在多重輸入-多重輸出多載波直序分碼多工擷取通訊系統中使用差動空時碼與軟式多使用者偵測技術的研究

(1)

行政院國家科學委員會專題研究計畫成果報告

研究成果報告(精簡版)

計畫類別：個別型

計畫編號： NSC 95-2221-E-011-026-

執行期間： 95 年 08 月 01 日至 96 年 07 月 31 日執行單位：國立臺灣科技大學電機工程系

計畫主持人：曾德峰

計畫參與人員：博士班研究生-兼任助理：蔡宗儒

碩士班研究生-兼任助理：林智謙、胡仕憲

報告附件：出席國際會議研究心得報告及發表論文

處理方式：本計畫可公開查詢

中華民國 96 年 10 月 27 日

(2)

- I -

行政院國家科學委員會補助專題研究計畫 █ 成果報告

□期中進度報告

研究使用空間頻率格子碼於正交分頻多工系統中結合殘存路徑處理及卡門濾波器追蹤多路徑塊狀時變衰減通道

計畫類別：■ 個別型計畫 □ 整合型計畫計畫編號：NSC 95－2221－ E －011－026－

執行期間： 95 年 08 月 01 日至 96 年 07 月 31 日

計畫主持人：曾德峰共同主持人：

計畫參與人員：林智謙、胡仕憲、蔡宗儒

成果報告類型(依經費核定清單規定繳交)：■精簡報告 □完整報告

本成果報告包括以下應繳交之附件：

□赴國外出差或研習心得報告一份

□赴大陸地區出差或研習心得報告一份

■出席國際學術會議心得報告及發表之論文各一份

□國際合作研究計畫國外研究報告書一份

處理方式：除產學合作研究計畫、提升產業技術及人才培育研究計畫、

列管計畫及下列情形者外，得立即公開查詢

□涉及專利或其他智慧財產權，□一年□二年後可公開查詢

執行單位：國立台灣科技大學電機工程學系

中華民國 96 年 10 月 25 日

(3)

- II -

行政院國家科學委員會專題研究計畫成果報告

研究使用空間頻率格子碼於正交分頻多工系統中結合殘存路徑處理及卡門濾波器追蹤多路徑塊狀時變衰減通道

Space-Frequency Trellis Codes in OFDM Systems with Kalman filter Tracking and Per-Survivor Processing over Frequency-Selective Time-Varying Fading Channel

計畫編號：NSC 95- 2221 – E - 011- 026 執行期限：95 年 08 月 01 日至 96 年 07 月 31 日

主持人：曾德峰教授國立台灣科技大學電機工程學系 Email: dtseng@mail.ntust.edu.tw

參與人員：林智謙、胡仕憲、蔡宗儒國立台灣科技大學電機工程所研究生一、中文摘要

在本篇結案報告中，我們將針對以空間頻率格子碼為基礎的多重輸入單一輸出正交分頻多工通訊系統，探討如何利用殘存路徑處理結合卡門濾波器在頻率選擇性時間變化塊狀衰減的通道模型中作通道估測。在傳輸端，先由簡易迴旋碼編碼器對資料位元串流編碼，接著經過解多工器將迴旋碼對映到星座點，並收集所需要的資料長度後再經由正交分頻多工調變後傳送塊狀資料。在接收端，首先利用己知的訓練序列得到卡門濾波器的通道估測初始值，並由殘存路徑處理結合卡門濾波器追蹤每一子載波通道狀態並且更新其通道估測值。藉由所得之估測結果利用在頻域通道響應各子載波彼此間的相關特性作最小均方差，進一步的使得每一子載波通道估測誤差值達到最小。解調端可利用殘存路徑處理卡門濾波器所估測的通道增益經由維特比演算法解碼出資訊位元。本計畫考量連續正交分頻多工浮元間的編碼以及通道的頻譜相關性，在快速的瑞雷衰減環境以及相當長的時間延遲效應下，與當今存在的方法相比，我們所提出的方法具有較低的接收端複雜度，較優良的系統良率。

