qiang assisted with experiments and writing. Dr. Yang
explained Bluetooth, Wi-Fi, and distributed systems design. Dr. Bai provided valuable
insights about mathematical rigor and research. Dr. Wang introduced me to graduate
study and machine learning. Cheng has illuminated deep learning, neural networks,
and artificial intelligence. Steve and Gang played vital roles in obtaining the WLAN
log dataset that Chapter 4 analyzes. Austin and Qiang assisted in SquawkComm’s
development and experiments. Dr. Xinfeng Li’s dissertation [3] was helpful in writing
mine. Dr. Theodore Pavlic’s dissertation [4] provided superb LATEX support; Dr. Chris
Monson’s LATEX Makefile [5] was very helpful. Gunter Lorenz provided the FMLIST
database [6] of FM radio stations for SquawkComm.
Finally, I thank my family for their continued love and support.
P.Y. Cao,∗ G. Li,∗ A.C. Champion,∗ D. Xuan, S. Romig, and W. Zhao, “On MobilityPredictability Via WLAN Logs,” in Proc. IEEE INFOCOM, 2017. (∗Co-primary firstauthor)
Q. Zhai, F. Yang, A.C. Champion, J. Zhu, D. Xuan, B. Chen, Y.F. Zheng, and W.Zhao, “S-Mirror: Mirroring Sensing Signals for Mobile Robots in Indoor Environ-ment,” in Proc. IEEE MSN, 2016.
F. Yang, Q. Zhai, G. Chen, A.C. Champion, J. Zhu, and D. Xuan, “Flash-Loc: Flash-ing Mobile Phones for Accurate Indoor Localization,” in Proc. IEEE INFOCOM,2016.
J. Hamm, A.C. Champion, G. Chen, D. Xuan, and M. Belkin, “Crowd-ML: A Privacy-Preserving Learning Framework for a Crowd of Smart Devices,” in Proc. IEEEICDCS, 2015.
vii
Z. Yu, F. Yang, J. Teng, A.C. Champion, and D. Xuan, “Local Face-View BarrierCoverage in Camera Sensor Networks,” in Proc. IEEE INFOCOM, 2015.
F. Yang, Y. Xuan, S. Ding, A.C. Champion, and Y.F. Zheng, “R-Focus: A RotatingPlatform for Human Detection and Verification Using Electronic and Visual Sensors,”in Proc. WASA, 2014. (Best Paper Award)
G. Li, J. Teng, F. Yang, A.C. Champion, D. Xuan, H. Luan, and Y.F. Zheng, “EV-Sounding: A Visual Assisted Electronic Channel Sounding System,” in Proc. IEEEINFOCOM, 2014.
B. Gu, X. Li, G. Li, A.C. Champion, Z. Chen, F. Qin, and D. Xuan, “D2Taint: Dif-ferentiated and Dynamic Information Flow Tracking on Smartphones for NumerousData Sources,” in Proc. IEEE INFOCOM, 2013.
Z. Yuan, W. Li, A.C. Champion, and W. Zhao, “An Efficient Hybrid LocalizationScheme for Heterogeneous Wireless Networks,” in Proc. IEEE GLOBECOM, 2012.
B. Gu, W. Zhang, X. Bai, A.C. Champion, F. Qin, and D. Xuan, “JSGuard: ShellcodeDetection in JavaScript,” in Proc. SecureComm, 2012.
A.C. Champion, X. Li, Q. Zhai, J. Teng, and D. Xuan, “Enclave: Promoting Unob-trusive and Secure Mobile Communications with a Ubiquitous Electronic World,” inProc. WASA, 2012. (Best Paper Runner-Up Award)
X. Li, J. Teng, B. Zhang, A.C. Champion, and D. Xuan, “TurfCast: A Service forControlling Information Dissemination in Wireless Networks,” in Proc. IEEE INFO-COM, 2012.
A.C. Champion, B. Zhang, J. Teng and Z. Yang, “D-Card: A Distributed MobilePhone Based System for Relaying Verified Friendships,” in Proc. IEEE CPNS, 2011.
Z. Yang, B. Zhang, J. Dai, A.C. Champion, D. Xuan, and D. Li, “E-SmallTalker:A Distributed Mobile System for Social Networking in Physical Proximity,” in Proc.IEEE ICDCS, 2010.
B. Gu, X. Bai, Z. Yang, A.C. Champion, and D. Xuan, “Malicious Shellcode Detectionwith Virtual Memory Snapshots,” in Proc. IEEE INFOCOM, 2010.
viii
X. Ni, Z. Yang, X. Bai, A.C. Champion, and D. Xuan, “DiffUser: Differentiated UserAccess Control on Smartphones,” in Proc. IEEE WSNS, 2009.
Z. Yang, A.C. Champion, B. Gu, X. Bai, and D. Xuan, “Improving the Security ofWireless Networks with Dummy Authentication,” in Proc. ACM WiSec, 2009.
X. Wang, W. Yu, A. Champion, X. Fu, and D. Xuan, “Detecting Worms via MiningDynamic Program Execution,” in Proc. SecureComm, 2007.
Research Publications – Journal
S. Shen, L. Huang, J. Liu, A.C. Champion, S. Yu, and Q. Cao, “Reliability Mea-surement for Clustered WSNs under Malware Propagation,” Sensors, vol. 16, no. 6,2016.
X. Li, J. Teng, B. Zhang, A.C. Champion, and D. Xuan, “TurfCast: A Service forControlling Information Dissemination in Wireless Networks,” IEEE Transactions onMobile Computing, Feb. 2014, pp. 250–262.
A.C. Champion, Z. Yang, B. Zhang, J. Dai, D. Xuan, and D. Li, “E-SmallTalker:A Distributed Mobile System for Social Networking in Physical Proximity,” IEEETransactions on Parallel and Distributed Systems, vol. 24, no. 8, Aug. 2013, pp. 1535–1545.
