基於人體通道傳輸之正交分頻多工基頻處理器設計

Abstract

無線近身網路已蓬勃發展，而其中的一個實體層—人體通道傳輸，在生醫行動看護上提供了我們不少的便利性；近年來我們更是逐步將其應用推廣至多媒體系統上的使用。但根據IEEE 802.15.6的標準定義，其傳輸率目前最大僅止於1.3125Mbps，遠遠不足以提供多媒體的影音傳輸應用。與此同時，除了要增進傳輸率之外，還需要保有可靠的訊息傳輸和低功耗的特性，故須設計一個新的通訊協定及模組來達到此一目標。 為了適用於高傳輸速率的應用，我們在此選用具有高頻譜效率和高傳輸速率的正交多頻分工調變來實作人體通道傳輸系統。但由於每個載波的功率大小變異性極高，造成訊號容易被截波而失真，對可靠傳輸來說可謂是一大阻礙。為了讓訊息能更穩定的被接收端所偵測到，我們設計了一套適用於我們系統的自動化增益控制器來使訊息在解調變之前能夠預先回到其該有的振幅上，讓封包沒被偵測到的機率降低；BER更可以在SNR從10到15dB的範圍中降低2到4個數量級。 另一個問題則為人體天線效應所產生的頻譜干擾。根據人體通道的量測結果，合適的頻帶落於30-50MHz之中；但是此頻帶已有其他許可的訊號在傳輸，造成人體通道傳輸中會有其他雜訊透過人體天線效應耦合進來，使得訊息有可能因此污染而失真。在正交分頻多工系統之中，此錯誤訊息將很難以被復原成正確訊號，因此我們設計了另一套適應性載波處理的選擇系統來避開被汙染的子通道，進而使BER能得到約2個數量級的下降率。 最後，我們以人體通道傳輸的原型模組來驗證基頻處理器設計；將基頻處理器搭配正交分頻多工調變燒錄至 FPGA中，再加上整合在PCB板上的前端電路完成實際傳輸。透過我們基頻處理器，偵測失敗率可降低至0.1%以下，而 BER 在SNR = 12dB下則可下降3個數量級。 此基頻演算法在電路的功率上幾乎沒任何提高，但由於偵測準確率的提高以及BER的下降，將其結果應用在實驗室下的晶片實現上(傳輸速率29.1Mbps)，能降低1.306mW的功率，使得能量/位元比降低約12%。

The Wireless Body Area Network (WBAN) has flourished nowadays. One of the physical layers (PHYs), Body Channel Communication (BCC), provides a more convenient life to us for the healthcare applications. The applications have even been extended to the multimedia information transmission. However, the maximum data rate is only up to 1.3125 Mbps according to the standardized definition of IEEE 802.15.6, which is not enough to support the application of the multimedia transmission. In addition to enhance the data rate, the property of the reliable signal transmission and the ultra-low power operation also need to be remained. Therefore, a new communication protocol and modules should be designed to achieve this goal. To support the multimedia transmission, the Orthogonal Frequency Division Multiplexing (OFDM) modulation is adopted due to its property of high data rate and high spectral efficiency. But the power of each carrier for OFDM has high variations, causing the signal chopped and distorted and the BCC transmission not reliable. To make the signals be detected more robustly, a low-cost Automatic Gain Control (AGC) is proposed, which is suitable for our BCC system. The AGC make the signals recover to its original amplitude before the demodulated progress and reduce the packet failed ratio. The BER performance can be improved 2 to 4 order when the signal to noise ratio (SNR) equals 10dB to 15dB. The other major problem is the frequency response interference generated from the body antenna effect. According to the measurement, the suitable transmission band of the body is located in the frequency between 30 to 50 MHz, but the band contains other licensed signals. The other noise will be coupled to the body channel through the body antenna effect, making the signal contaminated. The errors are hard to recover back to the original correct signal, so an adaptive technique is adopted to avoid the bad sub-channel, making the BER performance get 2 order improvement. Finally, the BCC prototype is used to verify the baseband processor design. The baseband processor with the OFDM-based modulation is programmed to FPGA; the PCB boards integrated with the front-end circuits are also used to reach the real transmission path. By our proposed baseband processer design, the packet detection failed ratio can be reduced to 0.1%, and the BER performance can get 3 order improvement when the SNR equals 12dB. The proposed algorithm in baseband only provides a little overhead in the power. However, due to the enhancement of the packet success detection ratio and the reduction of BER, the implementation result can reduce the power by 1.306mW, about 12 percentage point reduction of energy per bit when applying to the chip implementation in our lab under the data rate with 29.1Mbps.