T1E1.4/99-199R2. T1E1.4: VDSL and ITU. Title: VDSL System Requirements Proposals for ITU (99-199R2)

Transcription

1 May 31, ` T1E1.4/199R2 T1E1.4/99-199R2 Project: T1E1.4: VDSL and ITU Title: VDSL System Requirements Proposals for ITU (99-199R2) Contact: J. Cioffi (Editor) Information Systems Laboratory Stanford University Stanford, CA Phone: ; Fax: Date: April 20, 1999 Dist'n: T1E1.4 Abstract: This document repeats contribution 199 from interim meeting, where it was agreed with consensus that this document should be forwarded to ITU as American VDSL requirements. The ITU Boston meeting did not have time to review it, so we submitted it again through SG B, since it had consensus agreement, as a US position. Apparently, some companies reconsidered their endorsement at Study Group B meeting last week. This contribution can go to ITU as company or US position or not at all as the group decides. However, the editor requests we decide whether T1E1.4 has requirements with consensus support or not. Only this cover page is provided at the meeting because R1 was loaded as US SG B document prior to the electronic deadline for reduced copies, and it has been provided already at the April meeting, and has been available at both ITU and T1E1.4 web pages for months. NOTICE * Contact: John M. Cioffi Dept of EE, Stanford University Stanford, CA Tel: Fax: cioffi@stanford.edu

2 - 2 - This contribution has been prepared to assist Standards Committee T1 - Telecommunications. This document is offered to the Committee as a basis for discussion and is not a binding on any of the companies listed as authors. The requirements are subject to change after further study. The authors specifically reserve the right to add to, amend, or withdraw the statements contained herein.

3 - 3 - UIT - Secteur de la normalisation des télécommunications ITU - Telecommunication Standardization Sector UIT - Sector de Normalización de las Telecomunicaciones B Study Period Commission d'études ;Study Group;Comisión de Estudio } 15 Contribution tardive;delayed Contribution ;Contribución tardía } D.xxx Geneva, June 21 - July 1, 1999 Texte disponible seulement en ;Text available only in;texto disponible solamente en } E Question(s): SOURCE*: TITLE: United States of America G.vdsl: American T1E1.4 Consensus G.vdsl requirements restated Abstract: This contribution reformats some of the PO-023 G.vdsl requirements so that each requirement can be mapped to a specific issue in G.vdsl.

4 - 4 - G.vdsl: American T1E1.4 Consensus G.vdsl requirements restated J. Cioffi (T1E1.4 Editor, VDSL System Requirements) Department of Electrical Engineering Stanford University, Stanford, CA Phone: ; Fax: Introduction: This contribution reformats some of the PO-023 G.vdsl requirements so that each requirement can be mapped to a specific issue in G.vdsl. The following table of contents may be useful. 1. INTRODUCTION: PROPOSALS WITH CROSS REFERENCE TO VDSL ISSUES-LIST ITEMS: ITU VDSL ITEM OTHER SERVICES/CROSSTALKERS ITU VDSL ITEM RF EGRESS AND INGRESS ITU VDSL ITEM RANGES AND RATES ITU VDSL ITEMS 5.3 AND PROGRAMMABILITY ITU VDSL ITEM NOISE MODEL ITU VDSL ITEM LOOP MODELS ITU VDSL ITEMS VDSL TESTING ITU VDSL ITEMS OAMP NEW ISSUES FOR G.VDSL FROM [2] PERFORMANCE REQUIREMENTS NEW ISSUE ON ACTIVATION AND POWER CONTROL SUMMARY: REFERENCES Proposals with cross reference to VDSL issues-list items:

5 - 5 - The word "should" probably needs change to "shall" everywhere to follow differing rules of text for SG15/Q4 with respect to T1E1.4, but this editor will leave that for discussion later as the two words have different meeting in ITU. In T1E1.4, these were meant as requirements, but reports/documents in T1E1.4 can not use the word "shall." 2.1 ITU VDSL item other services/crosstalkers From Section of [2], the VDSL system-requirement proposal to ITU is POTS should be separated from VDSL by a splitter circuit that is possible to be implemented passively that prevents failure of the VDSL system from affecting the POTS service. Loss of VDSL service should not obstruct the operation of POTS. From Section of [2], the proposal to ITU is ISDN should be separated from VDSL by a splitter circuit that is possible to be implemented passively that prevents failure of the VDSL system from affecting the ISDN service. Local power failure should not obstruct the operation of ISDN. From Section of [2], the proposal to ITU is VDSL service should be compatible with other VDSL, ISDN, HDSL, ADSL and/or POTS services on twisted pair in other lines in the same binder. Co-existence of asymmetric and symmetric VDSL systems in the same multi-pair cable should be possible, possibly with performance penalty. No special arrangements should be required for pair selection. From Section of [2], the proposal to ITU is VDSL should be fully compliant with FCC part 15 with cable considerations for measurement From Section of [2], the proposal to ITU is VDSL should operate at bit error rate lower than 1e-7 with 6 db of margin in the presence of electromagnetic interference from AM radio, any amateur radio operators.

6 ITU VDSL item RF Egress and Ingress The following text from Section 6.22 of [2] should be included, perhaps with editorial adjustment, in the RF ingress portion of the ITU G.vdsl requirements: extracted Ingress section Radio Noise (A) General Requirements: (A) The VDSL system is required to meet its reach and quality of service requirements with adequate margin (6 db at 1e-7), considering crosstalk (see Section 6.2.1), impulse noise (see Section 6.2.3), system noise and broadband environmental noise (See Section 6.2.4) contributions, while at the same time the loop is subject to simultaneous RFI from multiple AM broadcast stations, and an adjacent amateur radio station. It may be necessary to employ special measures where customers aerial drop wire or unshielded riser cable is very close to the nearest interfering transmitter, i.e., within 300ft of a 50 kw AM broadcast transmitter. Similarly special measures may be necessary to deal with the situation where an unshielded part of the customer s loop or riser is within 200 ft. of an amateur station s antenna. Two worst cases are mutually exclusive since it is unlikely that an amateur station will be able to operate as close as 300 ft from an AM broadcast transmitter, so when worst-case AM broadcast interference is encountered, it can be assumed that the worst-case amateur radio interference is likely to be 10 db lower than worst-case, and the same argument can be applied reciprocally when the customer is close to an amateur station Test Requirements (A) The VDSL system should meet transmission performance objectives over the reference test loops with 6 db margin while subject to any of 3 simulated RFI threats, each comprising 13 different signals. Each simulated RFI threat includes 10 simulated AM broadcast stations in the band 535 to 1605 khz and one simulated Amateur Radio SSB transmission. Other sources of RFI can be considered in the VDSL standard, but were omitted here for lack of contributions in the area after 18 months of requests for such.

7 - 7 - The AM broadcast sources are modeled by a fixed frequency carrier 30% AM modulated with a flat (± 3 db) Gaussian noise source band limited to 0-5 khz. The average power of the modulated signal is specified in the Table below. The simulated amateur transmission is an SSB modulated carrier that changes frequency (F3) every 2 minutes, by at least 50 khz, and visits all amateur bands in the VDSL passband during each BER test. The baseband signal is speech weighted noise (ITU-T, Rec. G.227). This is interrupted on a 15-second period with 5 sec ON AND 10 seconds OFF to simulate speech activity. The resultant baseband signal is further interrupted on a period of 200 ms with 50 ms ON and 150 ms OFF -- approximately the syllabic rate. The doubly interrupted signal is then bandlimited to 4 khz and subject to 6 db per octave pre-emphasis. (This model combines the effects of transmit path speech processing for enhanced intelligibility and the spectral balance of sibilant unvoiced sounds.) The two types of RFI noises into VDSL are further specified below: AM Radio Noise Table: (A) (CM = common mode ; DM = differential mode) Models 1 and 2 represent urban high-density areas, while Model 3 represents a suburban environment. The following differential mode levels were proposed based on measurements Signal Strengths for AM Radio Noise into VDSL. distance to AM radio transmitter 300 ft ft ft ft -70 DM strength (dbm) Balance below 1.6 MHz was assumed to be -60 db. Model 1 has 2 AM stations at 300 ft., 2 more at 6000 ft, 4 more within ft, and 2 at ft. Model 2 has 1 AM station at 300 ft., 5 more at 6000 ft. (with 2 at the slightly higher level of -50 dbm), 3 at ft., and 1 at ft. Model 3 has 5 at 6000 ft. (two are slightly higher at -50 dbm), 3 at ft., and 2 at ft.

