我們利用傳感器來幫助我們監視我們周圍的環境事件。
在智能交通系統，智能家居，或天氣預測的情況下，
同質或異質傳感器被部署到不斷檢測
環境。然後將收集的傳感器讀數可以告訴我們，可能還需要進一步調查的任何事件。

傳感器網絡具有高風險的丟失數據的問題。它也面臨著節電的問題。為了節省功耗，我們經常喜歡用盡可能少的傳感器盡可能和傳感器可以在作有限的時間盡可能
同時保持來自傳感器的相同或相似的服務的性能。這種機制也將導致丟失數據的問題，因為我們會想念讀數值由一些傳感器。在這項工作中，我們建議，可以使用傳感器讀數的一小部分和傳感器讀數中未收集到的其餘部分的機構
可以從可用的傳感器讀數進行推算。

我們採用高斯過程回歸的預測模型。
一鍵具有給定的傳感器讀數數據的有效高斯過程預測
依賴於我們如何找到一個合適的核函數的過程。
更具體地，由於具有時間和空間關係的信息的傳感器數據，
我們提出了一個可整合不同的關係作為一種策略
我們可以成功地描述了不同的傳感器讀數之間的關係
對於閱讀的預測。更進一步，我們提出了一個各向異性核函數的高斯過程，對傳感器讀數造型空間和時間之間的關係平衡。

實驗評估是基於對天氣數據為例進行
這包括收集在台灣的溫度讀數。
實驗結果表明，提出的高斯過程
各向異性核函數來描述不同的傳感器讀數之間的時空關係，的確可以給我們很好的歸集。

We utilize sensors to help us monitor events in the environment around us.
In the cases of intelligent transportation systems, smart homes, or weather prediction,
homogeneous or heterogeneous sensors are deployed to constantly sense
the environment. Then the collected sensor readings can tell us for any events that may need further investigation.

Sensor networks have high risk in missing data issue. It is also confronted with power saving problem. To save power consumption, we often prefer to use as few sensors as possible and the sensors can be on for as limited time as possible
while keeping the same or similar service performance from the sensors. This mechanism will also lead to missing data issue, because we will miss reading value from some sensors. In this work, we propose a mechanism that can use a small subset of sensor readings and the rest of sensor readings that are not collected
can be imputed from the available sensor readings.

We adopt Gaussian process regression as the prediction model.
One key to have an effective Gaussian process prediction given sensor readings data
relies on how we find an appropriate kernel function for the process.
More specifically, given sensor data that have spatial and temporal relationship information,
we propose a strategy that can integrate different relationships as one
and we can successfully describe the relationship between different sensor readings
for the reading prediction. Further more, we propose an anisotropic kernel function for Gaussian process that balances between spatial and temporal relationships for sensor reading modeling.

The experiments for evaluation are conducted based on a case study on weather data
that consist of temperature readings collected in Taiwan.
The experiment results show that the proposed Gaussian process with
anisotropic kernel function to describe the spatio-temporal relationship between different sensor readings can indeed give us good imputation.

[1] G. Li, Z. Xu, C. Xiong, C. Yang, S. Zhang, Y. Chen, and S. Xu, “Energy-efficient wireless communications:
tutorial, survey, and open issues,” Wireless Communications, IEEE, vol. 18, pp. 28–35,
December 2011.
[2] Y. Li and L. E. Parker, “Nearest neighbor imputation using spatial–temporal correlations in wireless
sensor networks,” Information Fusion, vol. 15, no. 0, pp. 64 – 79, 2014. Special Issue: Resource
Constrained Networks.
[3] A. Manjhi, S. Nath, and P. B. Gibbons, “Tributaries and deltas: Efficient and robust aggregation
in sensor network streams,” in Proceedings of the 2005 ACM SIGMOD international conference on
Management of data, pp. 287–298, ACM, 2005.
[4] H. Jiang, S. Jin, and C. Wang, “Prediction or not? an energy-efficient framework for clustering-based
data collection in wireless sensor networks,” Parallel and Distributed Systems, IEEE Transactions on,
vol. 22, pp. 1064–1071, June 2011.
[5] J. Chen and J. Shao, “Jackknife variance estimation for nearest-neighbor imputation,” Journal of the
American Statistical Association, vol. 96, no. 453, pp. 260–269, 2001.
[6] E. R. Hruschka Jr, E. R. Hruschka, and N. F. Ebecken, “Bayesian networks for imputation in classification
problems,” Journal of Intelligent Information Systems, vol. 29, no. 3, pp. 231–252, 2007.
[7] J. Fox, Applied Regression Analysis and Generalized Linear Models. SAGE Publications, 2008.
[8] R. J. Little and D. B. Rubin, “Statistical analysis with missing data,” 2002.
[9] C. E. Rasmussen, “Gaussian processes for machine learning,” MIT Press, 2006.
[10] J. Yan, C. Tian, J. Huang, and Y. Wang, “Load forecasting using twin gaussian process model,” in
Service Operations and Logistics, and Informatics (SOLI), 2012 IEEE International Conference on,
pp. 36–41, July 2012.
[11] M. Girolami, “Mercer kernel-based clustering in feature space,” Neural Networks, IEEE Transactions
on, vol. 13, pp. 780–784, May 2002.
[12] V. N. Vapnik, The Nature of Statistical Learning Theory. Springer, 2 ed., November 1999.
[13] C. J. C. Burges, “A tutorial on support vector machines for pattern recognition,” Data Mining and
Knowledge Discovery, vol. 2, no. 2, pp. 121–167, 1998.