Docstoc

Mobile Social Networks Mobile Social Networks

Document Sample
Mobile Social Networks Mobile Social Networks Powered By Docstoc
					Lecture #3



Mobile Social Networks – A New
            Vision

            Dr. Kun Yang
  University of Essex, Colchester, UK


   Monday 25th October 2010 @ NII, Tokyo

                                           1




                 Agenda

Overview
A Socially-aware Mobility Model for
Delay-tolerant Mobile Social Networks
Vehicular Social Network
  Overall Architecture
  Centrality-aware Bandwidth Allocation
Q&A


                                           2




                                               1
         Online Social Networks (OSN)

       Social Networks have attracted billions of active users
       under major online social network (OSN) systems such
       as MySpace, Facebook, etc.
       These systems allow people with common interests to
       come together and form virtual communities.
       They largely rely on fixed network infrastructure and
       centralized servers to work.




                                                             3




              Mobile Social Networks

  Nowadays these OSNs are increasingly used on mobile
  devices thus rendering a new field of mobile social networks.
  There is a historic shift from PC’s to mobile devices for
  Internet access, as driven by the advent of smart phones such
  as Apple’s iPhone, Google’s gPhone, etc.

Apple’s iPhone:                                 Google’s gPhone:
first released at                               with Android
Jan, 2007 and was                               Operating System.
awarded Time                                    It was released in
Magazine's                                      October 2008.
"Invention of the
Year" in 2007

                                                             4




                                                                     2
   Mobile Social Networks: an Example


Mobage Town: Japan’s
biggest mobile-only
social network.
   By DeNA Company
   11 million users as of Aug. 2008
   Weblink: www.mbga.jp
Functions: exchange messages,
chat in communities, share
music, read pocket novels, and
blog.
The site’s “killer feature” is the
vast selection of free games!                                                5
                               Courtesy to Serkan Toto on http://techcrunch.com




  Mobile Social Networks: Another Example

                                      Loopt: help to find out who’s
                                      around, what to do and where
                                      to go.
                                      Visualize friend’s location and
                                      information on a map.
                                      Free Loopt application on:




                                      Many other location-based
                                      services.
                                                                            6
                                      Courtesy to www.loopt.com




                                                                                  3
 What’s Mobile Social Network (MSN)?

   Migrating OSNs such as Facebook to mobile devices?
      Much work on the application layer, or how to use mobile phones
      to collect user information
   A new vision: an inherent marriage of OSN with mobile
   wireless networks

                                                 Mobile Social
            Social Networks
                                                 Networks

Social Knowledge             Network Status Information

                                                   Social Mobile
        Mobile wireless networks                   Networks
                                                                        7




             Social Mobile Networks
 How to introduce social knowledge into the design of
 network protocols and algorithms.
 Socially-inspired mobile networks: (offline) use of a social
 phenomenon (e.g., community) or social theory (e.g., small
 world) to the network algorithm design
    No direct connection with the social community
 Socially-aware mobile networks: online (on-the-fly) use of
 social knowledge to the benefit of wireless network design
    Certain interaction with the social (application) layer is needed
    More challenging
 Network problems to be addressed:
    Network planning, configuration, resource allocation, scheduling,
    QoS, mobility, network management, etc.
                                                                        8




                                                                            4
 Social Mobile Networks: Infrastructure
 A combination of base stations (and centralized servers on
 the Internet) and delay-tolerant (DT) ad hoc mobile nodes.
                                            youTube
               Facebook
                                                        Mobile Social
                                Internet                Network Server


                                Content Provider
MSN software
        BS                        BS                                AP




                                                                         9
                     Delay-tolerant Ad hoc Networks




                              Agenda

  Overview
  A Socially-aware Mobility Model for
  Delay-tolerant Mobile Social Networks
  Vehicular Social Network
     Overall Architecture
     Centrality-aware Bandwidth Allocation
  Q&A


          Courtesy to my Phd student: Nikos Vastardis
                                                                         10




