Interdroid: a platform for distributed smartphone applications

1
Interdroid a platform for distributed smartphone
applications

• Henri Bal, Nick Palmer,Roelof Kemp, Thilo
  Kielmann
• High Performance Distributed Computing Group
• Department of Computer Science
• VU University Amsterdam
• bal_at_cs.vu.nl

2
Ibis framework for distributed applications

• Clusters, grids, clouds Thilo Kielmann
• Accelerators (GPUs) Rob van Nieuwpoort
• Heterogeneity (Jungle Computing) Frank
  Seinstra
• Clouds mobile devices Henri Bal

3
Computing Trends

• Mobile
• Personalization of Computing Devices
• Cloud
• Commoditization of Computation and Storage

4
Made For Each Other?
Mobile Cloud
Processor? Small Unlimited
Storage? Small Unlimited
Network? Constant Change! Constant
Power? Limited Battery Unlimited
Sensors? Lots! None
Personal? Very Not At All
5
Challenges

• Data Management
• Caching and Synchronization
• Mobile Resource Constraints
• Battery, Processor,Memory, Storage
• Making Smartphones Really Smart

6
Interdroid Platform
Novel Mobile Cloud Applications

Raven
SWAN
Cuckoo
Computation Communication Offloading
Mobile Sensing
Distributed Data Management
Distributed Communication Cloud Deployment
Android OS
Cloud Systems
7
Raven

• Framework for mobile management of shared
  versioned data collections
• Peer to Peer Sync On Ibis
• Mobile - Cloud Sync
• Commit and Branch Support
• Run Compile Time DB Definition
• List, View, and Edit UI Generation

8
SWANSensing With Android Nodes

• Framework for mobile sense applications
• Hardware, External, Fusion Network Sensors
• Data Storage Using Raven
• Domain Specific Expression Language
• soundlevel (ALL,1m) gt 3 gt call dad

9
Cuckoo - offloading

• Computation Offloading
• Select between Local or Cloud Execution of
  Workloads
• Communication Offloading
• Push Notification for Anything

10
Computation Offloading Framework

• Runs on Android, integrates with Eclipse
• Multiple implementations of compute intensive
  parts
• Remote and local implementation bundled together
• Deals with network connectivity issues (Ibis
  SmartSockets)

11
Computation Offloading
Remote
Activity
Stub
Proxy
Local
12
eyeDentify object recognition

• Compute intensive operation
• Local implementation is inaccurate
• Remote implementation runs in parallel and uses
  better accuracy parameters
• Results
• Up to 40 times less energy
• Up to 60 times faster
• better recognition quality

13
Communication Intensive Apps

• Applications and widgets that
• communicate periodically
• RSS Readers, Weather, Traffic,Social Networks,
  Sports Scores,News, Stock Market, etc.
• built by 3rd party developers
• Do Pull-based Web Requests
• Information Monitoring Applications

14
Communication offloading

• No Update Delays
• No Energy Waste
• No Server Modifications
• Works for 3rd party devs

15
Example App Web Page Widget (1/2)
16
Example App Web Page Widget (2/2)
17
More information

• http//interdroid.net/
• Ph.D. thesis Nick Palmer (2012)
• Papers
• IEEE Computer Aug. 2010 (Ibis),MDM 2011,
  MobiCASE 2010 2011