Transcript 下載/瀏覽
Optimizing Energy-Efficient Query Processing in Wireless Sensor Networks Ross Rosemark Wang-Chien Lee Bhuvan Urgaonkar Department of Computer Science and Engineering The Pennsylvania State University 任課老師:陳朝鈞 報告者:邱志銘、黃烱育 2016/7/13 1 Outline Introduction Query Optimization Performance Evaluation Conclusion 2016/7/13 2 Introduction Energy-efficient query processing for wireless sensor networks (WSNs) Choosing an efficient query plan Two primary contribute to the energy consumption of WSNs Sensing cost Communication cost 2016/7/13 3 Introduction Existing query optimizers for WSNs typically Sensing cost In this paper Query execution Including routing Sensing data/metadata collection 2016/7/13 4 Introduction data/metadata In query optimizers for database management systems 2016/7/13 Query optimization Remove the wireless sensor network data 5 Query Optimization Provides a highlevel overview of our query optimizer. Classification Stage Refinement Stage 2016/7/13 6 Query Optimization Preferences Q:a single query p:predicates n:node s:sensor m:total number of query plans generated 2016/7/13 7 Query Optimization Classification Stage Determine if metadata should be collected by evaluating each query plan within . Based on , the query optimizer chooses QPmin 2016/7/13 no metadata collection metadata collection 8 Query Optimization No metadata collection Finding a better query plan. Metadata collection Query optimizer enters the second stage 2016/7/13 9 Query Optimization Refinement Stage Collect metadata to re-evaluate each query plan in distributed to each node choosing Routes back the desired tuples to the AP 2016/7/13 10 PERFORMANCE EVALUATION Impact its total energy consumption The number of reportings The unit energy cost of each sensor sampling 2016/7/13 11 PERFORMANCE EVALUATION Number of Reports 2016/7/13 12 PERFORMANCE EVALUATION Unit Cost of Sensor Sampling 2016/7/13 13 PERFORMANCE EVALUATION Effectiveness of Query Optimizer query optimizer chooses most energy-efficient query evaluate over a thousand query plans 2016/7/13 14 PERFORMANCE EVALUATION Light Query 2016/7/13 15 PERFORMANCE EVALUATION Heavy Query 2016/7/13 16 PERFORMANCE EVALUATION Randomly Generated Queries Optimization on Communication Cost 2016/7/13 17 CONCLUSION In this paper they examine the design of a query optimizer for WSNs. The approach optimizes on routing、sensing and data/metadata collection. The approach can more accurately estimated query plans in potential gain. In the future they plan to extend our approach to support multiple queries simultaneously. 2016/7/13 18