關鍵詞：殘存路徑處理、卡門濾波器、空間頻率格子碼、維特比演算法、正交分頻多工塊狀傳輸系統

Abstract

In this report, a Kalman filter (KF) in conjunction with per-survivor processing

(PSP) is employed for space-frequency trellis coding (SFTC) multiple-input single-output (MISO) orthogonal frequency division multiplexing (OFDM) systems over frequency selective time-varying block fading channel. The data stream is first encoded by a simple convolutional encoder, followed by a demultiplexer, which maps the trellis output of convolution encoder into a vector of constellation points belonging to a selected modulation format. Depending on the number of transmit antennas, nT

modulated symbols are first collected for each symbol time instant. Then, with the length of an OFDM block, the symbol sequence is first implemented by an IFFT operation followed by pulse-shaping filtering.

At the receiver, Kalman filtering is utilized in conjunction with per-survivor processing (PSP) to predict the channel state information, assumed to be block-faded but featured with fairly long delay spread in comparison to symbol time interval. An initial channel estimate associated with the transmit antenna array for each receive antenna element is realized by the insertion of pilot symbols placed in the front end of each data frame.

Indeed, it is essential to differentiate channel state information aggregate from transmit antennas among each other upon initial prediction. Rather than straightforwardly employ a bank of Kalman filters in a recursive fashion, which is quite computationally demanding, we consider channel state update on a basis of per sub-carrier of transmit antenna array coupled

(4)

- III -

with an exploitation of correlation in frequency domain due to the delay spread.

Via the per-survivor processing coupled with Kalman filtering, one could track channel state in one OFDM symbol and update it in the next interval. Using the property of the correlation of sub-carrier, minimum mean-square error (MMSE) could refine channel estimates. Finally, the data demodulation could be performed by means of well known Viterbi algorithm.

Keywords: Per-survivor processing, Kalman filter, Space frequency trellis coded system, orthogonal frequency division multiple access, Viterbi algorithm

二、緣由與目的

Channel estimation in OFDM systems has been well studied in literature, for instance, [1]-[2], where training (known) symbols [3], inserted in the transmitted OFDM data block, are employed to gain insights into time-varying characteristics arising from Doppler frequency shift along with carrier frequency offset. As such, it is possible to identify the channel gains in frequency or time-domain by means of signal processing algorithms at the receiver through exploiting the knowledge of these known symbols. The idea of the algorithms in [4]

was to exploit OFDM channel correlation, thus the statistics (mean, variance) of the channel must be known as a-priori to make these algorithms feasible. Instead, channel statistics are not needed in [5]. The idea there was to build a system model with unknown coefficients, then exploit the information of the output symbols and known training symbols to estimate the coefficients of the model. Beek [1] proposed two OFDM channel estimation algorithms, namely the minimum mean square error (MMSE) algorithm and the least squares (LS) [6]

algorithm. The MMSE algorithm uses only the correlation of the channel in frequency domain (i.e. the correlation between sub-channels) and fails to address time-domain dynamics. The gains of the sub-channels are defined as the state variables, and an autoregressive (AR) process is used to model the dynamics of the channel. In this perspective, the Kalman

filtering can be applied to predict the channel in the frequency domain from received OFDM signals when pilot symbols are available. This Kalman filter is of a dimension equal to the product of the number of sub-carriers, Nsub, and the order of the AR model, which could be fairly large however.

