This article was downloaded by: [National Chiao Tung University 國立交通大學] On: 25 April 2014, At: 21:36
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of the Chinese Institute of Engineers
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tcie20
Throughput performance of DMT‐based VDSL system
at high sampling rates
Sun‐Ting Lin a & Che‐Ho Wei b
a
Department of Electronic Engineering , National Kaohsiung Normal University , Kaohsiung, Taiwan Phone: 886–7–7172930 ext. 7912 E-mail:
b
Department of Electronic Engineering , National Chiao Tung University , Hsinchu, Taiwan
Published online: 04 Mar 2011.
To cite this article: Sun‐Ting Lin & Che‐Ho Wei (2007) Throughput performance of DMT‐based VDSL system at high sampling rates, Journal of the Chinese Institute of Engineers, 30:6, 1109-1113, DOI: 10.1080/02533839.2007.9671339 To link to this article: http://dx.doi.org/10.1080/02533839.2007.9671339
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Short Paper
THROUGHPUT PERFORMANCE OF DMT-BASED VDSL
SYSTEM AT HIGH SAMPLING RATES
Sun-Ting Lin* and Che-Ho Wei
ABSTRACT
Very high-speed digital subscriber line (VDSL) is a cost effective solution for the last mile of the network over the existing twisted-pair telephone lines. The target of the VDSL system is to raise the transmission data rate up to 52 Mbps for short distances (300 m or less). This paper investigates the throughput of a DMT-based VDSL system at high sampling rates under the influence of various noises/interferences. The throughput limitation of the VDSL system is discussed and the optimal solutions of the sampling rates under various test loop lengths and environmental conditions are also investigated.
Key Words: cyclic prefix reconstruction, intersymbol interference (ISI), discrete
multitone (DMT), very high-speed DSL (VDSL).
*Corresponding author. (Tel: 886-7-7172930 ext. 7912; Email: stl@nknu.edu.tw)
S. T. Lin is with the Department of Electronic Engineering, National Kaohsiung Normal University, Kaohsiung, Taiwan
C. H. Wei is with the Department of Electronic Engineering National Chiao Tung University, Hsinchu, Taiwan
I. INTRODUCTION
Discrete multitone (DMT) modulation is a tech-nique in which a transmission channel is partitioned into a number of independent, parallel sub-channels, and each of them carries a lower-speed quadrature amplitude modulated (QAM) signal, as discussed by Chen (1998) and Cioffi (1997). The modulation scheme is now widely used in the DMT-based asymmetric digital subscriber line (ADSL), which can be referred to the ANSI standard T1.413 (1997). Due to the in-creased presence of optic fiber in the telecommuni-cation network, the last mile of a twisted-pair line has been gradually reduced, to below 1.5 Km, espe-cially in urban areas. The available frequency spec-trum of these short copper wires for transmission becomes much broader and its channel capacity grows dramatically. A new service called very high-speed DSL (VDSL), which is G.993.1 (2004), can provide
up to 52 Mbps or even more over a short telephone line to meet broadband access-network requirements. VDSL can be regarded as an evolution of the ADSL system, and several VDSL system architectures have been proposed, for example, Cioffi (1998), Lin and Wei (2001), Chang et al. (2002), and Nava and Del-Toso (2002), etc. These systems use the entire fre-quency band from a few hundred kHz to beyond 10 MHz over loops using various types of technology.
Previously, Lin and Wei (2001) proposed an asymmetric VDSL system, upgraded from a DMT-based ADSL system by multiplying the symbol rate. Its architecture and symbol format are the same as the original ADSL system except that both clock rate and frequency bandwidth are increased in the same ratio. At that time, the DMT technology was not as popular as today and there were many other candi-date technologies for the VDSL system, however, it has become the most popular technology, currently. In this paper, the throughput performance of this pro-posed system is analyzed over VDSL test loops at various sampling rates because there are shortages of these types of throughput performance analysis for the DMT VDSL system. This throughput has its up-per limitation due to the characteristics of the twisted-pair channel. In addition, some noise types such as
1110 Journal of the Chinese Institute of Engineers, Vol. 30, No. 6 (2007)
next-end crosstalk (NEXT), far-end crosstalk (FEXT) and other background noises, which lower the throughput of the system, are taken into consideration in the analysis. Recently, there have been studies to discuss the cancellation of FEXT noise, such as Leshem and Youming (2004), however, more types of interferences are discussed in our paper.
