A Energy-Efficient Precision-Constrained Data Aggregation Technique for Wireless Sensor Networks

• Main Authors: Cheng-Hsun Hsieh, 謝承勳
• Other Authors: Chin-Liang Wang
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/05283178790287845559

碩士 === 國立清華大學 === 通訊工程研究所 === 95 === Total energy consumption caused by data transmission and signal processing is always a major problem for the development of wireless sensor networks. In general, when sensor nodes are distributed densely, they will send all the sensing data to the processing center, and this may lead to packets contention or high loading for each relay node. However, some sensing data is not necessary for the precision of estima-tion made by the processing center; we call them the redundant data. If we can de-crease the redundant data appropriately, the total energy consumption can be effec-tively saved. In this thesis, we propose an energy-efficient architecture for wireless sensor networks. We first reduce the total energy consumption by a clustering algorithm that chooses the cluster head by vector quantization (VQ). In the VQ-based clustering al-gorithm, each cluster head is uniformly specified such that the distances between its members and itself are nearly the same. Furthermore, we use the precision-constrained data aggregation method to decrease the redundant data. Hence, we need to assign different error bound for each sensor node for adjusting the updating frequency, which is so-called precision allocation. To further alleviate the energy consumption of the training node for error bound estimation, we assign the error bound for each sensor node according to the number of hops. The proposed precision allocation method is easily realized in sensor nodes and the total energy consumption will be efficiently decreased. By using this architecture, we can save a large amount of energy consumption. Moreover, we can add a check construction in cluster heads; we check the averaged data in order to release the loading in cluster heads. Hence, the lifetime of networks will be further extended. Although this step may decrease the precision of estimated data slightly, we can decide if the system will use the check in cluster heads based on the application in practice. Computer simulation results show that as compared to the traditional architecture, the proposed scheme has both low-complexity in sensor nodes and high energy efficiency.