8 - 8 - Three AM Radio Noise Threats for VDSL. Frequency (khz) Model 1 (high-density urban with co-located transmitters) CM dbm (50 ) DM dbm (100 ) Model 2 (high-density urban with primary/secondary transmitters) CM dbm (50 ) DM dbm (100 ) CM dbm (50 ) Model 3 (suburban) DM dbm (100 ) 660 (AM) (AM) (AM) (AM) (AM) (AM) (AM) (AM) (AM) (AM) (Note: CM is a Longitudinal signal with the common mode terminated in 50. DM is a Differential signal with the differential mode terminated in 100.) (Note: The AM broadcast frequencies above may require modification to provide the worst-case scenario for specific VDSL frequency spectra use.) Amateur Radio Noise: (A)

9 - 9 - Amateur radio interference noise should be generated, with a DM of -10 dbm end of RF ingress section The following Egress section from Section of [2] should be incorporated. (N-ON corresponds to situations where RF egress is determined to be of concern). In the amateur RF bands in the Table, the PSD limit should be altered to 80 dbm/hz when option N-ON is selected. and is otherwise specified by the ADSL-compatible and PSD enhancement options. The following bands corresponding to Amateur Radio Bands. Amateur Radio Bands MHz MHz MHz MHz MHz MHz MHz MHz MHz 2.3 ITU VDSL item ranges and rates The following text from Section 5.1 of [2] is proposed for VDSL ranges and rates issue Section 5.1 VDSL should consider both asymmetric and symmetric transmission between the node and customer. The payload data rate combinations to be given primary evaluation are:

10 VDSL Payload Bit Rates Name of Service Type Downstream Data Rate (Mbps) Upstream Data Rate (Mbps) Range kft. (km) Asymmetric Short (.3) 34 or Medium (1) Long (1.5) or.8 6 (2) Symmetric Short (.3) Medium (1) Long (1.5) The data rates in the above table for system performance evaluation should be considered for purposes requiring greater bit-rate granularity as 52 Mbps = 51.2 Mbps, 26 Mbps = 25.6 Mbps, 19 Mbps = 19.2 Mbps, 13 Mbps = 12.8 Mbps, and 6.5 Mbps = 6.4 Mbps. These data rates are primarily intended for ATM transport and are for range-evaluation purposes only. The payload data rate of the EOC should not be less than 24 kbps and not more than 64 kbps. The EOC should be able to operate in a clear-channel mode that is duplex transparent bit or byte transmission. 2.4 ITU VDSL items 5.3 and programmability The following interpretation of the data rates in Section 5.1 of [2] should be forwarded to the ITU to address items 5.3 and 5.4.

11 The data rates of G.vdsl shall be programmable and encompass approximately those in the table above with control of said programmability being with the service provider. 2.5 ITU VDSL item noise model The following noise models from Section 6.2 of [2] are proposed as a subset of requirements for the ITU G.vdsl requirements: beginning of section VDSL Noises Crosstalk Noise (A) Near-end crosstalk or NEXT should be a Gaussian signal with power spectral density given by PSD NEXT = PSD disturber K next ( N / 49) f where N is the number of crosstalkers. Far-end crosstalk or FEXT is similarly PSD FEXT = PSD disturber K fext.6 2 ( N / 49) d 2 H ( f ) f where d is the length of the loop in feet. For category-5 twisted pair, which is listed in Section 8.5 of the appendix for information purposes, the coefficient K next = changes to K next cat 5 = and the coefficient K fext = changes to K fext cat 5 = The category-5 numbers may be more appropriate for new buildings or intra-campus/corporate symmetric transmission ISDN NEXT AND FEXT (A) For ISDN NEXT or FEXT,

12 PSD ISDN ( f ) = K ISDN 2 f 0 sin πf f 0 πf f f f 0 4 where f 0 = 80 khz, K ISDN = 5 9 V 2 p R, V p = 2.5 Volts, and R = 135 Ω HDSL NEXT AND FEXT (A) For HDSL NEXT or FEXT, PSD HDSL ( f ) = K HDSL 2 sin πf f 0 f 0 πf f f f 3 db 8 where f 0 = 392 khz, f 3 db = 196 khz, K HDSL = 5 V 2 p 9 R, V p = 2.7 Volts, and R = 135 Ω ADSL NEXT and FEXT (A) Upstream ADSL (A) The figure below shows the power spectral density (PSD) mask for the transmitted signal. The low frequency stop band is defined as the voice band; the high frequency stop band is defined as frequencies greater than 138 khz.

