Policy-Based Management for Resource-Constrained Devices and Systems

Abstract

The presented policy-based management system supports autonomous control and adaptation of a distributed system according to changing conditions and requirements by means of event-condition-action (ECA) rules. Furthermore, it supports policy-aware application programming. Application components can request evaluations of policy expressions and decisions in order to govern their behavior depending on global system state and environment conditions. That rich functionality has to be provided very efficiently since the distributed system consists of resource-constrained devices. The model-based management (MBM) approach is applied separating comfortable tool-assisted policy definition and refinement at design time from lightweight runtime policy enforcement. New enhancements of MBM extend the policy refinement to the derivation of ECA rules. The new backend functions generate executable Java byte code for policy expressions and decisions as well as for policy rules. The code is appropriately partitioned and allocated to the devices. A simplified healthcare scenario demonstrates the approach and its application.