W. Yu, X. Wang, A. Champion, D. Xuan, and D. Lee, “On Detecting Active Wormswith Varying Scan Rate,” Journal of Computer Communications, Elsevier, vol. 34,no. 11, Mar. 2011, pp. 1269–1282.
Fields of Study
Major Field: Computer Science and Engineering
Studies in:
Computer Networking Prof. Dong XuanProf. David Lee
Distributed Systems Prof. Feng QinProf. Srinivasan Parthasarathy
4.3 Stay time estimation for inter-building AP connections in a person’strajectory. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.4 Histograms of long-term entropy Std for different age groups. . . . . . 59
4.5 Average entropies across age groups for each day of the week. . . . . 61
4.6 Histograms of 20-year-olds’ (sophomores’) Std for six different groupsof students based on their majors. . . . . . . . . . . . . . . . . . . . . 62
1: Main Procedure:2: BTname ← ‘’; Wi− FiSSID ← ctrl frame; make discoverable;3: Start Wi-Fi thread and Bluetooth thread;4:5: Wi-Fi Thread: {6: Set up ad hoc Wi-Fi network with SSID CF ; // ctrl frame7: while (user does not terminate) do8: Do active scan, store received Wi-Fi SSIDs (CFi) in CFSet;9: for each CFi in CFSet do10: for each addrj in CFi do11: if addrj /∈ BTAddrSet ∧ addrj /∈ RecvMsgSet then12: CandidateAddrSet ← CandidateAddrSet ∪ {addrj};13: end if14: end for15: end for16: Select MsgID ∈ RecvMsgSet appearing most frequently in CFSet with flag ‘-’;17: Update Wi-Fi SSID and Bluetooth name with MsgID;18: end while }19:20: Bluetooth Thread: {21: iteration ← 0; codedBTAddrs ← ∅;22: while (user does not terminate) do23: Inquire nearby Bluetooth addresses, store in BTAddrSet;24: if CandidateAddrSet = ∅ then25: for each of Nr randomly chosen addresses addri do26: PerformPaging(addri)27: end for28: else29: for each discovered address in BTAddrSet do30: PerformPaging(BTAddri)31: end for32: end if33: Construct BT frame, encode it, and split into chunks;34: for each chunk do35: Place chunk and encoding row in BT name;36: Page each addrj ∈ codedBTAddrs37: end for38: end while }39:40: Subprocedure PerformPaging(addri):41: Page addri, store message in message42: if paging succeeds and isV alidChunk() then43: Decode chunk (frame)44: codedBTAddrs ← codedBTAddrs ∪ {addri}45: Update Wi-Fi SSID accordingly;46: end if
31
upon paging failure, we mark the ID with a ‘-’. We maintain codedBTAddrs, the set
of currently discovered Bluetooth addresses whose corresponding names are encoded
chunks. When we discover a new such Bluetooth address, we add it to codedBTAddrs.
In each iteration of the Bluetooth thread, we generate a Bluetooth frame, encode it,
and split the encoded frame into chunks. We alternately publish each chunk and page
each address in codedBTAddr to retrieve newly published chunks.
Remarks NameCast can only change Bluetooth device names and Wi-Fi SSIDs
names programmatically on certain mobile devices such as Android and Windows
Mobile 6.x smartphones. At this time, iOS, BlackBerry OS, and Windows Phone
do not allow users to programmatically change Bluetooth device names and SSIDs.
However, these devices can still interact with devices running NameCast without
installing the system. Users of such devices can disseminate information by manually
changing their devices’ Bluetooth names or Wi-Fi SSIDs.
NameCast implicitly assumes that users are willing to share some information
electronically with people in physical proximity. This has potential privacy implica-
tions as attackers can fingerprint the master or enclave device via their Bluetooth
device names and Wi-Fi SSIDs [102]. We point out that NameCast can be used only
for communication between the enclave device and external data sources, which lim-
its such fingerprinting to the enclave device. In such cases, the master device would
like to communicate with the enclave device without leaving an electronic “trace.”
PicComm provides such functionality, which we discuss in Section 3.2.2. Additionally,
incorporating strong privacy and security controls with NameCast forms an impor-
tant part of our future work. We consider leveraging further coding techniques and
32
power control to selectively send the NameCast information to a designated group of
people.
3.2.2 PicComm: Enclave-to-Master Communication
Enclave can realize secure communication between the master and enclave devices
via picture communication (PicComm), a type of visual communication based on
taking photos without wireless communications. We detail it in this section.
Design Rationale
We need a secure communication channel between the master device and the
enclave device in order to transmit useful information to the master. This is hard
to achieve. Such a channel needs to be protected from eavesdropping attackers, but
the master device cannot totally trust the enclave device as it is exposed to the
outside world where malicious code or indecent content may reside. We argue that
wireless-connection-based communication may not be secure enough due to security
problems [103] or vulnerabilities [104]. Further, some users may be concerned about
disclosing their electronic identities (e.g., MAC addresses) using Wi-Fi or Bluetooth.
Main Ideas
We propose PicComm in Enclave to establish such secure communication. In Pic-
Comm, the master device takes pictures of the enclave device’s screen, which displays
textual messages, and recognizes their contents using optical character recognition
(OCR). In this way, information transmission from the enclave to the master is very
secure. Besides security, PicComm needs to achieve high throughput. However, the
screen size of a device is limited and OCR results are not 100% accurate. Intuitively, if
33
the message on the screen has larger font size and letterspacing, OCR performs better,
but the message volume on-screen decreases. Thus, we propose a dynamic resolution
adjustment mechanism for PicComm to find a good tradeoff between message volume
and resolution (i.e., the font size and letterspacing). In dynamic resolution adjust-
ment, the master device sends feedback to the enclave (an ACK or NAK) and hash
values indicating the OCR error degree. Based on this information, the enclave device
dynamically adjusts its resolution of the message on-screen. We consider two options
for the feedback channel: wireless name communication and sound. Wireless name
communication is part of NameCast; we use the Wi-Fi SSID to transmit the feed-
back message (up to 32 bytes). But the master device has to disclose its Wi-Fi MAC
address, which may concern some users. Thus, we propose the second option: sound.
