Human-centered Computing Lab Contextual Inference and - - PowerPoint PPT Presentation

▶
human centered computing lab
SMART_READER_LITE
LIVE PREVIEW

Human-centered Computing Lab Contextual Inference and - - PowerPoint PPT Presentation

Sep 19, 2022 335 likes •567 views

Human-centered Computing Lab Contextual Inference and Characterization Derived from Wireless Data Mining Rute Sofia (rute.sofia@ulusofona.pt) http://copelabs.ulusofona.pt Agenda Background and COPELABS R&D Unit 2017 Projects and

slide-1
SLIDE 1

http://copelabs.ulusofona.pt

Human-centered Computing Lab

Contextual Inference and Characterization Derived from Wireless Data Mining

Rute Sofia (rute.sofia@ulusofona.pt)

slide-2
SLIDE 2

Agenda

  • Background and COPELABS R&D Unit
  • 2017 Projects and Main Outcome
  • Wireless Network Mining Tools
  • Nsense and PerSense Mobile Light as Examples
  • Wireless Network Mining
  • Tracking Indicators and Inference Examples
  • Ongoing Experiments
  • Summary, network operation applicability

2 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

slide-3
SLIDE 3

My Background

Rute C. Sofia, PhD

  • 2010-, Senior Researcher COPELABS/Associate

Professor University Lusofona, Lisbon, Portugal

  • COPELABS scientific coordinator (vice-director)
  • Before:
  • COPELABS Director (2014-2016)
  • Scientific Director SITILabs (2010 – 2013)
  • Area leader, INESCTEC (2007 – 2010)
  • Senior Researcher, Siemens AG CT IC2, DE

(2004 – 2007)

  • Senior Researcher, Bundeswehr Universitaet,

DE (2004)

  • WAN Engineer, FCCN, PT (1998-2003)
  • Visiting Scholar, Univ. Pennsylvania, EUA (2000

– 2003)

  • Visiting Scholar, ICAIR/Northwestern

University, Evanston, IL, USA (2000)

  • Grupo Forum, PT Networking systems

admin/Web team coordinator (1995-1998)

  • Packet-based networking: IPv4/IPv6;

carrier grade Ethernet; QoS; Mobility management and estimation

  • Current focus: IoT, wireless networks;

network architectural design that integrates social aspects (e.g. better spectrum sharing; social proximity)

3 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

slide-4
SLIDE 4

COPELABS R&D Unit

Structure

  • Multidisciplinary unit: informatics and psychology
  • 2013 FCT evaluation: “Good” (Multidisciplinary; small unit; high experimental intensity)
  • Private, not-for profit entity; secondary management institution: COFAC c.r.l.

(University Lusofona de Humanidades e Tecnologias)

Main Research Lines 2013-2020

  • Internet Science
  • Pervasive wireless systems
  • Networking dynamics
  • Cyberpsychology
  • Assessment/rehabilitation of psychological disorders

Focus

  • Promote well-being via pervasive, non-intrusive, and networked technology

Team

  • 28 reseachers (15 Ph.D.)
  • SITI: Informatics Systems and Technologies
  • CTIP: Cognitive and Technology Intensive Psychology

Data Mining, proximity (Proxemics Data Lab)

4 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

slide-5
SLIDE 5

Rute Sofia, PhD Networking dynamics Paulo Mendes, PhD Pervasive Wireless Systems Miguel Tavares People-centric sensing Software Engineer/Jr Researcher Francisco Pereira., MSc (2014/2015) Inferring behaviour Liliana Carvalho Pervasive sensing applied to social well-being Samrat Dattagupta. MSc Data Mining Community dynamics

  • Jr. Researcher

Seweryn Dynerowicz, PhD Opportunistic networking Researcher Jose Faisca, MSc (2012/2013) Personal Cloud Systems Rui Ribeiro, PhD Open-source systems Jose Rogado, PhD Security, Parallel Computing

PhD Students Researchers Jr Researchers Senior Researchers 17 researchers

7 PhD students

2017 Team

Preyesse Arquissandas, MSc (2013/2014) Sensor based Augmented Reality Aurea Costa, MSc (2014/2015) Sensor based Pervasive Interaction in Sport Monica Pedro, MSc (2014/2015) Implicit Interactions Omar Aponte, MSc Opportunistic networking, people-centric sensing Jr Researcher José Soares (undergrade) Software Engineering support Andres Mrad, MSc (2016/2017) Gaming Design framework Helder Valente (2016/2017) Unified communications in IoT

