mobile cloud

18
Mobile Cloud and a particular framework for intelligent local storage Seminario di Sistemi Middleware A.A.2014/2015 Andrea Sghedoni

Upload: andrea-sghedoni

Post on 14-Aug-2015

23 views

Category:

Mobile


0 download

TRANSCRIPT

Page 1: Mobile Cloud

Mobile Cloud and a particular framework for intelligent

local storage

Seminario di Sistemi Middleware A.A.2014/2015 Andrea Sghedoni

Page 2: Mobile Cloud

● Interaction between Mobile Devices and Cloud● New Era of computing● LBS (Location Based Services)

Introduction

2/18

Page 3: Mobile Cloud

● Network Connection - TCP - 3G,4G● Context Awareness - Sensors

- few user input- automatic configuration- intelligent output

● Limited processing capability ● Limited storage capability

Mobile Clients - Features

3/18

Page 4: Mobile Cloud

● Storage off-device- devices have limited storage for SO, app, picture,

video, multimedia- sharing data with other applications- Trade-off between packets size and

requirement/network bandwith- Local or Cloud storage? → WhereStore

- latency factors(Connection, transfer time,..)

Interactions with cloud(1)

4/18

Page 5: Mobile Cloud

● Processing off-device- complex and intensive tasks - “black box”- push technology- asynchronous response - device can process other tasks while wait

response from cloud server- user wait time - Amazon(EC2, EMR), Google, FlexiScale

Interactions with cloud(2)

5/18

Page 6: Mobile Cloud

● Study of social behavior - Activity context● Weather app, navigator, BlaBlaCar,

traffic, integration with bluetooth/beacons in a market

● Social network app, Restaurant reviews, Meeting on LBS

Application and Social Contexts

6/18

Page 7: Mobile Cloud

● OpenID- OpenID Provider keeps your password in a secure way- Provider tells the websites/resources you’re visting that you are

who you say you are

● OpenAuth- token based- easy to invoke- limited services - server side

Cloud security

7/18

Page 8: Mobile Cloud

● Different viewpoint from SOAP ws● Perfect for integration between cloud and mobile

devices● Aspects:

- Stateless- URL Based- Response HTTP-based- Easy invocation method

RESTful Web Services (1)

8/18

Page 9: Mobile Cloud

● Response minimal and discrete● HTTP standard● REST responses very easy to understand and use it● Rest request

- HTTP verbs (GET,POST,HEAD,PUT,DELETE) - CRUD operations

● Event-driven-model for XML response- DOM more memory usage and processing

RESTful Web Services (2)

9/18

Page 10: Mobile Cloud

● Framework for location-based data store in mobile and cloud environments

● Users use a LBS mobile app in a particular location (different from desktop app)

● Main goals of WhereStore:- predict future smartphone location- what data replicate in local device storage- what data store in cloud- data available in periods of no connectivity

WhereStore

10/18

Page 11: Mobile Cloud

● In many app, presence of data is a big plus when there is no connection- Web apps - Media Content- Live Applications

Example Applications

11/18

Page 12: Mobile Cloud

● Unused space on device for cache data● What data to replicate at a given

time/location (no user input - automatism)

● What data provide in the future● Determine the optimal moment to

interact with cloud (home/Wifi)

Challenges

12/18

Page 13: Mobile Cloud

● Synchronization between different node● Collection contains items (data + metadata)● Replica → local subset or entire collection● Filter → particular subset of a collection that

should be in a replica ● Each replica keeps data that match its filter● Version of the items must be the same in all

replica

Replication System

13/18

Page 14: Mobile Cloud

● GPS sensor ● Monitoring of device location● History → prediction of future location● Difference between GPS measurements (morning/afternoon,

weekend/weekday, feast day/ common day ...) ● WhereStore creates and update filters

continuosly where:- (l1,l2,...,ln) → future locations- (p1,p2,...,pn) → probability

Location Prediction

14/18

Page 15: Mobile Cloud

● Items → piece of data with a particular priority● Groups → set of items● Regions → geographical area

Data types

work home shopping

books reviewsdocument movies

Regions

Groups

15/18

Page 16: Mobile Cloud

Sync with Cloud● App receives all the items that match with its filters, in a

precise moment● Problem when smartphone storage capacity is not enough

for all replica items: - for each item, cloud site compute a rank:

cj = pi * kjwhere pi → probability of filter

kj → priority of item● Only the top n items are send (n → capacity of device

storage), the others remain in cloud

16/18

Page 17: Mobile Cloud

● Cloud and RESTful WS provide solutions to limited mobile capability

● Reducing of user input● WhereStore, in particular, provide a set of

guideline for intelligent local storage and prediction

Conclusions

17/18

Page 18: Mobile Cloud

● http://openidexplained.com/, description of OpenID ● http://hueniverse.com/oauth/, description of OpenAuth ● J.H.Christensen,“Using RESTful Web-Services and Cloud

Computing to Create Next Generation Mobile Applications”, Orlando, Florida, Oct. 2009

● P.Stuedi, I.Mohomed, D.Terry, “WhereStore:Location-based Data Storage for Mobile Devices Interacting with the Cloud”, San Francisco, USA, June 2010

Bibliography

18/18