Moreover, space-time coding has been recognized as an essential means of improving system capacity by realizing both coding gain as well as diversity gain in certain circumstances [13]. In this research project, per-sub-carrier Kalman filtering scheme is utilized in space-frequency coded OFDM systems for channel tracking and symbol detection with an aim at reducing computational complexity [7]. In principle, the Kalman filter channel estimator is applied to obtain the channel gain in each sub-carrier independently, followed by a minimum- mean-square-error (MMSE) combiner for a refinement of frequency-domain channel estimates with consideration of frequency selectivity arising from delay spread. Indeed, the channel cross-correlation between l-th and n-th sub-carriers could be decomposed into two separate terms as a product:

,

0 2

2 2

2

( ) ( ) ( )

1 2 ( )

(2 )

1 4 ( )

k n k n

t

d

t

r m E H l H l m

j n k

J f mT T

n k T π σ

π π σ

⎡ ∗ ⎤

= ⎣ − ⎦

− −

=

+ −

(1)

In view of (1), it is the per-sub-carrier Kalman estimator which explores the time-domain correlation of the channel, while the MMSE combiner explores the associated frequency domain correlation.

Per-survivor processing (PSP) [8]

coupled with Kalman filtering is an attractive approach to performing maximum-likelihood symbol sequence estimation (MLSE) over mobile radio channels that are rapidly time varying. In fact, per-survivor processing (PSP) has been known as an effective approach for simultaneous estimation of both data sequence and the unknown channel parameters in single-input single-output systems [9] without exhaustive search over trellis. In [10], a PSP-based receiver for decoding space-time trellis coded systems is proposed, in which an accelerated self-tuning least mean square (LMS) algorithm for

(5)

- IV -

tracking the fading channels is utilized.

Rather than that, we highlight a per-survivor processing (PSP) receiver that performs joint space-frequency trellis decoding and channel estimation in time-varying block fading channels by means of Kalman filtering.

Similar to [11], we incorporate the PSP algorithm into the Kalman filtering such that the Viterbi algorithm could thus be enabled.

However, the major contribution comes from the fact that we consider the time correlation between consecutive OFDM symbols and frequency correlation among sub-carriers.

Further, the information bit stream is assumed to encompass a couple of OFDM symbols such that PSP is performed over consecutive OFDM data blocks, rather than a single one, to the contrary of that in [11]. In addition, in order to conform to the nature of multiple transmit antennas, we design a simple pilot pattern so as to initialize the channel state variables of each Kalman filter.

Notation: Bold upper case letters denote matrices, bold lower case letters stand for column vectors; (⋅)^*, (⋅)^T ,and (⋅)^H denote conjugate, transpose, and Hermitian trans- pose, respectively; and E(⋅) for expectation, IK is an identity matrix of size K, FN denotes

an N×N FFT matrix with the

(^{p +1,q +1})_st ^entry ¹ ^N^{exp( 2}⁻ⁱ ^π^{pq N}^/ ⁾ [ ]

, ∀p q, ∈ 0,N−1 ; diag(X) stands for a diagonal matrix with X on its diagonal.

[ ]D denotes the_p (^{p +1})_st entry of a vector, and[ ]^D _{p q}_, denotes the (^{p +1,q +1})_st^{entry of}

a matrix.K : number of space-time symbol blocks, k: no. of block index, i: no. of frame size index. ε : transmit antenna index. ⊗ indicates Hadamard product.

The rest of this report is organized as follows. In Section III, the space-frequency coded OFDM system model is first described, and followed by the proposed PSP-KF receiver on a basis of per-sub-carrier. We then give numerical results in Section IV, where conclusions are drawn as well.

三、研究結果與討論

We now present numerical results for space-frequency coded OFDM systems by

means of our proposed receiver processing.