With sound, PicComm can achieve greater security without disclosing any electronic
information. The drawback is low channel capacity (1 or 2 bits per feedback message).
In the following, we discuss the design of our dynamic resolution adjustment mech-
anism.
Dynamic Resolution Adjustment A naıve approach to resolution adjustment
is to uniformly change the whole screen. If the enclave device receives an NAK, the
resolution setting is enlarged by one (i.e., the font size increases by one point). If
the device receives an ACK, the resolution setting will be decreased by one point.
However, such an approach is inefficient. We should not enlarge resolution settings
for error-free parts of the screen, which yields less space for the total message. Thus,
we introduce the concept of a “block,” an isolated area in the screen with its own
resolution settings. We perform resolution adjustment for each block.
Majors Engineering Science Health Business Social Education UndecidedMean Std 1.22 1.29 1.18 1.30 1.33 1.33 1.09Mean Sti 2.19 2.24 2.13 2.25 2.24 2.25 2.17
Table 4.2: Long-term metrics for each academic major.
Figure 4.6: Histograms of 20-year-olds’ (sophomores’) Std for six different groups ofstudents based on their majors. The two contours represent the long-term entropy of19-year-olds (freshmen, solid red lines) and 21-year-olds (juniors, dashed green lines).
Sti values vary within 4% of each other while variations of Std exceed 22%. Unde-
cided majors have the smallest Std and Sti compared to all other majors.
Another interesting observation is that the rate of entropic increase from 19-year-
olds (freshmen) to 20-year-olds (sophomores) differs significantly among majors. For
example, this rate of increase for engineering majors is small compared with that of
business majors. Figure 4.6 shows histograms among six different majors for their
20-year-old group (sophomores). Unlike the rate of change from age 19 to age 20, the
rate of entropic change from age 20 to age 21 is similar among majors. Undecided
majors are not shown due to limited data after dividing into individual age categories.
62
Discussion The most dramatic increase occurs with business majors whose entropy
mode shifts from left to right when students’ ages increase from 19 to 20.
The entropy mode shift for engineering and science majors is much smaller. We
calculate the average stay time of students with respect to their ages (19 and 20) and
majors. We find that business majors’ average stay times decreased from 72.9minutes
(age 19 group) to 52.5minutes (age 20 group), a 28% drop. Engineering majors’
average stay times decreased from 68.8minutes to 59.6minutes, a 14% drop.
We calculate the time-independent entropy Sti for all students and observe no dual
mode. The observed shifts between 19- and 20-year-olds across majors are minimal.
For example, Figure 4.7 shows 20-year-olds’ (sophomores’) Sti for different majors.
Figure 4.7: Histograms of 20-year-olds’ (sophomores’) Sti for six different groups ofstudents based on their majors. The two contours represent the long-term entropy of19-year-olds (freshmen, red solid lines) and 21-year-olds (juniors, green dashed lines).
Short-Term Entropy For every day of the week, undecided majors have low en-
tropies compared with those of every other major. Table 4.3 shows Std for each day
Table 4.4: Short-term metrics for each day of a week
4.4 Extension: Indoor WLAN Localization with Site SurveyData
Our WLAN dataset includes site survey data for current AP deployments in 73
buildings on the OSU campus. These buildings are used mainly for academic and
administrative purposes. We leverage these data and design a mobile application
70
(app) for indoor localization on campus. Figure 4.12 shows these buildings shaded
in black. Table 4.5 describes the number of floor plans for each building where AP
deployment information is available.
Figure 4.12: OSU buildings with AP deployment information (shaded). Other build-ings include: 22 E. 16th Avenue, 53 W. 11th Avenue, Knight House, North Commons,Northwood-High Building, Raney Commons, Riverwatch Tower, and the Wexner Cen-ter for the Arts (not shown). We generate the map using Mapzen [116] with Open-StreetMap data [117].
71
Building NameNumberof Floors
209 W. 18th Ave 422 E. 16th Ave 453 W. 11th Ave 1Arps Hall 5Baker Systems Engineering 6
Blackwell Inn 9Bolz Hall 4Bricker Hall 4Caldwell Laboratory 5CBEC 8Celeste Laboratory 5Central Services Building 3Cockins Hall 5Converse Hall 3Denney Hall 6Derby Hall 5Dreese Laboratories 9Dulles Hall 4Enarson Classroom Building 5Evans Laboratory 5
Faculty Club 3Fisher Hall 9Fontana Laboratory 3French Field House 1Gerlach Hall 4Hagerty Hall 5
Hale Hall 5Hayes Hall 4Hitchcock Hall 5Hopkins Hall 5Hughes Hall 5Ice Rink 1Independence Hall 2
Jesse Owens RecreationCenter North
1
Journalism Building 4Knight House 1Knowlton Hall 6
Koffolt Laboratory 4
Building NameNumberof Floors
Kuhn Honors and ScholarsHouse
2
Lincoln Tower 5Lincoln Tower Park 1MacQuigg Laboratory 6Maintenance Building 3Mason Hall 5Math Building 5Math Tower 7
McPherson Laboratory 4Mendenhall Laboratory 5Mershon Auditorium 6Newman and WolfromLaboratory
5
North Commons 2Northwood-High Building 1Ohio Stadium 10
Orton Hall 3Page Hall 4Pfahl Hall 4Physics Research Building 6Raney Commons 2Riverwatch Tower 1
Schoenbaum Hall 4Science and EngineeringLibrary
5
Scott Laboratory 5Smith Laboratory 6St. John Arena 4Stillman Hall 5Student Academic ServicesBuilding
6
Sullivant Hall 3Tuttle Parking Garage 3
University Hall 5Watts Hall 5Weigel Hall 4Wexner Ctr. for the Arts 4Women’s Field House 1
Table 4.5: Number of floors for each building at OSU with WLAN site survey data.
