Optimizing Component-Oriented Systems : A Case Study in Wireless Sensor Networks

In this demonstration we describe a systematic design process and a corresponding software system for automated planning, optimization and configuration of wireless networks according to user-specified scenarios and requirements. As an example we consider wireless sensor networks, but this methodology is also applicable to other kinds of networks. Using service-oriented approach the framework automatically “wires” components to generate a stack of components chosen from a wide variety of available software modules in order to maximally fulfill the user requirements. To reach a decision the system utilizes rule-based knowledge base reasoning, utility-based optimization, meta-heuristics and selflearning. After the network deployment the performance is further fine-tuned using a simplified version of the mechanisms employed at the pre-deployment stage. The runtime optimization includes both component-level parameter adjustment and cross-component manipulations. The presented system can be extended to optimize the performance of the network not only by choosing and configuring software modules, but also by altering the network topology.