In addition, a comparison with conventional schemes where no PSP was employed is demonstrated in the sequel. The simple convolutional code is a rate-2/4 code and could be referenced to Figs. 3 and 4, where the generator polynomial matrix is[1 0 2 0 ; 4 2 3 4]₈. The spectral efficiency is thus 1b/s/Hz. Note that the trellis termination is performed at the end of each transmitted data frame. Two-transmit and one-receive antennas are solely configured in the following simulations. An extension to a scenario with an arbitrary size would be straightforward once the corresponding channel encoder is selected. To illuminate time variation of channel due to mobility, the product of Doppler shift and OFDM symbol is set at 0.01. The modulation format selected is quadrature phase-shift keying (QPSK). A data frame size of 192 uncoded bits and a number of 16 sub-carriers are used. An initial channel estimate is aided by an orthogonal space-time matrix in a manner similar to the Alamouti code. We assume that the length of cyclic prefix is longer than delay spread, which is within 0.044 OFDM symbol time, such that no inter-block interference (IBI) is introduced. The channels in all simulations are dispersive and the amplitude of each multipath component is considered to be Rayleigh fading and randomly generated at each run. In the meanwhile, channel impulse response is normalized to be one for each one-to-one transmit and receiver antenna pair.

Moreover, a block fading channel environment is considered in such a manner that channel fading gains are assumed constant within an OFDM symbol time but are varied independently on a basis of symbol-by-symbol interval. At least one thousand independent runs are carried out to constitute a single frame or bit error rate point in all the below figures.

In Fig. 5, we compare the performance results of using PSP and non-PSP (conventional) receivers for a 4-state convolutional code in terms of frame error rate (FER), while its counterpart of 8-state is illustrated in Fig. 6. As one could observe in these two plots, with the aid of per-survivor processing for improved channel estimates in

(6)

- V -

one OFDM block the performance curve associated with is below its counterpart at the expense of computational complexity.

However, the enhancement translates to a better Kalman prediction of channel states on next OFDM symbol. Besides, with higher complexity from the larger number of trellis states it could be seen from Figs. 5 and 6 that the FER could be decreased to some extent.

We could also improve the performance result by means of an insertion of longer known symbol sequence at the beginning of each transmit OFDM symbol. Transmission efficiency is defined as the ratio of the number of known pilot tones to the number of transmitted symbols in an observed window of data blocks. Indeed, one could expect that with larger percentage of known data symbols transmitted the performance results in terms of both FER and bit error rate (BER), as shown in Figs. 7 and 8, respectively, are enhanced due to the fact that the receiver might be re-trained and could exploit channel state transition in a more precise manner. In both Figs. 7 and 8, lower transmission efficiency of 2/3 outperforms its counterpart of higher efficiency at 5/6.

四、計畫成果自評

We proposed a space-frequency coded OFDM system suitable for high-speed wireless data transmission over time-varying frequency-selective channels arising from fairly long delay spread in comparison with a transmit symbol time interval. In this perspective, channel modeling is performed in frequency domain and channel tracking is thus operated via Kalman filtering. Note that in order to enhance system capacity, space frequency trellis coded system is considered.

In view of this multi-input structure, per-survivor processing could be employed at the receiver end in realization of a simple trellis diagram due to a convolutional encoder established in the very front end of the transmit framework. Simulations conducted under block fading channel conditions are thereafter presented and exhibited superiority over conventional approach in terms of frame error probability.

The future work should encompass a more detailed derivation of performance

analysis in terms of bit error rate under block-fading channel conditions. Besides, a multiuser scenario should be devised to facilitate system bandwidth efficiency.

Space-Frequency Trellis Encoder

Symbol Mapper

Symbol Mapper Info.

Bits

S/P

Fig.1. Transmission block diagram

Fig.2. Trellis diagram for consecutive OFDM symbols in transmit structure

Fig.3. Rate 2/3 8-state convolutional encoder

0 1 2

3

00, 01, 02, 03 10. 11, 12, 13 20. 21, 22, 23 30, 31, 32, 33 22, 23, 20, 21 32, 33, 30, 31 02, 03, 00, 01 12, 13, 10, 11

Fig.4. Trellis diagram of rate-2/3 8-state convolutional code and output QPSK modulation

(7)

- VI - Fig.5. Performance comparison of PSP and

conventional receivers using 4-state covolutional encoding in terms of frame error rate (FER)

Fig.6. Performance comparison of PSP and conventional receivers using optimal 8-state covolutional encoding in terms of frame error rate

Fig.7. Performance comparison of PSP and conventional receivers using optimal 8-state covolutional encoding under different transmission efficiency in terms of FER

Fig.8. Performance comparison of PSP and conventional receivers using optimal 8-state covolutional encoding under different transmission efficiency in terms of bit error rate (BER)

Table 1: Simulation parameters for comparative receivers employed in space-frequency coded systems

五、參考文獻

[1] J.-J. van de Beek, O. Edfors, M. Sandell, S. K.