Algorithm 5.1 SquawkCode’s FM Carrier Frequency Selection
1: function getCandFmFreq(φ, λ)2: allFmFreqs ← {87.5, . . . , 107.9};3: usedFmFreqs ← getUsedFmFreqs(φ, λ)4: usedFmFreqs ← sort(usedFmFreqs ∪ {87.5, 107.9})5: candFmFreqs ← findEquidistFreqs(usedFmFreqs)6: candFmFreqs ← sort(candFmFreqs)7: for fmFreq ∈ candFmFreqs do8: if not isInUse(fmFreq) then9: fmCarFreq ← fmFreq10: break11: end if12: end for13: return fmCarFreq14: end function15:16: function getUsedFmFreqs(φ, λ)17: usedFmFreqs ← ∅
18: fmFreqs ← unique transmitter frequencies in bounding box (φ±Δφ, λ±Δλ)19: for freq ∈ fmFreqs do20: usedFmFreqs ← usedFmFreqs ∪ {freq}21: end for22: return usedFmFreqs23: end function24:25: function findEquidistFreqs(usedFmFreqs)26: availFmFreqs ← ∅
27: for each pair of consecutive frequencies freqi, freqj ∈ usedFmFreqs do28: newFreq ← (freqi + freqj)/229: availFmFreqs ← availFmFreqs ∪ {newFreq}30: end for return availFmFreqs31: end function
that can identify vehicles [95, 96]. There are four frame types as shown in Table 5.2,
the first three of which are self-explanatory. (We discuss the fourth one shortly.) The
urgent bit indicates whether the frame has high priority. High-priority frames are sent
without the RTS/CTS handshake; low-priority frames are sent with the handshake.
If the CRC type bit is 1, we use an 8-bit CRC checksum; if the bit is 0, we use a 1-bit
CRC checksum. We use a 1-bit checksum by default; the 8-bit checksum is used for
applications that require higher data integrity.
Algorithm 5.2 Channel Access Protocol
1: Generate random vehicle ID V ID2: if vehicle needs to send data then3: Generate new random V ID4: if frame has urgent status then5: Wait random backoff time twait ∼ U(0, tmax)6: Send frame7: if collision then8: tmax ← min{thmax, 2 · tmax}9: else10: tmax ← max{thmin, tmax/2}11: end if12: else13: Start RTS timer14: Wait short random time; send RTS, V ID, timestamp15: Wait until RTS timer expires16: Sort sent, received vehicle IDs ListV IDs17: if ≥ 2 vehicles and V ID = max{ListV IDs} then18: Send CTS, timestamp19: end if20: Wait until time slot21: Send frame for time slot duration22: Wait until all time slots finish23: end if24: end if25: if vehicle receives frame with type 11 then26: Generate new random V ID27: end if
96
Channel Access
SquawkComm faces an important challenge: preventing several vehicle occupants
from sending data at the same time. Each vehicle occupant on the road that wants to
send data may not see other senders nearby and transmit, leading to collisions. Since
no sender can “hear” the data it is sending, collision avoidance is desirable. No one
should be starved from sending data due to other senders constantly transmitting.
However, urgent frames should be delivered before other frames.
We address these challenges via SquawkLink’s channel access protocol shown in
Algorithm 5.2. Although Algorithm 5.2 builds atop CSMA/CA, we tailor it for com-
munication among rapidly moving vehicles. Initially, each vehicle occupant’s mobile
device chooses its own VID at random following the approach discussed previously.
The remainder of the protocol only applies to senders. For privacy reasons, each sender
generates a new temporal VID before sending (line 3). If a frame has high priority, we
choose a random wait time twait ∼ U(0, tmax) and wait, where tmax is the longest pos-
sible wait time. Next, we send the frame. We double or halve tmax based on whether
there is a collision (subject to thresholds thmin and thmax). Otherwise, communica-
tion takes place in rounds (lines 13–22). At the start of a round, after waiting a short
random period of time, each sender transmits an RTS frame while listening for other
senders’ RTS frames. Each RTS frame includes the sender’s VID and a timestamp.
When the RTS period ends, each sender has a list of all other senders’ VIDs. Each
sender sorts its list and determines the position of its VID in the list. If a sender
determines its VID is the maximum among all VIDs and there are least two VIDs,
the sender sends a CTS frame. Time is then divided into slots; each sender waits for
its slot and transmits its data. The round ends once all senders finish transmitting.
have studied communication in a laboratory environment. In future work, we
will evaluate SquawkComm in real-world vehicular environments and analyze
the resulting latencies and bit error rates. We will study the impact of distance
among nearby vehicles in stationary and mobile settings.
105
Appendix A: Glossary of Technical Terms
This appendix defines technical terms used in this dissertation.2
Access Point (AP) APs are wireless base stations in WLANs to which devices
connect for network access. Each AP provides wireless service within 30–70-
meter ranges (depending on the environment and number of concurrent users).
Bluetooth Bluetooth is a communications standard for short-range wireless personal
area networks (∼10 meters). This dissertation discusses “classic” Bluetooth.
Dedicated Short Range Communications (DSRC) DSRC is a vehicular safety
technology in which vehicles rapidly send safety messages to each other (at 1–
10Hz) regarding their motion.
Electronic world The electronic world consists of all mobile devices and wireless
infrastructure that transmit and receive wireless signals.
Enclave device In our Enclave work, the enclave device is a delegate mobile device
that users employ to interact with the electronic world on their behalf. For
example, enclave devices may be “rental” ones for tourists traveling abroad or
older devices without data plans. Chapter 3 describes such devices further.
2Trademarks are the property of their respective owners.
106
Entropy Entropy quantifies the uncertainty in a person’s or a group’s mobility. Sec-
tion 4.3.1 describes entropies used in our WLAN measurement study.
Frequency Modulation (FM) FM is a standard technique that automotive trans-
mitters use to transmit audio from mobile devices that is played on vehicle
stereos. Textbooks such as [145] provide details.
Institute of Electrical and Electronic Engineers (IEEE) IEEE is a profes-
sional society of engineers that defines communications standards [146].
Master device In our Enclave work, the master device is a user’s primary device
(such as a smartphone) that interacts with the electronic world via the user’s
enclave device. Chapter 3 provides further details.
