Abstract
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.
(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.
| Original language | English |
|---|---|
| Pages (from-to) | 269-287 |
| Number of pages | 19 |
| Journal | Design Automation for Embedded Systems |
| Volume | 20 |
| Issue number | 4 |
| DOIs | |
| Publication status | Published - 11 Aug 2016 |