Wilson, and P. O. Borjesson, “On channel estimation in OFDM systems,” Proc. IEEE VTC, Chicago IL, July 1995, pp. 815-819.

[2] O. Edfors,M. Sandell, J. -J van de Beek, S.

K.Wilson and P. O. Borjesson, “ OFDM channel estimation by singular value decomposition,”

IEEE Trans. on Commun., vol.46, pp. 931-939, July 1998.

[3] Y. Li, “Pilot-symbol-aided channel estimation for OFDM wireless systems”, IEEE Trans. on Vechicular Tech., vol. 49, no. 4, July 2000, pp.1131-1135

[4] Y. Li, L. J. Cimini,and N. R. Sollenberger,

“Robust channel estimation for OFDM systems with rapid dispersive fading channel,” IEEE Trans. Commun., vol.46, pp.902-915, July 1998.

0 2 4 6 8 10 12

10^-3 10^-2 10^-1

PSP-receiver v.s Non-PSP-receiver

SNR-(dB)

BER

trainng+MMSE-PSP training+MMSE-non-PSP trainng+MMSE-PSP training+MMSE-non-PSP

0 2 4 6 8 10 12

10^-2 10^-1 10⁰

PSP-receiver V.S Non-PSP-receiver

SNR-(dB)

FER

trainng+MMSE-PSP training+MMSE-non-PSP trainng+MMSE-PSP training+MMSE-non-PSP

0 2 4 6 8 10 12

10^-2 10^-1 10⁰

SNR-(dB)

FER

optimal-8state

trainng+MMSE-PSP training+MMSE-NON-PSP

0 2 4 6 8 10 12

10^-2 10^-1 10⁰

SNR-(dB)

FER

optimal-4state

trainng+MMSE-PSP training+MMSE-non-PSP

2 2

AR Model of order

2 2

Num of Transmit antenna

1 1

Num of Receive antenna

2.0 b/s/ Hz 2.0 b/s/ Hz

Bandwidth efficiency

Non-PSP Receiver PSP-Receiver

Receiver Structure

0~12 0~12

SNR (dB)

[ 1 j -1 -j ] [ 1 j -1 -j ]

Constellation

0.044 0.044

Delay spread / T

0.01 0.01

fd.T

2 / 3 2 / 3

Transmission efficiency

2 2

Training length (block)

16 16

Num of sub-carriers

192 192

Frame size (uncoded-bits)

2 / 4 2 / 4

Code rate Generator matrix Constraint length

Optimal-8state Optimal-8state

Num of state

2 2

AR Model of order

2 2

Num of Transmit antenna

1 1

Num of Receive antenna

2.0 b/s/ Hz 2.0 b/s/ Hz

Bandwidth efficiency

Non-PSP Receiver PSP-Receiver

Receiver Structure

0~12 0~12

SNR (dB)

[ 1 j -1 -j ] [ 1 j -1 -j ]

Constellation

0.044 0.044

Delay spread / T

0.01 0.01

fd.T

2 / 3 2 / 3

Transmission efficiency

2 2

Training length (block)

16 16

Num of sub-carriers

192 192

Frame size (uncoded-bits)

