黃寶儀臺灣大學：電機工程學研究所陳星豪Chen, Hsing-HauHsing-HauChen2007-11-262018-07-062007-11-262018-07-062007http://ntur.lib.ntu.edu.tw//handle/246246/52952本論文的目標是設計出一個加速度計型緊急系統。系統設計的要求有二：一是準確地判斷活動狀態，一是實現高能源效率。 我們使用多層次的感測資料融合機制(multi-level sensor data fusion)來準確地判斷活動狀態。為了實現高能源效率，我們利用融合機制的結果減少無線傳輸的次數，來降低無線電傳輸的耗電量。採用慣用手傾聽的機制(habitual hand listening)使我們可以進一步地降低由於閒置傾聽(idle listening)所造成的無線電能耗。 我們的實驗結果顯示，當人們有意識時，用以判斷活動狀態的不活動期間(inactivity period)所需的時間長度為30秒至4分鐘；當人們熟睡時,不活動期間則需要30分鐘到40分鐘。關於能源效率的模擬結果顯示，與單純地每一週期皆傳送加速度計資料相比，利用融合機制 的結果減少無線傳輸的次數可以使系統壽命延長39倍。採用慣用手傾聽的機制，系統壽命可以進一步再延長兩倍。 從實驗結果來看，我們發現使誤判率(false positive rate)最低所需的不活動期間除了因人而異，亦視所從事的活動而定；也就是沒有一個放諸四海皆準的不活動期間。本論文最大的貢獻在於，藉由顯著地延長系統壽命，我們建立了一個高能源效率緊急系統的典範。In this work, we aim to design an accelerometer-based emergency alarm system able to judge activity state accurately in the meantime achieve energy efficiency. We use the multi-level sensor data fusion scheme to judge activity state accurately. To achieve energy efficiency, we reduce radio power consumption by cutting down the number of times of transmission utilizing the fusion results. Moreover, we reduce radio power consumption due to idle listening by adopting habitual hand listening. Our results of human experiments show that for people being conscious, the inactivity period varies from 30 seconds to 4 minutes; for people being asleep, the inactivity period varies from 30 minutes to 40 minutes. The simulation results show that the lifetime of the system is prolonged 39 times utilizing the fusion results compared to naively sending accelerometer data periodically. Utilizing habitual hand listening, the lifetime of the system is prolonged twice compared to the scheme merely utilizing the fusion results. From the experimental results, we find that the inactivity period to minimize the false positive rate varies from person to person also from activity to activity, i.e., no concrete selection of inactivity period can be drawn. The most contribution of the work is that we set an energy efficient example of emergency alarm system by significantly prolonging the lifetime of system.1 Introduction 1 2 Related Work 4 2.1 Emergency Alarm Systems 4 2.2 Sensor Data Fusion 5 3 Multi-Level Sensor Data Fusion 6 4 Inter-Device Communication 9 4.1 Transmitting According to the First Two Levels of Fusion Results 9 4.2 Habitual Hand Listening 10 5 Experimental Results 12 5.1 Experimental Design 12 5.2 Threshold Determination 14 5.3 False Positive Rate of Human Subjects 14 5.4 False Negative Rate 19 5.5 Energy Efficiency 19 6 Conclusion and Outlook 23 Bibliography 24896182 bytesapplication/pdfen-US高能源效率感測資料融合加速度計緊急系統傳輸體制energy efficiencysensor data fusionaccelerometeremergency alarm systemcommunication scheme[SDGs]SDG7thesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/52952/1/ntu-96-R94921030-1.pdf