在間歇系統上最小化資料年齡

能源採集技術由於維修和充電成本低、永續性佳且不會受到壽命限制等等優點，逐漸取代傳統電池成為新興的供電方式。使用能源採集技術的偵測器也被用在各種即時系統的應用中，例如空氣品質、水庫水質、森林大火、交通狀況、人體數值監測等等。這些偵測器由於供電受到環境影響而較微弱且不穩定、造成系統間歇的執行，需要等待下次能源充足時方能繼續執行。採集的資料常常因為無法被如期的即時傳送而失真，因此資料新鮮度是一個很重要的考量。本文提出一種新穎的演算法能夠最小化系統的資料年齡，使得採集而來的資料得以盡可能的維持新鮮度，藉由對偵測器行為的預測和更新頻率的平均，本篇的方法經由實驗結果表明得以有效的降低資料平均年齡，相比其他現有方法更能維持資料新鮮度。

關鍵字

間歇系統；能源採集技術；資料新鮮度；排程演算法

並列摘要

Energy harvesting sensor nodes are widely used in real time systems due to its low maintenance and high sustainability. However, these nodes suffer from unstable power supply which causes the intermittent system execution. With the passage of time, the sensed data become obsolete which results in distortion of reality [1]. Without knowing the accurate time of recovery, data freshness is considered one of the critical attributes. We make use of the concept of Age of Information (AOI) which enables us to quantify the freshness of the sensed data. In this thesis, a new scheme for minimizing the average AOI has been investigated. The proposed scheme first predicts the pattern of every nodes utilizing moving average, then maintain a buffer in transmitter which keeps the latest updated data. Through our replacement technique implemented on the buffer, we minimize the average AOI of all data. Simulation results indicate that the proposed method achieve better average AOI as compared with the existing schemes.