2 / 4 2 / 4

Code rate Generator matrix Constraint length

Optimal-8state Optimal-8state

Num of state

4 8

3 2 4

0 2 0

1 ⎥

⎦

⎢ ⎤

⎣

⎡ 48

3 2 4

0 2 0

1 ⎥

⎦

⎢ ⎤

⎣

⎡

[2 3]8 [2 3]8

5/6

2/3

5/6

2/3

(8)

- VII - [5] M. X. Chang and Y. T. Su, “Model-based channel estimation for OFDM signals in Rayleigh fading,”

IEEE Trans. on Commun., vol 50, No. 4, pp 540-544, Apr 2002

[6] S. M. Kay, Fundamentals of Statistical Signal Processing: Estimation Theory, Prentice-Hall, NJ, 1993.

[7] C. Wei, Ruifeng Zhang, “Estimation of Time and Frequency Selective Channels in OFDM Systems:

A Kalman Filter Structure,” IEEE Globecomm., vol 2, Dec. 2004, pp.800 – 803.

[8] C. Cozzo and B. L. Hughes, “An adaptive receiver for space-time trellis codes based on per-survivor processing", IEEE Trans. Commun., vol. 50, no. 8, pp.1213-1216, Aug. 2002.

[9] R. Raheli, A. Polydoros, and C.-K. Tzou,

“Per-survivor processing: A general approach to MLSE in uncertain environments,” IEEE Trans.

Commun., vol. 43, pp. 354–364, Feb. 1995.

[10] Y. Xue and X. Zhu, “PSP decoder for space-time trellis code based on accelerated self-tuning LMS algorithm”, Electronic Letters, vol. 36, no. 17, pp.1472-1474, Aug. 2000.

[11] R. J. Piechocki , A. R .Nix , and J. P. Mcgeehan ,

“Joint semi-blind detection and channel estimation in space-frequency trellis coded MIMO-OFDM , “ in Proc. IEEC ICC , 11-15 May, 2003, pp. 3382-3386.

[12] S. Kay, J.Makhoul, ”On the Statistics of the Estimated Reflection Coefficients of an Autoregressive Process”, IEEE Trans. Acoustics, Speech, and signal processing, 1983, pp.1447-1455.

[13] V. Tarokh, N. Seshadri, and A. R. Calderbank,

``Space-Time Codes for High Data Rate Wireless Communication: Performance Criterion and Code Construction,'' IEEE Trans. Inform. Theory, vol.

44, no. 2,pp. 744-765, Mar. 1998.

(9)

- 8 -

(10)

表 Y04

行政院國家科學委員會補助國內專家學者出席國際學術會議報告

96 年 10 月 25 日報告人姓名

曾德峰服務機構

及職稱

國立台灣科技大學電機工程系助理教授

時間會議地點

自民國 96 年 06 月 24 日至民國 96 年 06 月 28 日英國，格拉斯哥

本會核定

補助文號 NSC 95－2221－ E －011－026－

會議名稱

(中文) 2007 國際電子電機工程師協會之通訊討論會

(英文) 2007 IEEE International Conference on Communications (ICC’07)

發表論文題目

(中文) 使用遞迴式雜訊抑制在空時塊狀編碼多直序展頻碼通訊系統的研究 (英文) Iterative Interference Suppression for High-Rate Single-Carrier

Space-Time Block-Coded CDMA

附件I

(11)