5 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

slide-6
SLIDE 6

Scientific and Technologic Focus 2016 / 2017

2016 2017

7 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

slide-7
SLIDE 7

Project Overview 2017

10

H2020 UMOBILE (2015-2018)

  • Pervasive wireless systems
  • Universal, mobile-centric and opportunistic

communications architecture

  • COPELABS: adding opportunistic communications

to Named-data networking CitySense (2015-2018)

  • People-centric sensing
  • Development of sensing middleware to analyze

and stimulate social interaction Proxemics Data Lab (2016-2018)

  • Social networking dynamics
  • Network mining: proxemics and social interaction

10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

slide-8
SLIDE 8

11 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

H2020 UMOBILE

Main Outcome

NDN-Opp: Opportunistic Routing

  • Pervasive wireless systems
  • Universal, mobile-centric and opportunistic communications

architecture

  • COPELABS: adding opportunistic communications to Named-

data networking

  • Contacts: Seweryn Dynerowicz, Paulo Mendes

Now@, Oi!

  • Opportunistic data dissemination
  • Contacts: Omar Aponte, Paulo Mendes

Contextualization Manager

  • Social networking dynamics
  • Network mining: proxemics and social interaction
  • Open-source module for usage and affinity network

(neighborhood) characterization

  • Contacts: Jose Soares, Rute Sofia
slide-9
SLIDE 9

CitySense

Main Outcome

mHealth: Elderly Social Stimulation Goal: Improve life experience

  • Detect isolation behaviors to trigger

alerts and actions

  • Detect common interests and behaviors

to stimulate social contacts

  • Increase social interaction by provide

tracking information in a controlled environment Connected Mobility Goal: Improve mobility in urban scenarios

  • Exploit the car as a Data Drone
  • Use social evidence to improve mobility

patterns.

  • Identify the best correlation among all

mobility forms used by a community

Nsense v1.0

  • Open-source pervasive wireless sensing middleware
  • Pipelines: motion; proximity; sound activity
  • Contextualization: social interaction level and probability of interaction
  • Sofia, Rute C.; Firdose, Saeik; Lopes, Luis Amaral; Moreira, Waldir;

Mendes, Paulo. NSense: A People-centric, non-intrusive Opportunistic Sensing Tool for Contextualizing Social Interaction. IEEE Healthcom 2016: 2016 IEEE 18th International Conference on eHealth Networking, Application, Services, pp 1-6, DOI: 10.1109/HealthCom.2016.7749490 NSense v2.0

  • Pipelines added: mobility
  • Features added: interest exchange

12

slide-10
SLIDE 10

Proxemics Data Lab (2016-2018)

http://copelabs.ulusofona.pt/~pdlab/ Behavior correlation and inference via pervasive, non-intrusive technology

Scientific Areas

Community/Group Dynamics

Physical psychological proximity patterns

Early detection of neuro/psychologica l disorders

Mild Cognitive Impairments

How

Traces Data Mining

13 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

slide-11
SLIDE 11

Proxemics Definitions

  • Dictionary (Merriem-Webster)

the study of the nature, degree, and effect of the spatial separation individuals naturally maintain (as in various social and interpersonal situations) and of how this separation relates to environmental and cultural factors

  • Origin: 1960-65; prox(imity) + -emics (extracted from phonemics);

apparently coined by U.S. anthropologist Edward T. Hall (born 1914)

  • Sociology, psychology: the study of the spatial requirements of humans

and animals and the effects of population density on behavior, communication, and social interaction.

  • Proxemics in COPELABS
  • Physical distance (personal, social and public spaces)
  • Study of nonverbal communication factors: Kinesthetic, voice, touching
  • Neuropsychology: personal space in terms of the kinds of "nearness" to the

body.

  • Cultural factors, namely adaptation: relationships may allow for personal

space to be modified, including friendships and close acquaintances. Physical and psychological aspects

14 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

slide-12
SLIDE 12

Proxemics Data Lab @COPELABS

Main Outcome

  • M. Bianchi, Anna Pegna, R. Sofia, Igor dos Santos, Ana Loureiro, Joana Santos, Ricardo Rodrigues,

Samrat Dattagupta. Social Interaction Analysis with non-intrusive wireless technology in children. 09.2017.

