標題: 數位行動無線電通訊之同步方法
Synchronization Methods in Digital Mobile Radio Communications
作者: 黃永亮
Yung-Liang Huang
黃家齊
Chia-Chi Huang
電信工程研究所
關鍵字: 同步;高斯最小相移鍵;頻率正交分割多工;快速傅立葉轉換;符號時序;頻率偏移;前序;保護間隔;Synchronization;Gaussian Minimum Shift Keying (GMSK);Orthogonal Frequency Division Multiplexing (OFDM);Fast Fourier Transform (FFT);Symbol Timing;Frequency Offset;Preamble;Guard Interval
公開日期: 1998
摘要: 在這本論文中,我們探討了兩種用於數位行動無線電通訊之同步方法,並且分別在採用高斯最小相移鍵 (Gaussian Minimum Shift Keying, GMSK) 及頻率正交分割多工 (Orthogonal Frequency Division Multiplexing, OFDM) 等調變方式的系統中評估這兩種同步方法。 首先,我們提出一種全數位式,可共同完成頻率偏移補償及符號時序恢復之非同調 (noncoherent) 及同調 (coherent) GMSK接收器架構。當採用同調解調變時,所須之載波相位偏移也可被估計出來。 接收到的基頻複數訊號先經頻率鑑別,再經過執行快速傅立葉轉換 (Fast Fourier Transform, FFT) 的數位濾波器;頻率偏移即可從FFT的直流項算出,而符號時序錯誤可從FFT的某一特殊頻率之相位角度算出,此頻率為位元傳輸率之一半之整數倍。這兩估計參數則可用來在前序 (preamble) 時期中做頻率偏移補償及符號時序恢復。在前序時期中做完頻率偏移補償後,則可以平均同相及正交相訊號來估算出粗略載波相位。我們以計算機模擬,在加成性白高斯雜訊 (additive white Gaussian noise, AWGN) 通道中評估這個 GMSK 接收器架構的位元錯誤率 (bit error rate, BER) 效能。計算機模擬結果顯示,當採用非同調解調變時,我們的接收器僅需12位元的訓練前序,故適用於叢發型 (burst-mode) 數據通訊;當採用同調解調變時,此接收器亦可達較佳之 BER 效能。 接著,我們提出一種可用於歐洲Eureka 147數位廣播系統中,共同完成符號、時框、及載波同步之方法。Eureka 147 採用OFDM調變。我們先讓保護間隔 (guard interval) 中之訊號與有效訊號之最後四分之一訊號做複數相乘,符號時序可藉偵測此相乘值 (亦即相似訊息) 的相位角度的突變估計出。此突變之偵測是基於最大相似度 (maximal likelihood, ML) 法則。分數載波間隔 (fractional carrier spacing) 之頻率偏移是在估出符號時序後,從上述相乘值之相位角度算出。粗略時框同步及虛符號之偵測可同樣地藉該相似訊息估計出。整數載波間隔 (integral carrier spacing) 之頻率偏移是藉接收到的相位參考符號與本地產生但已做頻率移之相位參考符號做迴旋 (convolution) 後決定出。我們發現保護間隔之長度是此同步演算法中最重要的參數。計算機模擬結果顯示,這個同步方法之BER 效能在 AWGN 通道及兩路瑞雷衰減 (two-path Rayleigh fading) 通道中趨近於理想同步。
In this dissertation, we investigate two different synchronization methods for digital mobile radio communications. The two methods are evaluated on systems that adopt Gaussian minimum shift keying (GMSK) modulation and Orthogonal Frequency Division Multiplexing (OFDM) modulation, respectively. We first proposed a fully digital noncoherent and coherent GMSK receiver architecture with joint frequency offset compensation and symbol timing recovery. Carrier phase offset can be estimated when the coherent demodulation mode is adopted. The down converted complex signal is first frequency discriminated and then passed through a digital filter that performs a Fast Fourier Transform (FFT). The frequency offset can be estimated from the DC component of the FFT and the symbol timing error can be estimated from the phase angle of the FFT at a specified frequency which is equal to an integral multiple of half the bit rate. These two estimated parameters are then used for frequency offset compensation and symbol timing recovery during a preamble period. Coarse carrier phase can be estimated by averaging sampled in-phase and quadrature-phase signals and finding its phase angle within the preamble period after carrier frequency offset is estimated and compensated. The BER performance of this GMSK receiver architecture is assessed for an AWGN channel by computer simulation. Simulation results show that our receiver requires only a 12 bits of training preamble in the noncoherent demodulation mode and its performance is suitable for burst-mode data communications. This receiver architecture can also achieve better BER performance with coherent detection. Next, we presented a joint symbol, frame, and carrier synchronization method for the Eureka 147 digital audio broadcasting (DAB) signal that adopts OFDM modulation. Symbol timing is determined first by detecting an abrupt change in the phase angle of the complex product between the last quarter of a useful symbol and its cyclic extension in the guard interval. The detection of this abrupt change is based on the maximal likelihood (ML) principle. Frequency offset of fractional carrier spacing is estimated from the phase angle of the autocorrelation after symbol timing is estimated. Coarse frame synchronization and null symbol detection can also be achieved through this correlation information. Frequency offset of integral carrier spacing is determined from the convolution outputs between a received phase reference symbol and several locally generated but frequency shifted phase reference symbols. We found the length of a guard interval is the most important parameter for the synchronization algorithm to work. Simulation results show that the performance of this synchronization method approaches to the ideal synchronization case in both an AWGN channel and a two-path Rayleigh fading channel.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT870435100
http://hdl.handle.net/11536/64560
Appears in Collections:Thesis