                                                                              5
                        Methodology

 In such a mobile social network, routing (or message
 relaying) is critical. And such a MSN is a delay-tolerant
 (DT) network.
 Methodology (-- for discussion):
     To extend a normal ad hoc routing (such as AODV) to be DT.
     To develop a socially-aware mobility model (to select a social
     model such as Caveman model and apply it into mobility pattern)
     To use this mobility model to generate trace file
     To replace the default mobility model (such as random waypoint)
     with this trace file.
     To evaluate the socially-aware DT routing algorithm
     To make further improvement on both social model, DT routing,
     etc
     Can we introduce social knowledge directly into routing?
                                                                              11




         Why a new mobility model?

    The are only a few data banks capturing movement in
   large scale real ad-hoc environments, and their
   parameters cannot be varied.
    In existing mobility models such as the Random Way
   Point, it is difficult to assess the validity.
    It is important to model the movement of the nodes, in
   and between groups as clustering.
    Although theoretical models presenting small world
   behavior exist, social networks have a greater level of
   clustering.
Ref. M. Musolesi and C. Mascolo. “Designing Mobility Models based on Social
Network Theory”, Mobile Computing and Communications Review. V11(3).
July 2007.
                                                                              12




                                                                                   6
                        Model Input

 A social network is used as input
 for the mobility model, therefore
 node movement is influence by
 the background social structure.
    A weighted graph is used to
    represent the social network.
     Social ties are valued in the range
    [0,1], and represented in the
    Interaction Matrix.
     To identify community structures,
    algorithms such as the Girvan-
    Newman or the Caveman model
    may be used.


                                                                   13




                     Mobility Model
A grid is placed over the simulation ground and each of the identified
communities is places on a specific square. Each node is assigned to a
point on the grid and moves towards that (initially inside the assigned
grid square).
 The choice of next moving object and target is randomly made according
to social attractivity. (Deterministic vs. Probabilistic)




                                                                   14




                                                                          7
                                  Agenda

 Overview
 A Socially-aware Mobility Model for
 Delay-tolerant Mobile Social Networks
 Vehicular Social Network
     Overall Architecture
     Centrality-aware Bandwidth Allocation
 Q&A


                                                                                        15




                             Introduction
Vehicles are an essential part of our everyday life.
One particular thing most commuters would like to do
during the usually long and boring journey is to socialize
with other commuters on the road.
     These commuters usually travel every working day between home
     and office on the same roads or highways at roughly the same time
     day in and day out.
Therefore, this is an opportunity for them to form a virtual
social community on the road, namely, Vehicular Social
Network or VSN [1].
     But [1] focuses only on the application layer.

[1] S. Smaldone, L. Han, P. Shankar, L. Iftode. “RoadSpeak: enabling voice chat on roadways
using vehicular social networks”, In SocialNets'08: Proceedings of the 1st workshop on Social
network systems (2008), pp. 43-48
                                                                                        16




                                                                                                8
A VSN Scenario & Network Infrastructure




IEEE 802.16j introduces a new role of stations called relay stations (RSs). A RS
has lower complexity than a standard 802.16 BS.
SS: Subscriber Station
David registers himself onto a VSN with the following major information: time
= [2pm-4pm, Sunday 8th August 2010], location = [A30], interest = [Cornwall] 17




             Scenario Observations (1)

    Two infrastructure are needed.
    Firstly, a social network software platform is needed to
    present the commuters with a graphical interface to get
    access to the VSN services
        located on the VSN provider’s website
        or are available at other commuter’s smart phones.
    This is a platform most of which is visible to the end
    users. There is also another platform, namely, network
    infrastructure, which, though is not exposed to the end
    users, eventually fulfils the transmission of end users’
    traffic.


