基於低功率藍⽛通訊協定之無線多導極⼼電圖系 統的網路設計

Abstract

我們所設計並實作的無線多導極⼼電圖（Electrocardiography, ECG）系統，是由三 個ECG 導極、⼀個低功率藍⽛閘道器（Bluetooth Low Energy gateway, BLE gateway）， 以及Google 雲端平台所組成。在這種架構下，三個ECG 導極扮演從屬節點（slave）， ⽽低功率藍⽛閘道器則是主控節點（master），他們會形成⼀組無線近⾝網路（Wireless Body Area Network, WBAN）。為了形成無線近⾝網路，我們以低功率藍芽通訊協定 為基礎設計了無線近⾝網路的網路架構，將主控節點的時間資源劃分成許多⻑度相同 的superframe，每個superframe 分成兩個階段，分別為控制階段和資料階段。在控制 階段中，主控節點會和ECG 導極建⽴連線形成無線近⾝網路，接著在資料階段，主控 節點除了會使⽤輪詢機制（polling）來接收每個ECG 導極所擷取的⼼電訊號之外，還 會和ECG 導極進⾏時間同步。由於控制階段分配的時間多寡，會影響主控節點在⼀個 superframe 中可以和多少數量的無線近⾝網路建⽴連線，因此，如何有效分配這兩段 時間是⼀項重要的議題。為了有效分配這兩段時間，我們設計了分析模型來分析低功 率藍⽛建⽴連線所需時間，並使⽤MATLAB 軟體來模擬低功率藍⽛的環境，將低功率 藍⽛建⽴連線所需時間之模擬結果，與我們的分析結果⽐較之後，誤差率約為2%-3%， 因此我們的分析結果是相當精準的。接著透過演算法分配這兩段時間，並計算出主控 節點可以容納的無線近⾝網路個數，也就是低功率藍⽛閘道器最⼤可服務⼈數。

A 100 percent wireless electrocardiography (ECG) system is prototyped and presented in this paper. Our wireless ECG system is composed of three wireless ECG sensor patches, a Bluetooth Low Energy (BLE) gateway, and a Google cloud platform. In this network architecture, the three ECG sensor patches which act as slaves and the BLE gateway which acts as master form a wireless body area network (WBAN). To archive this, time resource of master split into two phases: control phase and data phase. Master establishes connections with slaves to form the WBAN in the control phase and performs time synchronization as well as sensing data collection in the data phase. However, there is a challenge on time allocation between control phase and data phase, because the interval of control phase affects the number of WBAN which master can connect in a superframe. Therefore, we design an analysis model to analyze the connection delay of slaves in control phase to allocate time resource effectively. We use MATLAB to simulate the environment of Bluetooth Low Energy, and then compare the simulation results with our analysis results. Our analysis results are accurate because the error rate is 2%-3%. Finally, we can obtain the number of connected WBAN by time allocation algorithm after we get the connection delay from analysis model.