  • Tool: PerSense Mobile Light (Senception Lda) and surveys
  • Population: 80 children (10-12), 1 school in Lisbon
  • Duration: May 2017
  • Purpose: i) contact and prejudice in children; ii) well-being and spaces; iii) physical proximity and mobility
  • URL: September 2017
  • M. Tavares, P. Mendes, R. Brito. Nearness and Interests Traces. 2017.04
  • Tool: Nsense v2.0
  • Population: ,15 students (out of 50)
  • Duration: 05.04.2017-06.04.2017.
  • Purpose: study influence in psychological proximity
  • URL: http://siti2.ulusofona.pt:8085/xmlui/handle/20.500.11933/699

sofona.pt:8085/xmlui/handle/20.500.11933/699

  • S. Firdose, L. Lopes, W. Moreira, R. Sofia, P. Mendes. Data concerning social interaction and propinquity

based on wireless and bluetooth. 2017.01

  • Tool: Nsense v1.0
  • Population: 5 elements
  • Duration: 22 hours and 50 hours
  • URL: http://crawdad.org/copelabs/usense/20170127/
  • S. Firdose, L. Lopes, W. Moreira, R. Sofia, P. Mendes, Interpersonal space traces. 2017.01
  • Tool: Nsense v2.0
  • Population:9 elements
  • Duration: 12 days (12 days from 12th September to 23rd September 2016)
  • http://crawdad.org/copelabs/usense/20170127/NSense%20Data%20set%20II/

15 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

slide-13
SLIDE 13

NSense

  • Purpose:

Nearness inference

  • Status: traces;

publications; open- source software

  • Support: COPELABS

PerSense Mobile Light*

  • Purpose:

Analyze roaming habits and spaces

  • Status: traces;

publications; IPR*

  • Support: H2020

UMOBILE (gr nr 645124); EU IST FP7 ULOOP (gr. nr 721458)

  • *Patent pending

Oi!

  • Purpose:

Data transmission with intermittent Internet connectivity

  • Status: traces;

publications; open- source software

  • Support: H2020

UMOBILE

Ddicas

  • Purpose:

Recommendations

  • ver Wi-Fi
  • Status: Pilots -

LxFactory, Lisbon; Câmara Municipal de Portimão (Rua das Lojas)

  • Support: EU IST

FP7 ULOOP

Proxemics Data Lab

Mining Tools

16 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

slide-14
SLIDE 14

Mining Tools: Non-intrusive wireless technology NSense

17

*

  • Rute C. Sofia, Saeik Firdose, Luis Amaral Lopes, Waldir Moreira and Paulo Mendes, NSense: A People-centric, non-intrusive

Opportunistic Sensing Tool for Contextualizing Social Interaction (2016), in: IEEE Healthcom 2016: 2016 IEEE 18th International Conference on eHealth Networking, Application, Services

  • Luis Amaral Lopes, Saeik Firdose, Rute C. Sofia and Paulo Mendes, USENSE: a People-centric Opportunistic Sensing Tool (2016),

in: Infocom 2016

slide-15
SLIDE 15
  • Android App developed in the context of the H2020 UMOBILE project
  • Senception Lda (COPELABS spin-off)
  • What it does: mines wireless networks non-intrusively**
  • Wi-Fi and Wi-Fi Direct; Bluetooth (future)
  • Captures wireless foot printing aspects (distances, APs; visits’ type and

duration); and geo-location

  • All data stored LOCALLY and in accordance with European guidelines
  • Generates csv reports daily – researchers can get them via e-mail.
  • PML does not collect any personal data
  • Its Purpose: industrial investigation - scientific studies and traces

concerning roaming and interaction aspects

  • Can be extended upon request, to capture parameters relevant to

interested parties

  • Where it is being (further) applied:
  • PhD students, smart cities data extraction
  • UMOBILE project
  • Questions? info@senception.com

18

** patent pending.

Mining Tools: Non-intrusive wireless technology PerSense Mobile Light (PML)

  • Rute C. Sofia, A Tool to Estimate Roaming Behavior in Wireless Architectures (2015), in: WWIC2015, Wired/Wireless Internet