表 Y04

一、參加會議經過

2007 國際電子電機工程師協會之通訊討論會(2007 IEEE International Conference on Communications (ICC2007))於二 o o 七年六月二十四日至二十八日為期四天，假英國(蘇格蘭)，格拉斯哥市的國際會議中心舉行◦ 此會議由 IEEE Communications Society 與英國主要電訊公司如 Vodafone 以及蘇格蘭政府共同舉辦，是此學門、領域的年度盛會，其接受率約為百分之三十，精挑細選投稿的文章，因此廣受學術界及工業界的高度重視，筆者深感榮幸能參與並口語報告於國際知名學者之前◦ 今年的研討會議提供四大主要研究討論的領域: (1)未來無線通訊系統(The future of Convergence in the Communications industry)探討高品質、全 IP 為主軸的即時多媒體通訊網路架構 (2) 分散式合作通訊網路 (Towards Cooperative and Cognitive Wireless Networks)討論使用多數入多輸出 (MIMO)技術及設計應用於感應網路(sensor network)的 PHY 及 MAC 層 (3) 邁向具有感知偵測能力的無線網路(Towards Cognition in Wireless Networks) (4) 調適性多數入多輸出技術及效能 (Adaptive MIMO Techniques and Performance)提供系統容量於不同的傳輸環境，如相關性輸入多傳輸路徑下，切換傳輸模式已達成最佳化◦ 大會主席由 Tariq S Durrani 以及 Arun Sarin 主持，善盡地主之情誼及對蘇格蘭當地的歷史、

古蹟詳盡介紹 ̦ 留下深刻的印象◦ 此外，將近有來自世界各地三千人的參與，更激起學習新知的慾望與認是知名學者的興奮。

IEEE ICC，為 IEEE Communications Society 認可的唯二的旗艦會議之一，為一個目標明確、理論及實務並重的專業國際論文發表會議◦ 她提供了一個平台，使得來自全世界通訊相關的學界及工業界精英能夠交換彼此心得及發表最新的研究理念與結果，為未來的無線及有線網路系統及天線設計等技術做前瞻性的擘劃，使得網路效能大步的提升，傳輸技術更貼

(12)

表 Y04

切，頻寬的使用更加有效率，因而共同努力打造一個更舒適及有效率的地球村◦ 大會首日舉辦了八場針對先進無線通訊技術趨勢及應用的研討會，

由當今在各主要研究領域有顯著貢獻的學術專業人士與企業界精英講述，

並分成講授課程及研究講習會兩大類別◦第二天起共舉辦了九百篇(含壁報式)的專題論文發表會，每場發表八至十二篇的論文，當中並有社交時間討論、交換心得，預估共有超過兩千人出席了本屆 ICC 的會議◦

二、與會心得

此次會議所主導討論的主題，偏向分散式的無線網路系統，相當多數的報告文章與無限感測網路有直接或間接應用性的關聯。眾所周知，隨著資訊藉由無線傳播的速度與有限頻譜的分配使用有嚴重衝突的情形下，已經使得當今無線通訊系統的感知以及合作技術顯得非常關鍵且重要。除此之外，為了增進無線通訊產品的應用範圍，研究提高傳播速度及環境所衍生的無線通道不確定性質以及困難的預測方法，在加上考量硬體的複雜度都增加了提高服務品質的困難度及可行性◦ 更甚者，降低消耗功率以達成抑制產生干擾源，但又能符合在此複雜的分散式無線系統中維持整個系統的基本需要的良率，又能兼顧達到各個使用者的QoS 服務品質，還能將系統效能發揮到最佳化(如最節省的電源消耗模式)，便成了當今以及下一代無線通訊系統研究領域的其中之一要項。多數入多輸出技術 (MIMO) 是一種空間分集，不需加寬頻譜，被證實具有強大的反干擾的特性，也可解釋為低耗能、高傳輸的特性，但需要良好的傳輸環境◦ 爾後，Space-Time Coding 由貝爾實驗室裡的研究人員發明，使用通道編碼，防止錯誤率上升，代價是throughput 較低。渦輪原理 (turbo principle) 也被廣泛使用在簡易的多傳輸端、複雜的接收器使用。一般而言，將通道編碼、時空間、頻率分集效能結合，進而提升傳輸的速度或系統穩定。由於走向全IP 的無線網路模式，該次會議很多著名學者如喬治亞理工的Gordon Stuber 都

(13)