Multiple Access Control (MAC) address A MAC address is a 48-bit address
that identifies a network adapter at the link layer.
NameCast NameCast is our supporting technology for unobtrusive communication
among mobile devices without connection establishment. Section 3.2.1 explains
how it uses Bluetooth device names and Wi-Fi SSIDs for this purpose.
Optical Character Recognition (OCR) OCR is a computer vision technology
that parses textual content from imagery containing such content.
On-Off Keying (OOK) OOK is a modulation technique that encodes 1s and 0s as
the presence and absence, respectively, of sine waves sin(2πf) with correspond-
ing frequencies f . Section 5.2.3 explains how SquawkComm uses OOK.
Organizational Unit Identifier (OUI) An OUI comprises the three most signif-
icant bits of a network adapter’s MAC address that uniquely identifies the
107
adapter’s manufacturer. Manufacturers pay IEEE fees in order to use certain
OUIs. [147] lists all current OUIs in use.
Picture Communication (PicComm) PicComm is our supporting technology for
visual communication between the master and enclave devices. Section 3.2.2
explains how our Enclave work uses PicComm.
Service Set Identifier (SSID) An SSID is the name that a WLAN uses to adver-
tise network services (such as osuwireless). When mobile devices connect to
SSIDs, they can access the network via Wi-Fi.
SquawkComm SquawkComm is our work for cost-effective communication among
mobile devices in nearby vehicles. Chapter 5 provides further details.
Unobtrusive communication Unobtrusive communication is efficient wireless
communication on mobile devices that minimizes user interruption. For exam-
ple, device users should be able to discover nearby devices and networks without
waiting 10.24 seconds (for Bluetooth) or establishing network connections man-
ually (for Bluetooth and Wi-Fi). Chapter 3 elaborates such communication.
Vehicular Area Network (VANET) VANETs are ad hoc networks consisting of
vehicles that communicate with roadside infrastructure, other vehicles nearby, or
both. These cases are referred to as vehicle-to-infrastructure (V2I) and vehicle-
to-vehicle (V2V) communications, respectively.
Wireless Local Area Network (WLAN) WLANs are wireless networks that or-
ganizations deploy for Internet access on their premises. WLANs have one or
more APs. Wi-Fi (IEEE 802.11) is the standard for WLAN communication.
108
Bibliography
[1] Cisco Systems, Inc. Cisco visual networking index: global mobile data traf-fic forecast update, 2016–2021 white paper, 7 February 2017. https://www.
[2] Population Reference Bureau. 2016 world population data sheet, August2016. http://www.prb.org/Publications/Datasheets/2016/2016-world-
population-data-sheet.aspx.
[3] Xinfeng Li. Time-sensitive information communication, sensing, and computingin cyber-physical systems. PhD thesis, The Ohio State University, 2014.
[4] Theodore P. Pavlic. Design and analysis of optimal task-processing agents.PhD thesis, The Ohio State University, 2010. http://pubs.tedpavlic.com/
hg/Pavlic10/.
[5] Chris Monson. LaTeX Makefile, 2011. https://github.com/shiblon/latex-makefile.
[7] ITU. ITU releases 2016 ICT figures, 22 July 2016. http://www.itu.int/en/mediacentre/pages/2016-PR30.aspx.
[8] Gartner, Inc. Gartner says worldwide sales of smartphones grew 7 percent inthe fourth quarter of 2016, 15 February 2017. https://www.gartner.com/
newsroom/id/3609817.
[9] Pew Research Center. Mobile fact sheet, 12 February 2017. http://www.
pewinternet.org/fact-sheet/mobile/.
[10] Duygu Balcan, Vittoria Colizza, Bruno Goncalves, Hao Hu, Jose J. Ram-asco, and Alessandro Vespignani. Multiscale mobility networks and the spa-tial spreading of infectious diseases. Proc. National Academy of Sciences,106(51):21484–21489, 2009.
[11] Jing Yuan, Yu Zheng, and Xing Xie. Discovering regions of different functionsin a city using human mobility and POIs. In Proc. ACM KDD, pages 186–194,2012.
[12] Vincent D. Blondel, Adeline Decuyper, and Gautier Krings. A survey of resultson mobile phone datasets analysis. arXiv, 2015. https://arxiv.org/abs/
1502.03406.
[13] Diala Naboulsi, Marco Fiore, Stephane Ribot, and Razvan Stanica. Mobiletraffic analysis: a survey. Technical Report hal-01132385, Universite de Lyon,2015.
[14] Tristan Henderson, David Kotz, and Ilya Abyzov. The changing usage of amature campus-wide wireless network. In Proc. ACM MobiCom, 2004.
[15] Injong Rhee, Minsu Shin, Seongik Hong, Kyunghan Lee, Seong Joon Kim, andSong Chong. On the Levy-walk nature of human mobility. IEEE/ACM Trans-actions on Networking, 19(3):630–643, 2011.
[16] Felix Hernandez-Campos and Maria Papadopouli. A comparative measurementstudy of the workload of wireless access points in campus networks. In Proc.IEEE PIMRC, 2005.
[17] Magdalena Balazinska and Paul Castro. Characterizing mobility and networkusage in a corporate wireless local-area network. In Proc. ACM MobiSys, 2003.
[18] Marvin McNett and Geoffrey M. Voelker. Access and mobility of wirelessPDA users. ACM SIGMOBILE Mobile Computer and Communication Review,9(2):40–55, 2005.
[19] Wei-Jen Hsu, Debojyoti Dutta, and Ahmed Helmy. Structural analysis of userassociation patterns in university campus wireless LANs. IEEE Transactionson Mobile Computing, 11(11):1734–1748, 2012.
[20] Nils Aschenbruck, Aarti Munjal, and Tracy Camp. Trace-based mobility model-ing for multi-hop wireless networks. Computer Communications, 34(6):704–714,2011.
[21] Daimler AG. Mercedes Benz intelligent drive. https://www.mercedes-benz.com/en/mercedes-benz/innovation/mercedes-benz-intelligent-drive/.
