Case1: There is only one RFI in each HAM-band.
RFI_1 center frequency fc= 3.755MHz (tone index= # 871) RFI_2 center frequency fc= 7.05MHz (tone index= # 1635) (a)The placement of measurement tone = ( # 870 ,#1634) (b)The placement of measurement tone = ( # 860 ,#1624) The simulation shows as follow:
0 500 1000 1500 2000 2500 3000 3500 4000 4500
0 500 1000 1500 2000 2500 3000 3500 4000 4500
-200
Figure 5.2-1 Multiple RFI signal suppression by Model 1 algorithm In the Figure 5.2-1(b), the performance of RFI cancellation can achieve 30~40dB. And the performance of every modulated tones after RFI cancellation was almost reduced under AWGN power level (-140dBm/Hz).
Case2: There are more than one RFI in a HAM-band RFI_1 center frequency fc= 3.755MHz (tone index= # 871) RFI_2 center frequency fc= 3.6484MHz (tone index= # 846) (a)The placement of measurement tone = ( # 870 ,#845) (c)The placement of measurement tone = ( # 860 ,#835)
0 500 1000 1500 2000 2500 3000 3500 4000 4500
-200 -180 -160 -140 -120 -100 -80 -60 -40 -20
Power Spectral Density
dBm/Hz
subtone
residual RFI RFI
VDSL PSD mask AWGN
(a)distance = 5
400 600 800 1000 1200 1400 1600
0 500 1000 1500 2000 2500 3000 3500 4000 4500
-200
400 600 800 1000 1200 1400 1600 -140
-120 -100 -80 -60 -40
Power Spectral Density
dBm/Hz
subtone
residual RFI RFI
VDSL PSD mask AWGN
RFI_1
RFI_2
(d) Magnification of (c)
Figure 5.2-2 Multiple RFI signal suppression by Model 1 algorithm
Chapter 6
Conclusion
A more accurate RFI model is used in the development of a new RFI cancellation algorithm. It is shown that with proper assumptions of VDSL deployment environment, the model can be reduced to a commonly-used simplified RFI model and still remain its effectiveness if careful placement of measurement tones is conducted. Based on our simulation, the algorithm can achieve 40dB RFI cancellation at best, if the measurement tones are placed further away from the RFI-peak. But when we considered the influence of crosstalk noise on SNR= RFI
noise(=AWGN+Crosstalk) , the measurement tones cannot be further away form the RFI-peak without limitation. In Model 1 algorithm simulation the measurement tones should be placed 8 -15 tones away from the RFI-peak to give better performance of RFI cancellation; in Model 2 the measurement tones only be placed 5 tones away from the RFI-peak to satisfy RFI cancellation requirement. Therefore, Model 2 algorithm is more accurate than Model 1 and the placement of measurement tones is a trade off between low model error effects and high SNR.
Future Work
Digital RFI cancellation may be implemented in time domain. The following diagram shows the basic idea.
M odeling[ ]
rfi n
S o u r c e[ ] r fi n
Figure 6-1 RFI cancellation in time domain
In our approach, the basic idea is to model a constant wave as envelope of RFI in one DMT symbol, and then try to decrease RFI signal in time domain.
( )
T is the sampling period of ADC.
c o s ( 2 ) (6 -2 )
[ ]
iM o d e lin g c f
r fi n =
⋅π n T
+φ
After the FFT, we can obtain the RFI model in frequency domain.
[ ] * 0... -1 (6-3)
See (3.5-7),
index k index k
L f tone index
After the FFT, we can measure the RFI tones from the HAM band. Using RFI tones to calculate RFI model parameter (b,∆) in frequency domain. bcan convert to RFI model parameter (c,φ) in time domain.
RFI suppression in time domain. The simulation shows as follow.
Modeling[ ]
rfi n rfiS o u rce[ ]n rfiModeling[ ]n
(a) RFI cancellation in time domain
2910 2920 2930 2940 2950 2960 2970 2980 -0.02
-0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02
sampling point (sampling rate = 1/35.328MHz)
Voltage
1 DMT symbol RFI in time domain
RFI source RFI modeling residual RFI
(b) Magnification of (a)
0 500 1000 1500 2000 2500 3000 3500 4000 4500 -200
-180 -160 -140 -120 -100 -80 -60 -40
Power Spectral Density
dBm/Hz
subtone
RFI
VDSL PSD mask residual RFI AWGN
(c) PSD of (a)
Figure 6-2 RFI signal suppression by time domain cancellation algorithm
In Figure 6-2 (a) and (c), we can observe that the residual RFI signal in time domain decreases and consequently decreases the residual RFI signal in the frequency domain. But in some simulation cases shown as Figure 6-3 (a) this algorithm can’t reduce the residual RFI in the time domain. The simulation shows as follow.
