物聯網中基於能量感知之網路情境內容估測

Energy-aware Context Estimation for IoT Networks

陳彥賓(Yan-Bin Chen)

指導教授 ： 魏宏宇

國立臺灣大學/電機資訊學院/電機工程學研究所/博士(2016年)

https://doi.org/10.6342/NTU201603022

摘要

中文摘要
下世代網路 (例如：5G 網路) 正逐漸面臨一種需求，即需具備處理巨量資料能力與能源問題。巨量資料與能源問題在物聯網中極為重要。本博士論文包含兩則子研究內容，提供協助解決巨量資料與能源問題。當考量巨量資料與能源問題的可行性時，我們可將它具體化成兩個子研究內容，亦即內容情境估測和能量感知能力。首先我們專注的第一個子研究是內容情境的估測，而第二個則是具能量感知能力的網路最佳化。我們以目前較成熟的網路去模擬提議的解決方案架構，例如 IEEE 802.11 網路和無線感測網路 (wireless sensor network)。因為這些網路在網路演進到物聯網的過程中扮演著最重要的角色。

在第一個研究中，我們提出一個粒子過濾架構，用以實現立即性的動態估測。這估測使用在 IEEE 802.11 網路中無線終端機 (station) 的未飽和緩衝儲存器 (unsaturated buffer) 裏。採用本架構，存取點基地台 (access point) 可以對其下服務的無線終端機，動態地調整傳輸流量與組態相關的參數。所以增進了網路系統的資料傳輸率，也降低了封包的遞送延遲。目前在 IEEE 802.11 網路中分析未飽和條件的研究都是基於一種穩態模型 (steady-state model)，而我們提議的方法則是致力於對無線終端機的未飽和緩衝儲存器的機率分佈去做動態估測。這方法可用於同質或異質網路。本研究也採用了從貝氏推論 (Bayesian Inference) 一路到粒子過濾演算法 (particle-filtering algorithm) 的理論支持。用均方根誤差 (Root Mean Square Error) 和時間複雜度 (time complexity) 來評量估測的精確度和效力。此外，在分析時，也考慮了不同的網路負載和收斂速度。當跟其他傳統靜態流量模型去做比較分析時，我們提議的動態估測方法在各式無線網路中，對資料流量的變化展現出相當好的即時察覺能力。

另一方面，我們也發展一套新的統計性決策架構。在無線感測網路中含多個無線感測器的集合，它可以挑選出較佳的感測器子集合後，再啟動它們。而且是在滿足使用者指定的各種服務品質 (Quality of Service (QoS)) 標準下挑選出。感測器節點的用電完全由周遭環境以能量擷取 (energy harvesting) 方式供電。感測器節點依據可用的電池能量以有效又經濟的方被啟動。只是電池能量的多寡無法被決策器直接觀察取得。我們的決策架構包含兩方面：第一是每一個感測器目前可用電池能量的估測；第二是感應器選擇策略。能量估測的步驟是依據累積的能量擷取程序 (Cumulative Energy Harvesting Process)，將其實現於合作型感測網路上 (collaborative wireless sensor networks)。感應器選擇策略依據上述估測的電池能量，採用交叉熵 (Cross-Entropy) 方法實踐。交叉熵方法有效地解決某些問題造成的組合問題 (combinatorial problem)，選擇出具有較長能量壽命的感應器集合。本研究的結果讓網路具有較長的能量壽命，以順應未來巨大資料網路的需求。我們也研究各種參數的效應，深入地洞悉我們提議的架構在不同條件下操作，依然堅實有效。

關鍵字

貝式推論 ； 802.11標準分散式協調功能 ； 馬可夫鍊蒙地卡羅 ； 粒子濾波器 ； 估測 ； 非飽和資料暫存器 ； 物聯網 ； 合作型感測網路 ； 高斯隨機程序 ； 能量擷取 ； 交叉熵

並列摘要

Abstract
Next generation network (e.g., 5G networks) are increasingly confronted with demands to deal with the huge data and energy problems. Huge data and energy problems are of vital importance in the Internet of Things (IoT) network. This dissertation includes two sub-works which could lend supports to the solutions to huge data and energy problems. As coming down to the feasibilities of the huge data and energy problems, we may substantiate them to the two sub-works which are context estimation and energy-aware. The first sub-work we bend our mind to is the context estimation, whereas the second one is the network optimization with the ability of the energy-aware. We simulate our proposed frameworks on the current mature networks, say, IEEE 802.11 DCF network and wireless sensor network, due to these two networks have been playing the most important roles on the wireless network evolution process toward the IoT.