                                                                              18




                                                                                   9
           Scenario Observations (2)

 (1) a graphical user interface (GUI) needs to be presented to the
 commuters for them to interact with the concerned VSN and so does a
 backend web server to provide VSN services;
 (2) the virtual social activities in this VSN involve transmission of text,
 voice and video
      this implies QoS (Quality of Services) needs to be provided
 (3) transmission occurs both uplink (e.g., Maria uploading photos) and
 downlink (e.g., David downloading photos);
 (4) there are both big vehicles such as coaches and small vehicles such
 as personal cars;
 (5) David may use 3G network technology whereas Maria might take
 benefit of the existence of WiFi in her proximity.



                                                                         19




          VSN Network Architecture
(Referring to the VSN scenario diagram earlier)
This hybrid vehicular network architecture embraces
multiple wireless technologies (both WiFi and WiMAX) and
supports both
  BS-based communications for quick access to the outside Internet
  and ad-hoc communication amongst vehicles to facilitate the
  exchange of content generated by vehicles themselves.


Here only the PMP (point to multiple point, i.e., WiMAX BS
to multiple SSs) mode is employed.
The ad hoc or mesh connection amongst vehicles is
conducted by WiFi.

                                                                         20




                                                                               10
     Overall System Architecture




                                          21




                 Agenda

Overview
A Socially-aware Mobility Model for
Delay-tolerant Mobile Social Networks
Vehicular Social Network
  Overall Architecture
  Centrality-aware Bandwidth Allocation
Q&A


                                          22




                                               11
                                 Aims

        Aims to making some preliminary exploration into this
        exciting field of vehicular social networks by
        investigating into how social centrality, an important
        concept in social network analysis, makes impact on
        dynamic bandwidth allocation (DBA).
        In particular the paper introduces degree centrality into
        the utility function’s formula, based on which DBA is
        carried out in IEEE 802.16j-enabled vehicular networks.




                                                                            23




                      Social Centrality
   Centrality is widely used by social network analysis to describe
   relations among individuals and groups, aiming to identifying the most
   important actors.
   An actor is considered to be central if the ties it has with the other actors
   in the network concerned make the actor more visible or prominent
   than the others in the network.
                                      Network graph illustrating centrality




a: all nodes are equally central (as they are equally interchangeable)
b: Node A completely outranks the others,
c: centrality of A and F (two end nodes) is smaller than the other nodes. 24




                                                                                   12
    Centrality in Vehicular Networks?

Looking into the above vehicular network where there are
both large coaches and small cars on the road, all
communicating to the Internet, a coach usually have more
passengers and possesses more resources than a car.
   Having more passengers means there are more chances of the coach
   either providing information or consuming information, i.e., more
   likely to become more prominent social roles.
   Possessing more physical resources means, from vehicular network
   operation’s perspective, these coaches shall be encouraged to carry
   out more relaying.
Both demonstrate that coaches tend to have more ties with
other vehicles (especially cars which want to use coaches to
relay their traffic). Namely, coaches will exhibit higher
centrality.
                                                                           25




                   Degree Centrality
There are various different types of measures for centralities, such as
degree centrality, closeness centrality, betweenness centrality, etc.
Here degree centrality is used to express the popularity of a vehicle in
terms of relaying as it is intuitive and natural that a coach that has more
links (i.e., degrees) with other vehicles is more active or central.
The centrality of each node (e.g. SS or RS) is measured by the
standardized actor degree index as follows:




where xij = 1 if there is a link between nodes i and j; otherwise, xij = 0. ni
denotes node i. N is the total number of nodes under the coverage of a
related roadside BS.