13 PSD (dbm/hz) peak 21.5 db/oct 48 db/oct -90 peak peak +15 dbrn 0-4 khz peak -50 dbm max power in any 1 MHz sliding window above 1630 khz Freq (khz) ADSL Upstream Power mask FREQUENCY BAND (khz) EQUATION FOR LINE (dbm/hz) 0 < f < , with max power in the in 0-4 khz band of +15 dbrn 4 < f < log 2 (f/4) < f < < f < log 2 (f/138) 307 < f < < f < peak, with max power in the [f, f + 1 MHz] window of log 2 (f/1221) + 60) dbm 1630 < f < peak, with max power in the [f, f+1mhz] window of -50 dbm NOTES 1. All PSD measurements are in 100 ohms; the voice band aggregate power measurement is in 600 ohms. 2. All PSD and power measurements should be made at the U-R; the signals delivered to the POTS are specified in ANNEX E. 3. The breakpoint frequencies and PSD values are exact; the indicated slopes are approximate. 4. Above khz, the peak PSD should be measured with a 10 khz resolution bandwidth. 5. The power in a 1 MHz sliding window is measured in 1 MHz bandwidth, starting at the measurement frequency.

14 Downstream ADSL (A) The Figure below shows the power spectral density (PSD) mask for the transmitted signal. The low frequency stop band is defined as the voice band; the high frequency stop band is defined as frequencies greater than MHz. PSD (dbm/hz) peak 21 db/oct 36 db/oct -90 peak peak +15 dbrn 0-4 khz peak -50 dbm max power in any 1 MHz sliding window above 4545 khz Freq (khz) ADSL Downstream Power Mask FREQUENCY BAND (khz) EQUATION FOR LINE (dbm/hz) 0 < f < , with max power in the in 0-4 khz band of +15 dbrn 4 < f < log 2 (f/4) < f < < f < log 2 (f/1104) 3093 < f < peak, with max power in the [f, f + 1 MHz] window of log 2 (f/1104) + 60) dbm 4545 < f < peak, with max power in the [f, f+1mhz] window of -50 dbm NOTES 1. All PSD measurements are in 100 ohms; the POTS band aggregate power measurement is in 600 ohms. 2. All PSD and power measurements should be made at the U-C interface; the signals delivered to the PSTN are specified in ANNEX E. 3. The breakpoint frequencies and PSD values are exact; the indicated slopes are approximate.

15 Above khz, the peak PSD should be measured with a 10 khz resolution bandwidth. 5. The power in a 1 MHz sliding window is measured in 1 MHz bandwidth, starting at the measurement frequency VDSL NEXT or FEXT VDSL NEXT and FEXT is transmission-technique dependent, but the VDSL standard should consider VDSL self-next and VDSL self-fext using the same NEXT and FEXT transfer functions as all of the other DSL crosstalkers earlier in this section. VDSL NEXT and FEXT into other services is included in the considerations for the PSD mask specified in Section more VDSL noise info from Section Noise Masks A & B Noise Masks A and B are the power spectral densities of additive Gaussian noise used in some of the tests in Section 6.5 and appear in the Figures here. These noise models include the background noise levels of Section Thus, generators G3 and G4 shouldn t be applied simultaneously. Noise models A and B possibly reflect European noise and can be used until similar data for US cables is available Noise PSD in dbm/hz Noise PSD in dbm/hz frequency (MHz) Noise Model A for VTU-O (LT) frequency (MHz) Noise Model A for VTU-R (NT)

