Show Less

Application Coordination in Pervasive Systems


Verena Majuntke

Pervasive applications are designed to support users in their daily life. For this purpose, applications interact with their environment, i.e. their context. They are able to adapt themselves to context changes or to explicitly change the context via actuators. If multiple applications are executed in the same context, interferences are likely to occur. To manage interferences, a coordination framework is presented in this thesis. Interferences are detected using a context model and information about applications’ interaction with the context. The resolution of interference is achieved through a coordinated application adaptation. The thesis introduces the theoretical concepts, presents a prototypical implementation and evaluates the prototype through extensive measurements.


Show Summary Details
Restricted access

8. Evaluation


This chapter evaluates the prototype which was presented in Chapter 7. The goal of this chapter is to assess the concepts which were developed in this thesis and to show the utilizability of the coordinator for practical pervasive systems. For this purpose, Section 8.1 first discusses the memory requirements of the coordinator in detail in dependence on the number of active applications and their context configurations. Subsequently, Section 8.2 evaluates the prototype with respect to its performance. At first, the critical path of application coordination is analyzed. Afterwards, measurements for the implemented interference detection and interference resolution algorithms are conducted and the results are discussed. 8.1. Memory Requirements and Overhead The first evaluation addresses the memory requirements of the coordinator and the over- head it causes. Firstly, the memory requirements of the classes that compose the coor- dinator are determined in dependence of active and alternative context configurations. Secondly, conceivable examples of context configurations and the resulting load for the coordinator are discussed. Moreover, the resulting load is set in relation to the memory requirements of the middleware BASE to determine the overhead. Finally, the message overhead that is caused through the use of the coordinator in a pervasive system realized with BASE is analyzed. In order to determine the memory requirements, a footprint of the prototype has been measured using the Java profiler Java VisualVM which is part of JDK. The classes that compose the prototype and their memory requirements are shown in Table 8.1. The table 134 8....

You are not authenticated to view the full text of this chapter or article.

This site requires a subscription or purchase to access the full text of books or journals.

Do you have any questions? Contact us.

Or login to access all content.