Energy‐efficient and localized lossy data aggregation in asynchronous sensor networks‡. (2nd February 2012)

• Record Type: Journal Article
• Authors: Zhang, J.; Shen, X.; Zeng, H.; Dai, G.; Bo, C.; Chen, F.; Lv, C.
• Is Part Of: International journal of communication systems. Volume 26:Number 8(2013:Aug.)
• Page Start: 989
• Page End: 1010
• Publication Date: 2012-02-02
• Subjects: lossy data aggregation -- energy efficiency -- data sampling -- link unreliability -- wireless sensor networks; Telecommunication systems -- Periodicals; 621.382
• DOI: 10.1002/dac.1384
• Languages: English
• ISSNs: 1074-5351

Abstract

SUMMARY: In wireless sensor networks, most data aggregation scheduling methods let all nodes aggregate data in every time instance. It is not energy efficient and practical because of link unreliability and data redundancy. This paper proposes a lossy data aggregation (LDA) scheme to reduce traffic and save energy. LDA selects partial child nodes to sample data at partial time slots and allows estimated aggregation at parent nodes or a root in a network. We firstly consider that all nodes sample data synchronously and find that the error between the real value of a physical parameter and that measured by LDA is bounded respectively with and without link unreliability. Detailed analysis is given on error bound when a confidence level is previously assigned to the root by a newly designed algorithm. Thus, each parent can determine the minimum number of child nodes needed to achieve its assigned confidence level. We then analyze a probability to bound the error with a confidence level previously assigned to the root when all nodes sample data asynchronously. An algorithm then is designed to implement our data aggregation under asynchronization. Finally, we implement our experiment on the basis of real test‐beds to prove that the scheme can save more energy than an existing algorithm for node selection, Distributive Online Greedy (DOG). Copyright © 2012 John Wiley & Sons, Ltd. Abstract : This paper proposes a lossy data aggregation (LDA) scheme to reduce traffic and save energy by allowing partial child nodes to sample data at partial time slots and estimated aggregation at parent nodes or a root in a network under synchronization and asynchronization. Under both cases, the sample errors are bounded with a confidence level previously assigned to the root. We implement our experiment on real test‐beds to prove that LDA can save more energy than an existing protocol DOG.