16 Noise psd in dbm/hz Noise psd in dbm/hz frequency (MHz) Noise Model B for VTU-O (LT) Noise Models A and B frequency (MHz) Noise Model B for VTU-R (NT) For mask A, the table of frequencies and magnitudes are Noise Model A LT NT 300 khz -97 dbm/hz 300 khz -90 dbm/hz 1.1 MHz -103 dbm/hz 700 khz -108 dbm/hz 2.0 MHz -137 dbm/hz 1.1 MHz -112 dbm/hz 2.6 MHz -140 dbm/hz 1.5 MHz -135 dbm/hz 2.4 MHz -140 dbm/hz and For mask B, the table of frequencies and magnitudes are Noise Model B LT NT 300 khz -90 dbm/hz 300 khz -90 dbm/hz 1.1 MHz -83 dbm/hz 400 khz -94 dbm/hz

17 MHz -115 dbm/hz 1.1 MHz -90 dbm/hz 4.9 MHz -140 dbm/hz 2.0 MHz -120 dbm/hz 2.6 MHz -130 dbm/hz 4.6 MHz -140 dbm/hz end of noise model proposal ITU VDSL item loop models The following loop models from section 6.1 of [2] are proposed for G.vdsl requirements: beginning of loop section VDSL Loops (A) The loops in the Figure below characterize distribution node twisted-pairs and will be used for testing and competitive evaluations. The RLCG parameters for the different types of wires used are in the Appendix. The VDSL Loops table below summarizes the purpose of test for each of the 8 VDSL loops. The table below enumerates short-, medium, and long-range values for a nominal length variable in VDSL1 through VDSL4. VDSL Loops No. VDSL0 VDSL1 VDSL2 VDSL3 VDSL4 Rationale null loop (also in ETSI) range stress limit (also in ETSI), underground cable flat-wire vertical drop (also in ETSI), horizontal aerial cable on other section Reinforced-wire vertical drop (also in ETSI), horizontal aerial cable on other section bridged tap, horizontal aerial cable

18 VDSL5 VDSL6 VDSL7 short loop test with bridged taps and various crosstalk (may also sometimes represent CPE wiring) medium loop test with bridged taps and various crosstalk (may also sometimes represent CPE wiring) long loop test with bridged taps and various crosstalk (may also sometimes represent CPE wiring) Nominal Lengths for asymmetric VDSL Loops Variable Name Short Value Medium Value Long Value x (VDSL1) 1000 ft. (304.8m) 3000 ft. (914.4m) 4500 ft. ( km) y (VDSL1) 1500 ft. (457.2m) 3000 ft. (914.4m) 4500 ft. ( km) z (VDSL2) 1500 ft. (457.2m) 3000 ft. (914.4m) 4500 ft. ( km) u (VDSL3) 1500 ft. (457.2m) 3000 ft. (914.4m) 4500 ft. ( km) v (VDSL4) 1000 ft. (304.8m) 3000 ft. (914.4m) 4500 ft. ( km)

19 VDSL0 (null loop) 6.5 ft (2m) TP2 TP1 ~.4mm or 26-gauge - see appendix for rlcg TP2 ~.5mm or 24-gauge - see appendix for rlcg TP3 ~ DW 10 - se e appe ndix for rlc g FP - flat untw isted pa ir - see a ppendix for rlcg VDSL1 (ra nge te st for given data rate) VDSL2 (fla t w ire in drop) VDSL3 x (ft) TP1 ; y (ft) TP2 underground cable z (ft) of TP2 aerial cable u (ft) of TP2 aerial cable 250ft (76.2m) FP (vertical) 250ft (76.2m) TP3 (vertical) (re inforce d tp in drop - "K evin's Castle") VDSL4 (bridge ta p) VDSL5 (short - "The Little De mon") 550 ft (167m) TP2 (underground cable 20-pair) v (ft) of TP1 aerial cable (underground,5- pair) 300 ft (91.4m) TP2 150 ft (45.7m ) TP2 100 ft (30.4m ) TP2 250 ft (76.2m) TP2 (overhe ad ae rial) 150 ft (45.7m) TP2 90 o 50 ft (15.2m) TP2 horizontal 50 ft (15.2m) TP3 horizontal VDSL6 (medium ) 1650ft (503m ) TP1 (underground cable 100-pair) 650ft (198m) T P2 (unde rground cable 100-pair) V D SL 5 VDSL7 (long) 1650ft (503m ) TP1 (dist'n cable 100-pair) 2300ft (701m ) TP2 (underground cable 100-pair) VD SL5 VDSL test loops.