                                                                           26




                                                                                 13
                                                    Utility Function
As far as the communication demand is concerned, there is an
intention to allocate more bandwidth to “popular nodes”.
However, centrality is not the only factor affecting the
bandwidth allocation.
There are other two major factors that make impact on
bandwidth allocation, i.e.,
           the requested bandwidth from a particular node and
           the types of services a request is composed of.
Combining all these factors, the utility function is defined as
follows: (i: node ID; k: type of service)




                                                                                                        27




 Relationships among utility, allocated
       bandwidth and centrality


                   1



                  0.8
  Utility Value




                  0.6



                  0.4



                  0.2



                   0
                   1
                         0.8
                                                                                         Requested Bandwidth
                                  0.6

                                              0.4
                                                     0.2

                                                           0   0
                        Centrality of nodei                        Allocated Bandwidth


                                                                                                        28




                                                                                                               14
                                                                    DBA Algorithms

                                    The objective of the proposed DBA algorithm is to
                                    allocate bandwidth to different types of services of each
                                    node with QoS considerations from the BS.
                                    This is carried out by maximizing the total utility of the
                                    overall network as defined below:




                                                                                                                                                                                                                  29




                                    Evaluation Results - Throughput
                            Throughput of different types of services under different traffic load                                               Throughput of different types of services under different centralities
                     45                                                                                                              30
                            rtPS                                                                                                                 rtPS
                            nrtPS                      Saturation                                                                                nrtPS
                     40     BE                                                                                                                   BE
                                                                                                                                     25
                     35
Throughput (M bps)




                                                                                                           T h ro u g h p u t (M b p s )




                     30                                                                                                              20


                     25

                                                                                                                                     15
                     20


                     15
                                                                                                                                     10
                                                                                                                                                             Very preliminary
                     10

                                                                                                                                           5
                     5


                     0
                      10       20      30       40         50       60     70       80       90      100                                   0
                                                                                                                                           0.1     0.2      0.3      0.4      0.5      0.6      0.7      0.8      0.9     1
                                                      Number of SSs
                                                                                                                                                                           Centrality of RS

                          The throughput for different service types changes as the average centrality of
                          RS increases under a fixed traffic load (roughly with 65 SSs – slightly above
                          the number of SSs at the saturation point, which is 60 in the first figure).
                          As more SSs are connecting to, i.e., relayed by, RSs (as indicated by an
                          increased value of RS centrality), the throughput for any type of service
                          increases. (but only slightly). In particular, the rtPS service has bigger   30
                          throughput increase.




                                                                                                                                                                                                                              15
                                                         Evaluation Results – Average Delay
                                          Average Queuing Delay of different Types of Services under different traffic load
                                          5                                                                                                                  Average Queuing Delay of different Types of Services under different centralities
                                     10                                                                                                                      50
                                                 BE                                                                                                                 BE
                                                 nrtPS                                                                                                              nrtPS
                                                 rtPS                                                                                                        45
                                          4                                                                                                                         rtPS
                                     10
                                                                                                                                                             40




                                                                                                                              Average Q ueuing Delay (m s)
A v e ra ge Q u e u in g D ela y (m s )




                                                                          Saturation
                                          3
                                     10                                                                                                                      35


                                                                                                                                                             30
                                          2
                                     10
                                                                                                                                                             25

                                          1
                                     10                                                                                                                      20


                                                                                                                                                             15
                                          0
                                     10
                                                                                                                                                             10

                                          -1
                                     10                                                                                                                      5


                                                                                                                                                             0
                                     10
                                          -2                                                                                                                 0.1       0.2     0.3     0.4      0.5     0.6      0.7      0.8     0.9       1
                                          10        20      30      40         50      60     70       80       90      100                                                                  centrality of RS
                                                                          Number of SSs

                                               However, this delay reduction is not significant
                                                  Possibly because it is compromised by one more hop to the RS on the way
                                                  to the BS in comparison to direct transmission to the BS??

                                                                         More investigations are needed.                                                                                                                        31




                                                                                               Contact,Q&A


                                               Dr Kun Yang, Reader
                                               School of Computer Science & Elec. Eng. (CSEE),
                                               University of Essex, Wivenhoe Park, Colchester,
                                               CO4 3SQ, UK
                                                 Email: kunyang@essex.ac.uk
                                                 http://privatewww.essex.ac.uk/~kunyang/




                                                                                                                                                                                                                                32




                                                                                                                                                                                                                                                 16

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:4
posted:7/30/2012
language:English
pages:16