[22] BMW. Active cruise control. http://www.bmw.com/com/en/insights/
[24] The iPhone Wiki. Malware for iOS. https://www.theiphonewiki.com/wiki/Malware_for_iOS.
[25] Kimberly Tam, Ali Feizollah, Nor Badrul Anuar, Rosli Salleh, and LorenzoCavallaro. The evolution of android malware and android analysis techniques.ACM Computing Surveys, 49(4), January 2017.
[26] Sanjit Biswas, John Bicket, Edmund Wong, Raluca Musaloiu-E, Apurv Bhartia,and Dan Aguayo. Large-scale measurements of wireless network behavior. InProc. ACM SIGCOMM, 2015.
[27] Fosca Giannotti, Mirco Nanni, Dino Pedreschi, Fabio Pinelli, Chiara Renso,Salvatore Rinzivillo, and Roberto Trasarti. Unveiling the complexity of hu-man mobility by querying and mining massive trajectory data. VLDB Journal,20(5):695–719, 2011.
[28] Marta C. Gonzalez, Cesar A. Hidalgo, and Albert-Laszlo Barabasi. Understand-ing individual human mobility patterns. Nature, 453(7196):779–782, 2008.
[29] D. Brockmann, L. Hufnagel, and T. Geisel. The scaling laws of human travel.Nature, 439(7075):462–465, 2006.
[30] Chaoming Song, Zehui Qu, Nicholas Blumm, and Albert-Laszlo Barabasi. Lim-its of predictability in human mobility. Science, 327(5968):1018–1021, 2010.
[31] Shao-Meng Qin, Hannu Verkasalo, Mikael Mohtaschemi, Tuomo Hartonen, andMikko Alava. Patterns, entropy, and predictability of human mobility and life.PLoS One, 7(12), December 2012.
[32] Xin Lu, Erik Wetter, Nita Bharti, Andrew J Tatem, and Linus Bengtsson.Approaching the limit of predictability in human mobility. Scientific Reports,3, 2013.
[33] Eunjoon Cho, Seth A. Myers, and Jure Leskovec. Friendship and mobility:User movement in location-based social networks. In Proc. ACM KDD, pages1082–1090, 2011.
[35] Tom H. Luan, Xuemin Shen, Fan Bai, and Limin Sun. Feel bored? Join Verse!Engineering vehicular proximity social networks. IEEE Transactions on Vehic-ular Technology, 64(3):1120–1131, March 2015.
[36] Android Open Source Project. Issue 82: Support Wi-Fi ad hoc networking,2015. https://code.google.com/p/android/issues/detail?id=82.
[37] Shaohan Hu, Hengchang Liu, Lu Su, Hongyan Wang, Tarek F Abdelzaher, PanHui, Wei Zheng, Zhiheng Xie, and John A Stankovic. Towards automatic phone-to-phone communication for vehicular networking applications. In Proc. IEEEINFOCOM, 2014.
[38] Nathan Eagle and Alex Pentland. Social serendipity: mobilizing social software.IEEE Pervasive Computing, pages 28–34, April–June 2005.
[39] Nokia Corporation. Nokia Sensor. http://www.nokia-asia.com/support/
download-software/nokia-sensor.
[40] Ranveer Chandra, Jitendra Padhye, Lenin Ravindranath, and Alec Wolman.Beacon-stuffing: Wi-Fi without associations. In Proc. ACM HotMobile, 2007.
[41] Ranveer Chandra, Jitendra Padhye, and Lenin Ravindranath. Wi-Fi neighbor-cast: enabling communication among nearby clients. In Proc. ACM HotMobile,2008.
[42] MaJose Villanueva, Carlos T. Calafate, Juan-Carlos Cano, and Pietro Manzoni.Solving the MANET autoconfiguration problem using the 802.11 SSID field. InProc. ACM MoMM, 2010.
[43] Jose Cano Reyes, Eduardo Burgoa, Carlos T. Calafate, Juan-Carlos Cano, andPietro Manzoni. A MANET autoconfiguration system based on Bluetooth tech-nology. In Proc. IEEE ISWCS, 2006.
[44] Nigel Davies, Adrian Friday, Peter Newman, Sarah Rutlidge, and Oliver Storz.Using Bluetooth device names to support interaction in smart environments.In Proc. ACM MobiSys, pages 151–164, 2009.
[45] David W. McDonald, Joseph F. McCarthy, Suzanne Soroczak, David H.Nguyen, and Al M. Rashid. Proactive displays: Supporting awareness in fluidsocial environments. ACM Transactions on Computer-Human Interaction,14(4):16:1–16:31, 2008.
[46] Jin Teng, Boying Zhang, Xinfeng Li, Xiaole Bai, and Dong Xuan. E-Shadow:Lubricating social interaction using mobile phones. In Proc. IEEE ICDCS,2011.
[49] Samuel David Perli, Nabeel Ahmed, and Dina Katabi. PixNet: Interference-free wireless links using LCD-camera pairs. In Proc. ACM MobiCom, pages137–148, 2010.
[50] Xian Chen, Ruofan Jin, Kyoungwon Suh, Bing Wang, and Wei Wei. NetworkPerformance of Smart Mobile Handhelds in a University Campus WiFi Network.In Proc. ACM IMC, 2012.
[51] Aaron Gember, Ashok Anand, and Aditya Akella. A Comparative Study ofHandheld and Non-Handheld Traffic in Campus Wi-Fi Networks. In Proc.PAM, 2011. Springer.
[52] Xuetao Wei, Nicholas C. Valler Valler, Harsha V. Madhyastha, Iulian Neamtiu,and Michalis Faloutsos. A Behavior-aware Profiling of Handheld Devices. InProc. IEEE INFOCOM, 2015.
[53] Nathan Eagle and Alex Pentland. Reality mining: sensing complex social sys-tems. Journal of Personal Ubiquitous Computing, 10(4):255–268, March 2006.
[54] CRAWDAD. http://crawdad.org.
