Optimizing the Quality of Dynamic Context Subscriptions for Scarce Network Resources

Scalable access to dynamic context information is a key challenge for future context-sensitive systems. When increasing the access frequency, the information accuracy can improve but at the same time the additional context management traffic may reduce network performance, which creates the opposite effect on information reliability. In order to understand and control this trade-off, this paper develops a model that allows to calculate context reliability, captured by the so-called mismatch probability, in relation to the network load. The model is subsequently used for a real time algorithm for admission and configuration control for incoming context subscriptions. Two different variants of the algorithm are analyzed in simulation results while the basic version determines an optimal configuration only for a newly incoming context subscription, the second version also iterates over and potentially reconfigures already established context subscriptions. The results show the gain of the total mismatch probability when selecting the context access configuration intelligently.