People

Zola Mahlaza
Jarvis Mutakha

Supervisors

Professor E. Blake
Thomas Reitmaier
Pierre Benz

Components

Cloudlet platform
Android application

Downloads

Jarvis' Lit eview
Jarvis' Report
Zola's Lit review
Zola's Report
Research Proposal
Code



Home

Android application


The cloud is generally used for data storage and service provision. Cloudlets can also be used to provide the mentioned things. The provision of a large number of cloud services in a cloudlet scenario would be beneficial, however, for this project a single service will be offered due to time constraints. There are numerous aspects the current cloud computing paradigm is lacking in, these limitations carry over to cloudlets. The following are the motivational factors of the project.

  1. Information Ownership, Control and Security: Data centres for cloud computing service providers can be located in any country in the world. Users may be concerned about where their information is stored. Users may want the ability to control where their data is stored and who has access to it. This may be due to privacy laws and government organizations in different countries.
  2. Cellular networks: Connecting to the Internet via cell phone networks is generally slower than connecting to local WiFi networks. They also introduce high charges from cell phone providers in some cases. These two factors therefore limit the use of cloud services by users of mobile devices.

Research Questions

The project focusses on the following two research questions:
  1. Can we create an effective common sandbox for data?
  2. What interface conceptual metaphors are effect in conveying the properties of the cloudlets as ephemeral data stores?

Implementation

An application was developed that allowed file sharing between Android devices connected to the cloudlet.

Evaluation

This investigation set out to evaluate the feasibility of a Cloudlet running on the Raspberry Pi to allow co-located file sharing between users with smartphones. An experimental design process was used which allowed for incremental improvements on the system and its interface after each phase. Here is a general overview of the design process that occurred:

  1. Design of initial paper prototypes (P1);
  2. Heuristic evaluation of those paper prototypes (using P1);
  3. Re-design of the paper prototypes (P2);
  4. Carrying out participant observations with the paper prototypes (using P2);
  5. Re-design of the paper prototypes to get final paper prototypes (P3);
  6. Implementing the system (I1 using P3);
  7. Carrying out the first HCI expert evaluation on the implementation (using I1);
  8. Re-design of the implementation (I2);
  9. Carrying out a second HCI expert evaluation (using I2)(It should be noted at this point a third implementation (I3) couldn’t be developed due to time constraints).
  10. Carrying out usability testing (using I2)
The key research questions included:
  1. Determining if a common sandbox (public digital table/shelve) for ephemeral data (i.e. data that’s not permanently on the cloudlet) could be implemented and used in a practical setting. This was a success but it could be investigated further because although it was implemented on the Android OS and tested on Samsung Galaxy Pockets, to fully test the practicality of a common sandbox, the application would have to be implemented on other mobile devices running on different software (e.g. having a cloudlet with Apple, Android and BlackBerry devices all sharing files). However, implementing it on different mobile operating systems was out of the projects scope.
  2. Determining if there is an effective interface for conveying the cloudlet and the ephemeral nature of the file shared. This was fairly successful because most users understood the functions of the cloudlet immediately after completing a few short tasks on the cloudlet in one session. However, the final HCI expert evaluation pointed out ways in which the interface could be improved slightly to convey the conceptualisation better. The Cloudlet app was met with fair enthusiasm with the user groups and there was genuine interest in using the Cloudlet beyond the scope of the project. The success can be attributed to carrying out evaluations and getting feedback along the design process which led to revamping the application a few times.
An interesting lesson learned during this project was the effectiveness of using different ways to evaluate the system. This allowed for flaws to be pointed out which were missed by other methods on previous phases. For example, users didn’t notice that the conceptualisation of the cloudlet (namely, the conference table) may have been difficult to get in the first participant observation however this issue was brought to the light right after in the first expert evaluation. third implementation (I3) couldn’t be developed due to time constraints).



- Jarvis Mutakha