ATP: 用於無電池物聯網裝置的自適應傳輸策略

Abstract

能量收集技術 (Energy harvesting) 實現了物聯網 (IoT) 系統中自我維持的願景。開發人員可以利用此項技術從環境中收集能量，以補充裝置在運行過程中消耗的能量。為了支持長時間或耗電的任務執行，多數的無線感測器節點選擇電池作為其電源。然而，電池對環境是有害的，需要定期地人力維護以保持電池的清潔和無腐蝕。近年來，使用電容代替電池的無電池物聯網系統在降低成本和提高安全性方面帶來了好處。這些系統在任務執行後將感測器節點設置於睡眠模式以幫助電容充電，並調整佔空比 (duty cycle) 以實現長期運行。與電池相比，電容的能量儲存空間很小，裝置很容易因為能量不足而出現斷電的情形。因此，更有效率地利用有限的能源並確保任務能夠成功執行是很重要的。對於無電池的感測應用，維持感測速率有助於評估環境中事件發生的前因後果與隨時間的變化。若能根據環境條件調整節點的傳輸間隔，有助於平衡能量供需。本論文分析了無電池的物聯網感測裝置任務執行所需的電容電壓，並提出了自適應傳輸策略 (稱為 ATP) 及應用批處理 (batch processing) 來保持電容的能量水平。ATP 的目標是保持感測速率，並儘可能地提高有效吞吐量 (goodput)。ATP 有三種實作方法：基於閾值 ATP、數學預測 ATP 與改良的基於閾值 ATP。實驗證明，當所需的環境條件得到滿足時，所提出的方法有助於感測器節點維持其感測速率。模擬結果表明，ATP 有足夠的穩定性來適應環境的變化; 就傳輸性能（即傳輸時間間隔）而言，數學預測 ATP 有比相關研究與 ATP 的其他兩種方法更好的結果。

Energy harvesting technology achieves the vision of self-sustaining in the Internet of Things (IoT) system. Developers can use this technology to harvest energy from the environment to supplement the energy consumed by devices during operation. In order to support long-term or power-consuming tasks, most wireless sensor nodes select batteries as their power supply. Batteries are harmful to the environment, however, and regular maintenance is required to keep the batteries clean and corrosion-free. In recent years, battery-less IoT systems that use capacitors instead of batteries have brought benefits in reducing costs and improving safety. These systems put the sensor nodes in sleep mode for charging after task execution and adjust the duty cycle for long-term operation. Compared with batteries, the energy storage of capacitors is tiny, and the device is prone to power failures due to insufficient energy. Therefore, it is important to use energy more efficiently and to ensure that tasks can be executed successfully.For battery-less sensing applications, maintaining sensing rate helps to assess environmental changes. Adjusting the transmission interval according to environmental conditions is beneficial in balancing energy supply and demand. This thesis analyzes the required capacitor voltage for task execution and proposes a transmission strategy for battery-less IoT sensing devices called Adaptive Transmission Policy (ATP), which applies batch processing to keep the energy level of the capacitor. The objective of ATP is to maintain the sensing rate and increase the throughput as much as possible. ATP has three approaches: threshold-based ATP, mathematical prediction ATP, and modified threshold-based ATP, with mathematical prediction ATP relying on more local computations. It is empirically demonstrated that the proposed approaches is beneficial to maintain the sensing rate when the required environmental conditions are met. The simulation results also show that ATP is robust enough to adapt to environmental changes. Furthermore, in terms of transmission performance (i.e., transmission interval), the mathematical prediction ATP shows better results than both baseline approach and the other two approaches of ATP.