In the first work, we proposes a particle filter framework to perform an online estimation of the unsaturated buffers of the stations in the IEEE 802.11 DCF network. Using this framework, an access point can adapt flow control to its serving stations and configure related parameters dynamically, thus improving the system throughput and reducing the packet latency. Current research analyzing the unsaturated condition in the IEEE 802.11 DCF network is based on the steady-state model, whereas this proposed method is devoted to the dynamic estimation for the probability distribution of the unsaturated buffer in the stations, in either homogeneous or heterogeneous networks. This study also employs theoretical support from the Bayesian Inference to the particle-filtering algorithm. The estimation accuracy and effectiveness were evaluated via Root Mean Square Error and time complexity. Furthermore, we considered different network loads and the convergence speeds in our analysis. Our analysis demonstrated that the dynamic estimation scheme we are proposing has a greater awareness of the traffic changes in the varying wireless networks, when compared to the traditional static traffic model.

On the other hand, we also develop a new statistical decision making framework to select the optimal subset of wireless sensors to activate sensors, while meeting various Quality of Service (QoS) criteria specified by users queries. The sensor nodes are powered solely by energy harvested from the environment and should be activated in an efficient and economical manner based on the available battery energy, which may not be directly observed by the decision maker. Our decision making framework consists of two aspects: the first is the estimation of the current available battery levels of each of the sensors; and the second is a sensor selection policy. The energy estimation step is based on the Cumulative Energy Harvesting Process which is carried out over the collaborative wireless sensor networks. The sensor selection policy is based on the estimated battery levels and uses the Cross-Entropy method which efficiently solves the resulting combinatorial problem to select the sensor set with the long lifetime. This work exhibits the long lifetime of the network for accommodating the large data network in the future. We also investigate the effect of various parameters, which provides insights into the robustness and effectiveness of our framework under different operational conditions.

並列關鍵字

Bayesian inference ； IEEE 802.11 distributed coordination function (DCF) ； Markov Chain Monte Carlo (MCMC) ； particle filter ； estimation ； unsaturated buffer ； Internet of Things (IoT) ； collaborative wireless sensor networks ； Gaussian process ； Cross-Entropy method

參考文獻

[1] D. Bol, G. de Streel, and D. Flandre. Can we connect trillions of iot sensors in a sustainable way? a technology/circuit perspective (invited). In 2015 IEEE SOI-3D-Subthreshold Microelectronics Technology Unified Conference (S3S), pages 1–3, Oct 2015.

[2] A. Rahmati and L. Zhong. Context-based network estimation for energy-efficient ubiquitous wireless connectivity. IEEE Transactions on Mobile Computing, 10(1): 54–66, Jan 2011.

[4] Simiao Niu, Xiaofeng Wang, Fang Yi, Yu Sheng Zhou, and Zhong Lin Wang. A universal self-charging system driven by random biomechanical energy for sustainable operation of mobile electronics. Nature Communications, 6(8975), Dec 2015.

[6] Gil Reiter. Wireless connectivity for the Internet of Things. Technical report, Texas Instruments, June 2014. http://www.ti.com/lit/wp/swry010/swry010.pdf.

[10] Ming-Yuan Cheng, Guan-Yu Lin, Hung-Yu Wei, and A.C.-C. Hsu. Overload control for Machine-Type-Communications in LTE-Advanced system. IEEE Communications Magazine, 50(6):38–45, Jun 2012.