Gen Z Social Networks using Collaborative Spatial Alarms

advertisement
Gen Z Social Networks using Collaborative Spatial Alarms
Kinshuk Mishra, Vivek Raju
{kinshuk, vivek.mupalla}@cc.gatech.edu
1> Motivation and Objectives:
Social Networking is going wireless. Existing web based services like MySpace and
Orkut are being extended for use over handheld devices. This is a fast growing market
and considering some surveys by companies like M: Metrics show that 45% of the
internet users have at some point in time used a social networking site and these users
welcoming the feature in handheld devices is reason enough for us to invest time into this
field of study.
We are keen to implement a system which alerts all the people, connected in a social
network and who use a hand-held device, based on their proximity to certain location of
interest. This is where the motivation of using location based services (LBS) to introduce
the concept of “Spatial Alarms” arises from.
So what is a spatial alarm? Now it is regular to say pick up laundry at 6:00 pm as an
alarm setting, a spatial alarm allows us to say pick up laundry from the store when we
pass by the store when we are in x miles of radius. Now it not just time dependent but
extended to space too.
The above example is one of the simplest ways of implementing the idea and a very naïve
form of alarm. The idea of these spatial alarms can be extended to collaborative behavior
amongst peers in a network in various situations, for example:
• Find a place on campus location where free food is available leave an alarm your
friend when passing by will be notified.
• Like a restaurant leave a review/rating alarm for someone.
• Tourism can have a booming advantage with this new e-tour guide with inputs
from locals.
Our system looks at implementing such a collaborative spatial alarm based system to
build the next generation of social networks.
2> Related work:
The proposed works brings to the research in the field of spatial alarms to a new
dimension. Currently there exists no published work in this area. The only base that is
used by our system is from “Spatial alarms, Anand Murugappan and Sarang Karandikar
2006”. We propose to extend the work done by Murugappan and Karandikar to build a
collaborative system and propose changes to exiting alarm detection methods in order to
perform a comparative study.
3> Proposed work:
We propose to add collaboration as a feature for the spatial alarms by introducing a TAG
based system and user profiling system.
Figure 1 : Prototype Architecture
Map Rendering
Module
Position Locater Module
Social Network
Module
MapPoint
Server
GPS
server
Social
Network
Server
Alarm Trigger
Figure 1 shows the prototype architecture of the proposed system. The main components
of the work can be explained as follows:
Social Network Server:
The Social Network Server consists of mainly three components:
Network Module: This takes care of connections with the mobile hosts and is responsible
for the data communication with the clients and other parts of the server.
Host Module: This takes care of all the computation related to alarms, like storage
maintaining tags and other related work. The host module also takes care of aspects like
the wakeup time for the client.
Expert Module: This is the learning module which is used for user profiling and learning
from past behavior.
A new alarm can be set by a client by sending a packet over the network as following:
Type
Size
Data
Location
TAG
User Profile
Group ID
Where Type is for type of packet e.g. Type = 0 implies it is packet to set an alarm, size is
the packet size. Location contains the location information in coordinates. TAG is a
feature which will allow user to filter results on search, reducing spamming of alarms,
TAG = ‘Restaurant’ or TAG = ‘Publix’ searches only for the relevant data.
User Profile is a unique system override feature provided for group members say A and B
are ‘Metallica’ Fans A has set an alarm for a concert at place X, B does not have a clue
about this and is only looking for TAG = Books in the area however due to matching
profiles a system override take place to send out this information too. The user B then
sends a packet saying if the information is useful or not allowing the learning engine to
store the case for future behavior.
Group ID is again a feature so that all people of a particular group (extending the
community concept from Social Networks) can share alarms.
An alarm check usually takes place by sending the packet:
Type
Size
Location
TAG
Here Location and TAG are used to return the information. The server typically replies
with a
Type Size
Alarm Next Check
Here Alarm is the relevant alarm which is set to NULL if the there is no alarm and Next
Check is a server based calculation based on client data an optimized energy conservation
mechanism.
Figure 2 gives an example of dataflow for a typical scenario in our proposed system.
The other components for the system are the GPS server for LBS and the Map Point
Server for rendering the maps along with the client devices.
Figure 2: Example Data Flow.
Client A
Client B
Set Alarm
Client C
Server
Check Alarm
No Alarm Check After T=N
Check Alarm
Return Alarm
Check Alarm after N
4>Plan of Action:
a> Schedule
Week
1
2
3
4
5
6
7
8
9
10
11
Milestones
Analysis of Data Structures For Spatial indexing
Continuous Query Analysis Technique Study
Design Optimizations
Building the Network Module & Host Module
Building the Expert System Module & Host Module
Integrating the Expert System Module in Server
Integrating Map point & GPS servers
Testing
Testing
Performance Analysis
Extendables
b> Software to Be Used:
We will be using the MapPoint Web services by Microsoft for fetching the maps and
rendering them. C# would be used for the map display functionality over the .NET
compact framework.
Our Social Network Server would be implementing the logic and simulations in java. The
network connection modules within the client will also be implemented in java.
c> Hardware:
We will be using the emulator for initial testing and simulations. If time permits we plan
to test it on Hand-held devices too.
5> Evaluation and Testing:
The most significant testing and analysis will be based on optimization of the various
software modules and their place in the architecture i.e. client side or server side.
The performance of the system will be based on three major parameters:
1. The efficiency of setting/ triggering an alarm.
2. The ease of use.
3. Performance based on the power consumption.
6> Bibliography
[1] The TPR* - Tree : Yufei Tao, Dimitris Papadias, Jimeng Sun
[2] Spatial Alarms – Anand Murugappan and Sarang Karandikar
[3] MapPoint - msdn.microsoft.com/mappoint/
Download