Application modeling for performance evaluation on event-triggered wireless sensor networks

• Published in: Design automation for embedded systems Vol. 20; no. 4; pp. 269 - 287
• Main Authors: Brisolara, Lisane, Ferreira, Paulo R., Indrusiak, Leandro Soares
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.12.2016; Springer Nature B.V
• Subjects: CAE) and Design; Circuits and Systems; Computer simulation; Computer-Aided Engineering (CAD; Configuration management; Engineering; Load balancing; Performance evaluation; Remote sensors; Special Purpose and Application-Based Systems; Wireless sensor networks; Workload; Load; Workload characterization; Sensor networks; Load balancing; High-level simulation; Application; Modeling
• ISSN: 0929-5585; 1572-8080
• DOI: 10.1007/s10617-016-9177-1

This paper presents an approach for event-triggered wireless sensor network (WSN) application modeling, aiming to evaluate the performance of WSN configurations with regards to metrics that are meaningful to specific application domains and respective end-users. It combines application, environment-generated workload and computing/communication infrastructure within a high-level modeling simulation framework, and includes modeling primitives to represent different kind of events based on different probabilities distributions. Such primitives help end-users to characterize their application workload to capture realistic scenarios. This characterization allows the performance evaluation of specific WSN configurations, including dynamic management techniques as load balancing. Extensive experimental work shows that the proposed approach is effective in verifying whether a given WSN configuration can fulfill non-functional application requirements, such as identifying the application behavior that can lead a WSN to a break point after which it cannot further maintain these requirements. Furthermore, through these experiments, we discuss the impact of different distribution probabilities to model temporal and spatial aspects of the workload on WSNs performance, considering the adoption of dynamic and decentralized load balancing approaches.