Communications Volume 9071 of the series Lecture Notes in Computer Science, 9071(pp 247-258) Rute C. Sofia and Paulo Mendes, A Characterization Study of Human Wireless Footprints with PerSense Mobile Light (SHORT VERSION UNDER SUBMISSION), COPELABS, University Lusofona & Senception Lda, number COPE-SITI-TR-16-01; Senception TR-16-01, 2016

18

slide-16
SLIDE 16

Network Wireless Mining: What can we track ? The UMOBILE Contextual Manager Example

  • Set 1: Affinity Network Characterization Data
  • Peer status over time and space; affinities between

source node and peers

  • Affinity network information concerns, i.e., peer

status over time and space as well as affinities (matches) between source nodes and peers.

  • EXAMPLES of INDICATORS that can be passed
  • Peer list (bluetooth and Wi-Fi Direct) at instant t
  • r over time window T.
  • Interests associated to each peer.
  • Battery status of each peer.
  • Average, max, min connectivity duration over

period T.

  • Average. Max, min contact duration.
  • Average node degree over time and space.
  • Cluster distance.
  • Visited networks (Aps, SSID, etc)

characterization.

  • Set 2: Usage and Similarity Characterization Data
  • Indicators that can be provided and that concern

usage and similarty characterization are built upon data collected internally (in the device)

  • EXAMPLES of Indicators that can be passed
  • Preferred visited network and/or geo-location.
  • Type (category) of preferred application (e.g.

most used over time window T).

  • Time spent per application category (e.g. per

day).

Periodically (regular scans, 30s)

Storage Visited Networks

Bluetooth

Affinity network

Wi-Fi Wi-Fi Direct

Contextual Manager Service Resource usage Periodically (regular checks, e.g. 60 minutes)

19 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

slide-17
SLIDE 17

20

Network Wireless Mining: What can we Infer? Example: Physical Proximity and Mobility

  • M. Bianchi, Anna Pegna, R. Sofia, Igor dos Santos, Ana Loureiro, Joana Santos, Ricardo Rodrigues, Samrat Dattagupta. Social

Interaction Analysis with non-intrusive wireless technology in children. 09.2017.

Data collected, 1 day (05.05.2017), PML connected (1) vs crossed access points (0 – blue) Data collected, 1 day (05.05.2017) distribution of visited APs over time Data collected, 1 day (05.05.2017) 1 single scan, peers around

  • Data from multiple days and multiple devices – correlation being currently analyzed.

20 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

slide-18
SLIDE 18

21

Network Wireless Mining: What can we Infer? Example: Social Proximity

  • Rute C. Sofia and Paulo Mendes, A Characterization Study of Human Wireless Footprints

with PerSense Mobile Light (SHORT VERSION UNDER SUBMISSION), COPELABS, University Lusofona & Senception Lda, number COPE-SITI-TR-16-01; Senception TR-16- 01, 2016

  • Tool: PML

21 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt) Clustering, latitude vs. longitude

slide-19
SLIDE 19

22

Network Wireless Mining: What can we Infer? Example: Physical Proximity and Mobility

  • S. Firdose, L. Lopes, W. Moreira, R. Sofia, P. Mendes. Data

concerning social interaction and propinquity based on wireless and bluetooth. 2017.01

  • Tool: NSense

22 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

slide-20
SLIDE 20

23

Network Wireless Mining: What can we Infer? Example: Physical Proximity and Mobility

  • M. Tavares, P. Mendes, R. Brito. Nearness and Interests
  • Traces. 2017.04
  • Tool: NSense

23 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt) Radical sports’ interest per connected devices Social interaction vs. distance

slide-21
SLIDE 21

Network Mining, Summarizing

Applicability in the Context of Networking

  • Analyse social behavior (e.g. skype calls within a specific area)
  • Considers traffic locality
  • Place networking nodes accordingly to user perceived usefulness
  • Improve adoption modelling

Connectivity Modeling

  • Improve dissemination of information (e.g. opportunistic data

dissemination based on contextualization)

  • Social interaction as the vehicle for dissemination

Routing

  • Improve resource allocation (QoS)
  • Improve user satisfaction (QoE)

Resource Management

  • Estimate movement
  • Anticipate movement
  • Improve mobility management
  • Develop new services

Mobility management and modeling

24 10.07. 2017 R. Sofia (rute.sofia@ulusofona.pt)

slide-22
SLIDE 22

25