Generic benchmarking for application specific wireless sensor networks multi criteria performance

Dong Yu · UTS ePRESS (University of Technology Sydney) · 2012

Due to stringent energy constraint and demand for performance requirement, a generic architecture like TCP/IP or Internet is not feasible with sensors used across various applications.Instead, application specific design methodology is the de facto consensus accepted among Wireless Sensor Network (WSN) community.While it wins WSN performance gains for individual applications, the methodology sacrifices all plausible attributes a generic architecture can contribute.Without a unified reference model as comparing foundation, the profound problem in true protocols contribution evaluation and comparison remains challenging.Moreover, the stochastic and statistical nature of WSNs makes realistic performance analysis fairly complex.In multi criteria QoS context, this problem is further magnified by big design space with not yet fully understood parameters and the competing relationship between multi objective performance metrics.This work introduces a generic wireless-benchmarking methodology not only qualitatively evaluation from high level abstraction, concerning only profound pros and cons from a general viewpoint of tradeoffs between generality, performance and cost, but also a set of practical workflows that are designed to support quantitative evaluation and analysis of WSN protocols for application-specific objectives.This methodology and the accompanying new benchmark concepts, such as performance efficiency, development efficiency and performance stability, are designed to gain new insight of the dynamic behavior of WSN protocols in a systematical way vi compared to the current ad-hoc evaluation approaches applied by most of the community.vii

Read the paper · More papers on PaperTik