[55] Cristian Tuduce and Thomas Gross. A mobility model based on WLAN tracesand its validation. In Proc. IEEE INFOCOM, 2005.
[56] Aaron Clauset, Cosma Rohilla Shalizi, and M. E. J. Newman. Power-law dis-tributions in empirical data. SIAM Review, 51(4):661–703, November 2009.
[57] Minkyong Kim, David Kotz, and Songkuk Kim. Extracting a mobility modelfrom real user traces. In Proc. IEEE INFOCOM, volume 6, pages 1–13, 2006.
[58] Wei-Jen Hsu, Thrasyvoulos Spyropoulos, Konstantinos Psounis, and AhmedHelmy. Modeling spatial and temporal dependencies of user mobility in wirelessmobile networks. IEEE/ACM Transactions on Networking, 17(5):1564–1577,2009.
[59] Chaoming Song, Tal Koren, Pu Wang, and Albert-Laszlo Barabasi. Modellingthe scaling properties of human mobility. Nature Physics, 6(10):818–823, Oct2010. http://dx.doi.org/10.1038/nphys1760.
[60] Jungkeun Yoon, Brian D. Noble, Mingyan Liu, and Minkyong Kim. Buildingrealistic mobility models from coarse-grained traces. In Proc. ACM MobiSys,2006.
[61] Mukesh Saini, Abdulhameed Alelaiwi, and Abdulmotaleb El Saddik. How closeare we to realizing a pragmatic VANET solution? A meta-survey. ACM Com-puting Surveys, 48(2), November 2015. article 29, 40 pages.
[62] Riccardo Crepaldi, Ryan Beavers, Braden Ehrat, Matt Jaeger, Steven Bier-steker, and Robin Kravets. LoadingZones: Leveraging street parking to enablevehicular internet access. In Proc. ACM CHANTS, 2012.
[63] Riccardo Crepaldi, Mehedi Bakht, and Robin Kravets. QuickSilver:Application-driven inter- and intra-cluster communication in VANETs. In Proc.ACM MobiOpp, 2012.
[64] Jing Zhao and Guohong Cao. VADD: Vehicle-assisted data delivery in vehicularad hoc networks. IEEE Transactions on Vehicular Technology, 57(3):1910–1922,2008.
[65] Lu Han, Stephen Smaldone, Pravin Shankar, James Boyce, and Liviu Iftode.Ad-hoc voice-based group communication. In Proc. IEEE PerCom, 2010.
[66] Bret Hull, Vladimir Bychkovsky, Yang Zhang, Kevin Chen, Michel Goraczko,Allen Miu, Eugene Shih, Hari Balakrishnan, and Samuel Madden. CarTel: Adistributed mobile sensor computing system. In Proc. ACM SenSys, 2006.
[67] Nilanjan Banerjee, Mark D. Corner, and Brian Neil Levine. Design andfield experimentation of an energy-efficient architecture for DTN throwboxes.IEEE/ACM Transactions on Networking, 18(2):554–567, April 2010.
[68] Fan Bai, Daniel D. Stancil, and Hariharan Krishnan. Toward understandingcharacteristics of Dedicated Short Range Communications (DSRC) from a per-spective of vehicular network engineers. In Proc. ACM MobiCom, 2010.
[69] Wen-Long Jin, Candy Kwan, Zhe Sun, Hao Yang, and Qijian Gan. SPIVC: Asmartphone-based inter-vehicle communication system. In Proc. TRB, 2012.
[70] Yongtae Park, Jihun Ha, Seungho Kuk, Hyogon Kim, Chieh-Jan Mike Liang,and JeongGil Ko. A feasibility study and development framework design forrealizing smartphone-based vehicular networking systems. IEEE Transactionson Mobile Computing, 13(11):2431–2444, November 2014.
[71] Pilsoon Choi, Jason Gao, Nadesh Ramanathan, Mengda Mao, Shipeng Xu,Chirn-Chye Boon, Suhaib A. Fahmy Fahmy, and Li-Shiuan Peh. A case for
114
leveraging 802.11p for direct phone-to-phone communications. In Proc. ACMISPLED, 2014.
[72] National Highway Traffic Safety Administration. Transportation Sec. Foxx an-nounces steps to accelerate road safety innovation, 13 May 2015. http://www.nhtsa.gov/About+NHTSA/Press+Releases/2015/nhtsa-will-accelerate-
v2v-efforts.
[73] U.S. Department of Transportation. http://www.its.dot.gov/factsheets/
dsrc_factsheet.htm.
[74] M. I. Hassan, H. L. Vu, and T. Sakurai. Performance analysis of the IEEE802.11 MAC protocol for DSRC safety applications. IEEE Transactions onVehicular Technology, 60(8), October 2011.
[75] Nicholas D. Lane, Emiliano Miluzzo, Hong Lu, Daniel Peebles, Tanzeem Choud-hury, and Andrew T. Campbell. A survey of mobile phone sensing. IEEECommunications Magazine, 48(9):140–150, September 2010.
[76] Delphine Christin, Andreas Reinhardt, Salil S. Kanhere, and Matthias Hollick.A survey on privacy in mobile participatory sensing applications. Journal ofSystems and Software, 84(11):1928–1946, November 2011.
[77] Prashanth Mohan, Venkata N. Padmanabhan, and Ramachandran Ramjee.Nericell: Rich monitoring of road and traffic conditions using mobile smart-phones. In Proc. ACM SenSys, 2008.
[78] Raghu K. Ganti, Nam Pham, Hossein Ahmadi, Saurabh Nangia, and Tarek F.Abdelzaher. GreenGPS: A participatory sensing fuel-efficient maps application.In Proc. ACM MobiSys, 2010.
[79] Rajesh Krishna Balan, Khoa Xuan Nguyen, and Lingxiao Jiang. Real-time tripinformation service for a large taxi fleet. In Proc. ACM MobiSys, 2011.
[80] Jules White, Chris Thompson, Hamilton Turner, Brian Dougherty, and Dou-glas C. Schmidt. WreckWatch: Automatic traffic accident detection and no-tification with smartphones. Mobile Networks and Applications, 16:285–303,2011.