表 Y04

提出跨階層(PHY, MAC, Routing) 的分散式無線網路設計架構，讓筆者深感國外學者追求前瞻性研究的旺盛企圖心。參加完這次的國際會議，再次打開自己的國際視野，知道自己未來是要置身在國際中，與這些學者先進一同競爭，就更提醒自己，不能以現有的研究結果為滿足，並且要更積極參與國際學術組織的會議，盡所能擷取新知。

筆者在此次會議中，有著無限的榮幸受邀為【正交分碼多工多重接取系統】單元中的多使用者技術Multiuser Techniques 的會議時段以口頭方式發表與前碩士班指導的兩位畢業學生(賴威諭、蔡宗儒)所著作的論文，也是國科會個人型專題計畫內容 ◦筆者著重於使用 multi-sequence 的概念於直序分碼多工系統的多維訊號空間方法來增進訊號彼此的距離，並且創造出更多可用的優良的空時正交碼，進而提升整體傳輸的速率。此種概念應用於直序分碼多工非同步系統，筆者與之前畢業的碩士生(盧耿志)已經發表於IET Communications 2007 年八月號。除此之外，吾等更利用串連式傳輸方塊架構，並且結合使用渦輪碼的解碼原理與多使用者偵測的技術，

在多次遞迴的進行後，與傳統的分碼多工系統有明顯的改善系統容量◦ 詳細的內容，經過校正後，已經被日本 IEICE 期刊在 2007 年十一月的專刊登。有鑑於多使用者的趨勢在未來 4G 的系統角色，與會的澳洲雪梨大學電機工程學系(University of Sydney, Australia) 的 Branka Vucetic 教授與我探討如何將此多維訊號空間應用於多重輸入及多重輸出的空時塊狀編碼 (space-time block coding) 模式在快速時變通道情況下，使我獲益良多。

至於如何將此概念隱含於調適性傳輸(adaptive modulation) 模式應用於跨階層(cross-layer design) 設計，則是一個非常值得深入研究的題目，需要時間做更進一步的思考及搜尋資料。

(14)

表 Y04

三、建議

本屆大會在籌備工作進度論文、線上投稿系統、議程安排註冊報名，會後雞尾酒活動以及預定旅館資訊及對蘇格蘭國家地理的資訊等等，都提供與會人士最大的協助◦ 此種體貼入微的服務及對國際人士的包容、體貼，非常值得政府相關單位的派人考察。也希望我國在不久的將來，也能多多爭取到舉辦類似的大型、有指標性的國際型會議來提昇我國通訊產業的學術、工業界的全球競爭力◦

四、攜回資料名稱及內容

整個會議的論文集 CD-ROM 一片。

五、結語及其他

筆者參加此會議，巧遇昔日在學的老師、學長以及曾在研究所共同學習的同學，及來自國內台大、交大‛ 清大‛ 中央、中正等大學的教授，工研院的經理等，學習甚多，受益廣闊◦ 筆者非常感謝國際合作處的前輩能夠補助筆者的國科會個人型計畫裡的出國開會項目，參加此次的 IEEE 2007 ICC 在蘇格蘭舉辦的國際會議◦ 除了得以口頭發表研究成果, 並能藉此機會與著名國內外學者專家（如台灣大學陳教授光禎、鍾教授嘉德、交通大學張教授仲儒、王教授蒞君、清華大學的王教授晉良、雪梨大學Branka Vucetic 教授、Carleton U.的 David Falconer 教授、Georgia Tech 的 Gordon Stuber 教授、HKUST 的 Lateiff 教授）交換最新研究心得及分享研究成果◦

這對從事無線通訊系統研究有興趣者如筆者而言，有著無與倫比的鼓舞、

啟發及幫助, 再次十二萬分感謝國科會委員及相關業務人員的辛勞及經費上的支持◦ 經由與前述專家的討論後，筆者正著手下ㄧ階段的研究，希望明年能再接再厲為國爭光。

(15)

表 Y04

(16)

表 Y04

(17)

表 Y04

(18)

表 Y04

(19)

表 Y04

(20)

表 Y04