II. PROPOSED VDSL SYSTEM DESCRIPTIONS
In this section, the VDSL system described in this paper has the same architecture as the ADSL sys-tem discussed by Lin and Wei (2001), as shown in Fig. 1. All the functional blocks in Fig. 1 are built for computer simulations by using MATLAB. The basic DMT-based ADSL transceiver system includes 512-point FFT, and its symbol rate is 4 kHz on the downstream side. If the cyclic prefix length is 32, this DMT system includes 256 sub-channels, at band-widths of 4.3125 kHz each, for data carrying and other overheads such as cyclic prefixes or synchronization symbols. Since the maximum throughput supported by this transmission system depends on the charac-teristics of the channel, the channel modeling and bit-loading of each sub-channel must be computed first. After these calculations are performed, the through-put of this DMT system is determined by the product of the symbol rate multiplying the total bit-loading summation in a symbol.
The throughput supported by this system de-pends on the channel characteristics and the band-width used. For example, if the FFT size is fixed and the symbol rate is doubled, the bandwidth of each sub-channel should be increased to 8.625 kHz, which is twice the original system configuration. Therefore, the total bandwidth used is doubled. Otherwise, if the bandwidth of each sub-channel is fixed as 4.3125 kHz, then the FFT size must be increased to enlarge the system throughput.
III. INFLUENCE OF NOISES/INTERFER-ENCES ON THE THROUGHPUT
In this section, some computer simulations are used to analyze the relationship between throughput performance and sampling rate. From these results, it can be seen that the throughput of a DMT-based VDSL system depends on the test loop length and sampling rate as well as working environment.
The curve of throughput performance and sam-pling rate can be divided into two parts: linear region and saturation region. In the linear region, the throughput increases with the sampling rate, while in the saturation region, the throughput remains nearly flat. In general, the transition point from the linear
region into the saturation region depends on the loop length. As the test loop gets longer, the throughput of the DMT-based VDSL system becomes smaller and the saturation point occurs at a lower sampling rate. We can see that the throughput also depends on the SNR and channel characteristics, such as bridged taps and noise/ interference sources like AWGN, ADSL NEXT, VDSL FEXT, etc.
1. AWGN vs. Crosstalk
In ITU-T G.996.1 (2001), many crosstalk inter-ferences are included in the test procedure for DSL transceivers. Here, different crosstalk interferences, such as HDSL, ADSL NEXT and VDSL FEXT, etc. are selected to simulate their influences on the sys-tem throughput. In general, NEXT interference is much larger than FEXT for the same type of signals, therefore, the NEXTs from other DSL systems play a dominant role for system throughput if they coexist. The first case simulated is studying the influence of the HDSL system. The power spectrum of the HDSL signal is concentrated below 700 kHz. Doubling the sampling rate of the VDSL system, with bandwidths about 4.4 MHz, the effect of the HDSL system is then insignificant, with degradation less than 5%. The T1 NEXT, with stronger power spectral density and broader bandwidth has more effects on the system throughput. The ADSL system with bandwidth of 2.208 MHz has similar ef-fects on VDSL performance. In Fig. 2, the efef-fects of various crosstalk interferences are displayed by plotting, together with the original AWGN corrupted signal, at test loop length 300 m and length 600 m.
In examining the self-crosstalk interferences, the NEXT problem can be discarded and only the FEXT of the VDSL system should be considered. It is clear that this FEXT depends on the loop length. The shorter the loop is, the larger the degradation will be, as also shown in Fig. 2 The formulas used in the above simu-lations refer to the ITU-T standard G.996.1 (2001). From those formulas, it can be observed that the amount of induced FEXT depends on the loop length while
x0 y0 xk yk xN–1 yN–1 rL r0 z0 rN+L–1 zN–1 s(t) r(t) P(t) N(t) H(t) FFT + FEQ Serial-to-parallel P arallel-to-serial DMT symbol generator IFFT Frame synchronizer TEQ
Fig. 1 Block diagram of VDSL system
that of NEXT does not. Our simulation results indi-cate this phenomenon. These results are reasonable because the coupling of the interference from the far-end will decay more through the longer loop.
2. AWGN vs. Bridged Taps
It has been observed that bridged taps can cause notches in the frequency response of VDSL loops. If some bridged taps of various lengths are added into the middle of a test loop, the shortest bridged tap will cause the deepest null as described by Wang et al. (1999) and Im (2002). In our test case, we use the simple bridged tap configuration with both vertical and horizontal sections being 15.2 m each. The simu-lation results are also shown in Fig. 2.