[81] Arvind Thiagarajan, Lenin Ravindranath, Katrina LaCurts, Samuel Madden,Hari Balakrishnan, Sivan Toledo, and Jakob Eriksson. VTrack: Accurate,energy-aware road traffic delay estimation using mobile phones. In Proc. ACMSenSys, 2009.
[86] Ke-Chian Su, Hei-Min Wu, Wei-Ling Chang, and Yao-Hsin Chou. Vehicle-to-vehicle communication system through Wi-Fi network using Android smart-phone. In Proc. IEEE ICCVE, 2012.
[87] Sacha Trifunovic, Bernhard Distl, Dominik Schatzmann, and Franck Legendre.WiFi-Opp: Ad-hoc-less opportunistic networking. In Proc. ACM CHANTS,2011.
[88] Ahmad Rahmati, Lin Zhong, Venu Vasudevan, Jehan Wickramasuriya, andDaniel Stewart. Enabling pervasive mobile applications with the FM radiobroadcast data system. In Proc. ACM HotMobile, 2010.
[89] Christopher P. Paolini, Christopher Aguirre, Mahasweta Sarkar, and SantoshNagaraj. An ad hoc broadcast scheme using the Radio Data System (RDS) inthe FM band. In Proc. IEEE WCECS, 2013.
[90] Hang Yu, Ahmad Rahmati, Ardalan Amiri Sani, Lin Zhong, Jehan Wickrama-suriya, and Venu Vasudevan. Data broadcasting using mobile FM radio: Design,realization and application. In Proc. ACM UbiComp, 2011.
[91] Jiao Wang, Fang Cheng, Jiangshan Wang, Guobin Shen, and Xiaofan Jiang.Genius-on-the-Go: FM radio based proximity sensing and audio informationsharing. In Proc. ACM SenSys, 2012.
[92] Liqun Li, Guoliang Xing, Limin Sun, Wei Huangfu, Ruogu Zhou, and HongsongZhu. Exploiting FM Radio Data System for adaptive clock calibration in sensornetworks. In Proc. ACM MobiSys, 2011.
[93] Marcel Flores, Uri Klarman, and Aleksandar Kuzmanovic. Wi-FM: Resolvingneighborhood wireless network affairs by listening to music. In Proc. ACMICNP, 2015.
[94] Jonathan Petit, Florian Schaub, Michael Feiri, and Frank Kargl. Pseudonymschemes in vehicular networks: A survey. IEEE Communications Surveys andTutorials, 17(1):228–255, 2015. Q1.
[95] A. B. M. Musa and Jakob Eriksson. Tracking Unmodified Smartphones UsingWi-Fi Monitors. In Proc. ACM SenSys, 2012.
[96] Ben Greenstein, Ramakrishna Gummadi, Jeffrey Pang, Mike Y. Chen, Ta-dayoshi Kohno, Srinivasan Seshan, and David Wetherall. Can Ferris Buellerstill have his day off? Protecting privacy in the wireless era. In Proc. ACMHOTOS, 2007.
[97] Shravan Gaonkar, Jack Li, Romit Roy Choudhury, Landon Cox, and Al Shmidt.Micro-Blog: Sharing and querying content through mobile phones and socialparticipation. In Proc. ACM MobiSys, 2008.
[98] Emiliano Miluzzo, Nicholas D. Lane, Krisof Fodor, Ronald Peterson, Hong Lu,Mirco Musolesi, Shane B. Eisenman, Xiao Zheng, and Andrew T. Campbell.Sensing meets mobile social networks: the design, implementation and evalua-tion of the CenceMe application. In Proc. ACM SenSys, 2008.
[99] Boying Zhang, Jin Teng, Xiaole Bai, Zhimin Yang, and Dong Xuan. P 3-Coupon:A probabilistic system for prompt and privacy-preserving electronic coupondistribution. In Proc. IEEE PerCom, 2011.
[100] David J. C. MacKay. Fountain codes. IEE Proceedings-Communications,152(6):1062–1068, 2005.
[101] M. Luby. LT codes. In Proc. ACM Foundations of Computer Science, pages271–282, 2002.
[102] Jeffrey Pang, Ben Greenstein, Ramakrishna Gummadi, Srinivasan Seshan, andDavid Wetherall. 802.11 user fingerprinting. In Proc. ACM MobiCom, 2007.
[103] Tom Rowan. Negotiating wifi security. Network Security, 2010(2):8–12, 2010.
[104] Chloe Albanesius. HTC smartphone vulnerability exposes your personal data.http://www.pcmag.com/article2/0,2817,2393999,00.asp.
[105] Manabu Ohta, Atsuhiro Takasu, and Jun Adachi. Retrieval methods for Englishtext with missrecognized OCR characters. In Proc. IEEE ICDAR, pages 950–956, 1997.
[114] Ioannis Kontoyiannis, Paul H. Algoet, Yu M. Suhov, and Abraham J. Wyner.Nonparametric entropy estimation for stationary processes and random fields,with applications to English text. IEEE Transactions on Information Theory,44(3):1319–1327, 1998.
[122] NHTSA. 2013 motor vehicle crashes: Overview. Technical Report DOT HS812 101, NHTSA, December 2014. http://www-nrd.nhtsa.dot.gov/Pubs/
812101.pdf.
[123] BusinessWire. Millennials most likely to indulge in luxury hotel services, ChaseMarriott rewards survey reveals, 6 June 2014. http://www.businesswire.
[134] Nicola Bellomo and Christian Dogbe. On the modeling of traffic and crowds: Asurvey of models, speculations, and perspectives. SIAM Review, 53(3):409–463,2011.
[135] U.S. Department of Justice. Americans with disabilities act title III, 1991.http://www.ada.gov/reg3a.html#anchor9.
[136] U.S. Department of Transportation Federal Highway Administration. Federalsize regulations for commercial motor vehicles, 2004. http://ops.fhwa.dot.