(a) RFI cancellation in time domain
(b) PSD of (a)
Figure 6-3 RFI signal suppression by time domain cancellation algorithm
In Figure 6-3 (a) the residual RFI signal increases, however, in the Figure 6-3 (b) the PSD of residual RFI signal decreases. This is because the residual RFI signal has satisfied certain conditions.
The satisfying conditions are:
1. The sub-carrier index of residual RFI signal is integer.
2. Envelope of residual RFI signal is constant.
(The residual RFI signal is a single carrier.)
( ) b f N T
c⋅ ⋅ − ⋅ ⋅ =
sf N T
c s0 ( ) a f N T
c⋅ ⋅ − ⋅ ⋅ =
sf N T
c s0.5
Figure 6-4 Spectral leakage in the digital domain caused by a single carrier (a) fc between two sub-carrier frequency (b) fc equal to a sub-carrier frequency.
Once the two conditions are satisfied, the RFI spectral leakage can be effective reduced shown as Figure 6-4 (b). Therefore, if we can make residual RFI to fit the two conditions, we can achieve the goal of RFI signal suppression, too. Until now this algorithm seems to be effective in RFI cancellation. But in Figure 6-3 simulation case could be a special one. We need to do more simulations or modify RFI signal model to support that RFI cancellation in time domain can work.
Appendix A.
1 2
( cos( 2 )) cos( 2 ). (3.5-4)
rfi[ ] e i
S n = c + d π f nT +φ π fnT +φ
After N (=2N’)-point Fast Fourier Transform (FFT), the RFI signal is represented as:
And Where means the center frequency
After N (=2N’)-point Fast Fourier Transform (FFT), the RFI signal is
represented as:
2 2
{ } { }
Reference
[1] John A. C. Bingham, “RFI suppression in multicarriers transmission systems”. Globecom,vol.2.pp. 1026-1030, Nov, 1996
[2] John A. C. Bingham, ”ADSL, VDSL, and Multicarrier Modulation-10.6 Dealing with RFI form Ham and AM Radio,” Copyright 20000 by John Wiley
& Sons, Inc.
[3] Committee T1-Telecommunications Working Group T1E1.4 ” Very high speed Digital Subscriber Line (VDSL) ;Part 1: Functional requirements and Common Specification-12 Test procedures and requirements ”, 2002
[4] ITU, “Conventional telephone signal”, Recommendation G.227, ITU,1968.
[5] Martin Sehlstedt, “RFI Cancellation in VDSL “, Master’s Thesis. Lulea University of Technology.2000.
[6] Yaohui Liu, Timo I. Laakso, and Paulo S.R. Diniz,”Adaptive RFI Cancellation in VDSL Systems,” European Conference on Circuit Theory and Design, Aug 28-31, 2001, Espoo, Finland.
[7] Chen-Chin Huang. “Design and Implementation of an RFI Cancellation Scheme and a UPBO Scheme for DMT-Based VDSL Transmission Systems,”
Master’s Thesis. National Tsing Hua University, Hsinchu, Taiwan, ROC, June, 2002.
[8] Frank Sjoberg, Rickard Nilsson, Per Ola Borjesson “Digital RFI Suppression in DMT-Based VDSL Systems” IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS-I:REGULAR PAPERS. 51, NO. 11, NOVEMBER 2004.
[9] Byung-Jang Teong and Kyung-Hyun Yoo, ”Digital RFI canceller for DMT based VDSL.” IEEE Electronics Letters, vol.34, No. 17, Aug. 1998.
[10] B. Wise and J. Bingham, “Digital radio frequency cancellation for DMT VDSL,” Tech. Rep. T1E1.4/97-460, ANSI, Sacramento, CA, Dec.1997
[11] Arthur J. Redfern, Member, IEEE ”Receiver Window Design for Multicarrier Communication Systems.” IEEE Journal on selected areas in communications, Vol.20, NO. 5, JUNE 2002
[12] Dr. Dennis J. Ranuschmayer, “ADSL/VDSL Principles” Copyright 1998 by Macmillan Technical Publishing U.S.A.
[13] Ray Andraka, “A survey of CORDIC algorithms for FPGA based computer” Copyright 1998
[14] Vladmir Oksman, “Standard VDSL Technology Overview of European (ETSI), North American (T1E1.4) and International (ITU) VDSL standard development” IEEE 802.3 EFM SG, Broadcom Corporation, July 2001
簡歷
姓名: 江啟仁 性別: 男
生日: 民國六十六年一月十八日 學歷:
2003.9~2005.7 國立交通大學電子工程學系碩士班 畢業 1996.9~2000.7 國立台灣科技大學電資工程學系學士班 畢業 1993.9~1996.7 國立彰化師大附工 畢業
1990.9~1993.7 彰化縣立員林中學 畢業 1984.9~1990.7 彰化縣立靜修國小 畢業