For the bridged taps located at the same distance from the end point of the test loops, it can be observed from the simulation curves that the influence of the bridged taps depends on the sampling rate as well as the loop length. In the lower sampling rate case, the bandwidth used does not cover the notch caused by the bridged taps; therefore, the power spectra of these two test loops, with or without bridged taps, are ap-proximately the same. In contrast, if the used band-width covers one notch or more, the bit-loading number under those certain frequency band decrease toward zero. Thus, the throughput of the whole system is degraded. The ratio of the degradation depends on the width, depth and number of notches covered.
IV. ANALYSIS OF THROUGHPUT PERFORMANCE
In Fig. 3, the percentages of the channel capacities under various interference effects to the AWGN-only
case are plotted. From the simulation results, T1 NEXT degrades the system throughput seriously. Since the T1 system is the first generation digital carrier, it does not consider spectral compatibility with other DSL systems. Therefore, it is very noisy when combined with a new DSL system. For example, the shaping filter of the T1 system is not so sharp that its sidelobe signal in-duces severe interference in other systems. In the real world, the single-pair high-speed digital subscriber line (SHDSL: a modulation scheme for repeater-less T1 service) is replacing T1 service in main lines, therefore, the influence of the T1 system can be avoided.
The influence of the HDSL NEXT on the VDSL system is small, especially in a high sampling rate system. The ratio of system throughput reaches above 90%, even to 95% under this NEXT influence while the sampling rate is above 5.5 MHz.
The notches caused by the bridged taps will lower the system throughput if the used band covers these notches. From our simulation, the effect of the bridged taps causes about 10% to 15% penalty to the whole system throughput at high sampling rate, as shown in Fig. 3. It can be seen that if the sampling rate is below 5.5 MHz, the system throughput under the influence of one short bridged tap (15.2 m) can be ignored.
The influence of ADSL NEXT is concentrated at low sampling rate, especially below 2.2 MHz since the shaping filter of an ADSL system is sharp enough to cut its power spectral density under -140 dBm/Hz if the bandwidth is above 2.2 MHz. The throughput degradation caused by the ADSL system is about 10 to 20 percent while the sampling rate is above 11 MHz. From Fig. 3, it can be seen that other VDSL systems induce a large amount of crosstalk especially for a short loop case. The degradation can be more
160 140 120 100 80 60 40 20 0 0 5 10 15 20 25 Sampling rate (MHz) 30 35 40 45 Mbps AWGN Bridge Tap ADSL-NEXT VDSL-FEXT 300 m 600 m
Fig. 2 Throughput comparison of the tested loops in various
en-vironments 150 100 50 0 0 5 10 15 20 Loop length - 300 m 25 30 45 Percentage 120 100 80 60 40 20 0 5 10 15 20 Loop length - 600 m Sampling rate (MHz) 25 30 35 40 45 Percentage HDSL Bridge Tap ADSL-NEXT VDSL-FEXT T1
Fig. 3 Throughput percentage comparison of various noises over
AWGN
1112 Journal of the Chinese Institute of Engineers, Vol. 30, No. 6 (2007)
than 30 percent if the test loop is 300 m long. From the simulations, it is observed that, al-though many effects lower the throughput of a DMT-based VDSL system, the throughput can be increased with the sampling rate until reaching the saturation point. The optimal sampling rate of the system is selected by finding the minimum sampling rate of a test loop with throughput up to 90 percent of its opti-mal value. The influence of an HDSL system is sopti-mall and can be ignored. The influence of a T1 system on a VDSL system is so significant that it is not practi-cal to bundle these two types of service together.
From the simulations, if the loop length is be-low 800 m, increasing the sampling rate will improve the system throughput significantly. The optimal sam-pling rate for the VDSL system varies from 5.5 to 40 MHz if the loop length is below 800 m, as shown in Fig. 4. Otherwise, for loop length above 800 m, 4.4 MHz sampling rate is good enough to reach the opti-mal throughput of a DMT-based xDSL system. The optimal throughput of the DMT-based VDSL system also depends on the loop length. From the simula-tion results, the throughput, including both upstream and downstream, can reach above 50 Mbps only if the loop is shorter than 600 m, as shown in Table 1. From Table 1, it can be seen that the influence of ADSL NEXT is very significant for a long loop, especially one longer than 800 m; while the influ-ence of VDSL FEXT is substantial for short loops.
