Prognostic of feature interactions between independently developed pervasive systems

Abstract

Statistics show an aging trend in the world population, which will progressively overload existing health systems. Therefore, we believe that ubiquitous computing will play an important role in domicile settings, coping with the growing need for automated home healthcare support, especially for the sick and elderly. The integration of independently developed off-the-shelf systems (e.g., health-monitoring, entertainment, communications, home automation, etc.) may cause unplanned interactions between them (cf. feature interactions). This is a major concern since the correct/expected behavior of an isolated system may not be the same when deployed in conjunction with other systems, causing interferences, i.e., unexpected outcomes or misbehaviors. The Safe Home Care project tackles this problem to pursuit the safe deployment and reconfiguration of home healthcare smart-spaces. We propose the use of state graphs to represent off-the-shelf systems and predict the occurrence of intra-system's feature interactions. We use pre-deployment simulations to forecast feature interactions before deployment. We assess the applicability and correctness of this approach through a set of simulated home assisted living scenarios.

Interference Free Integration of Pervasive Applications

“…FI Problem). The intrasystem approach has three steps (see Figure 2) [7]: i) classify -creates a graph with observed OTS system states; ii) pruning -based on the knowledge of expected state, identifies unexpected states; and iii) reasoning -identifies unexpected behavior or malfunctions using the pruned results. For example, assume that the current GoOS sequence is <S 3 ,S 6 ,S 5 ,S 7 ,S 1 ,S 2 > (see Figure 1b).…”

“…We have applied our FI detection approaches (see Section II) to different scenarios, equipped with diverse systems (see Table III). For each scenario we identify possible FI outcomes (see [7] for more description of the scenarios).…”

Graph-based approach for interference free integration of pervasive applications

“…In our previous papers, we present a 3D simulation framework [1], which will be used to generate system state data; and we propose a first approach to represent and analyze in realtime the behavior of an OTS system using graphs [2]. In this paper, we extend this approach for detecting unwanted behavior in home care introducing relations between independent systems during their execution through a media (i.e., akin to physical, user, time, history and location ubicomp's context).…”