V. CONCLUSIONS
The throughput performance of a DMT-based VDSL system under various constraints, such as the effects of crosstalk and bridged taps, is analyzed. It is observed that raising the sampling rate can improve
the system throughput, especially when the loop length is short. However, this method has its limitations as shown in the previous simulations, and the optimal solutions are calculated according to the defined constraints. From our simulation results, if the loop length is below 800 m, increasing the sampling rate will improve the system throughput significantly. Otherwise, for loops with lengths from 800 m to 1500 m, a 4.4 MHz sampling rate is good enough to reach the optimal throughput of the DMT-based VDSL system.
REFERENCES
ANSI Standard T1.413, 1997, “Standard Project for
Interfaces Relating to Carrier to Customer Con-nection of Asymmetrical Digital Subscriber Line (ADSL) Equipment,” Issue 2.
Chang, C. C., Wang, M. S., and Chiueh, T. D., 2002, “Design of a DMT-based Baseband Transceiver for Very High-speed Digital Subscriber Lines,”
IEEE Asia-Pacific Conference on Communications,
pp. 367-370.
Chen, W. Y., 1998, DSL Simulation Techniques and
Standards Development for Digital Subscriber Line Systems, Macmillan Technical Publishing,
Indianapolis, IN, USA.
Cioffi, J. M., 1997, A Symmetrical Digital Subscriber
Lines in The Communications Handbook, J. D.
Gibson ed., CRC Press Inc., pp. 450-479. Im, G. H., 2002, “Performance of 51.84-Mb/s VDSL
Transceiver over the Loop with Bridge Taps,” IEEE
Transaction on Communications, Vol. 50, No. 5,
pp.711-717.
ITU-T Standard G.992.5, 2003, Asynchronous
Digi-tal Subscriber Line (ADSL) Transceivers 2,
In-ternational Telecommunication Union, Geneva,
Table 1 Optimal throughput of various DMT VDSL test loops (Mbps)
Length (m) AWGN BT ADSL VDSL
100 227.41 213.37 223.28 155.34 200 179.73 164.61 174.64 124.62 300 137.49 124.08 131.37 98.86 400 103.97 92.12 96.68 79.03 500 78.83 68.82 70.33 62.61 600 61.72 54.63 52.50 51.11 700 50.17 43.26 40.48 42.96 800 28.32 28.25 17.85 28.01 900 24.03 24.47 13.46 23.70 1000 19.78 19.82 9.20 19.50 1100 17.42 16.91 6.89 17.19 1200 15.23 14.26 4.90 15.06 1300 13.04 12.45 2.96 12.91 1400 11.63 11.43 2.49 11.53 1500 10.10 10.37 2.06 10.03
Fig. 4 Optimal sampling rate and throughput of various DMT
VDSL test loops 40 35 30 25 20 15 10 5 0 0 500 Loop length (m) 1000 1500
Optimal sampling rate (MHz)
AWGN Bridge Tap ADSL-NEXT VDSL-FEXT
Switzerland.
ITU-T Standard G.996.1, 2001, Test Procedures for
Digital Subscriber Line (DSL) Transceivers,
In-ternational Telecommunication Union, Geneva, Switzerland.
Leshem, A., and Youming, L., 2004, “A Low Complex-ity Coordinated FEXT Cancellation for VDSL,”
IEEE International Conference on Electronics, Cir-cuit and System, Vol. 11, pp. 338-341.
Lin, S. T., and Wei, C. H., 2001, “Precursor ISI-Free Frame Synchronization for DMT VDSL System,”
IEEE 2001 Globecom, San Antonio, TX, USA, pp.
389-393.
Nava, M. D., and Del-Toso, C., 2002, “A Short
Overview of the VDSL System Requirements,”
IEEE Communications Magazine, Vol. 40, Issue
12, pp. 82-90.
Standard G.993.1, 2004, Very High-speed Digital
Subscriber Lines, International
Telecommunica-tion Union, Geneva, Switzerland.
Wang, A., Werner, J. J., and Kallel, S., 1999, “Effect of Bridged Taps on Channel Capacity at VDSL Frequencies,” IEEE International Conference on
Communications, Vol. 1, pp. 236-245.
Manuscript Received: Apr. 19, 2005 Revision Received: Mar. 15, 2007 and Accepted: Apr. 02, 2007