20 Insertion loss of Test Loops.

21 (length labels are approximate see Table above for exact values)

22 - 22 -

23 Delay of VDSL test loops. (length labels are approximate see Table above for exact values

24 Impedance of test loops. (length labels are approximate see Table above for exact values

25 Effect of Bridged Taps Advisory Note: Bridged taps may have a very significant effect on reach in VDSL. A bridged-tap is an open-circuited twisted pair, which is connected in parallel with a working loop. Loops VDSL4-VDSL7 all contain bridged taps. At the bridging/splitting location, the signal separates into two components. The component traversing the bridged-tap line is reflected at the open circuit and recombines with the first component. At those frequencies for which the reflected component is approximately 180 degrees different in phase than the first component, the two components destrutively intefere. This destructive interference manifests itself as a notch at all such frequencies in the channel insertion-loss characteristic. These notch frequencies correspond to those for which the length of the bridged tap is one-quarter of a wavelength. More precisely, assuming d = length of the bridged tap (ft) c=speed of light, m / s v=phase velocity in the twisted pair = the wavelength ε r relative dielectric constant of the loop 1 khz Frequency 1 MHz 100 MHz Polyethylene PVC Pulp (paper) relative permittivity The wavelenth = v/f and velocity v = c = c ε µ ε r r r lead to an expression for the frequencies (in MHz, polyethylene) at which the bridged-tap length is an odd number of quarter wavelength(s) that is equal to f notch = m d d ( 2 m + 1) = = 1,2,... feet For VDSL, many of these frequencies are less than 30 MHz, the upper limit for transmission, leading to severe notching within the transmission band. Furthermore, long bridged taps introduce up to 3.5 db degradation in meters

26 insertion loss across the enitre band. The insertion loss characteristics of VDSL 4,5,6 and 7 appear above and illustrate the effect end of proposed loop section It is further proposed that the loop RLCG models from the appendix of [2] be also used for ITU G.vdsl at least as a regional specific annex for America, if not world wide. 2.7 ITU VDSL items VDSL testing The following text from Section 6.5 in [2] is proposed for G.vdsl testing, and may also be appropriate for G.test from section Testing Methods (A) The testing methods described in this section is to help define and compare standard-compliant VDSL products. Detailed performance tests on actual loop plants are encouraged for actual deployment of VDSL systems. In particular, noise injection, which is assumed throughout this section, may or may not be representative of all field conditions with induced crosstalk Method of Performance Measurement The purpose of this section is to provide an unambiguous specification of the test set-up, the insertion path, and to define signal and noise levels. The tests focus on the noise margin, with respect to the crosstalk noise or impulse noise levels when VDSL signals are attenuated by standard test-loops. This noise margin indicates the amount of increase of crosstalk noise or impulse noise level that tolerates under operational conditions while still ensuring sufficient transmission quality. NOTE: the interpretation of noise margin, and the development of deployment rules based on minimum margin requirements under operational conditions, are not the responsibility of transceiver manufacturers. Nevertheless, it is recommended that manufacturers provide Network Operators with models that enable them to perform reliable predictions on transceiver behavior under deviant insertion loss or crosstalk conditions. Different linecodes or duplexing techniques may behave differently. The figure below illustrates the functional description of the test set-up. It includes: The test loops, as specified in the Table;

27 An adding element to add the impairment noise (a mix of random, impulsive and harmonic noise), as specified in the Table; A high impedance, and well balanced (e.g., better than 60 db across the whole VDSL band) differential voltage probe connected with level detectors such as a spectrum analyzer or a true RMS volt meter. Test Loop VDSL0 should always be used for calibrating and verifying the correct settings of noise (impairment) generators when performing performance tests. The two-port characteristics (transfer function, impedance) of the test-loop is defined between port Tx (node pairs A1, B1) and port Rx (node pair A2, B2). The consequence is that the two-port characteristics of the test cable the figure below must be properly adjusted to take full account of non-zero insertion loss and non-infinite shunt impedance of the adding element and impairment generator. This is to ensure that the insertion of the generated impairment signals does not appreciably load the line. Test Loop A1 A2 Modem + Splitter Tx Test "Cable" Adding Element Rx Modem + Splitter B1 B2 Voltage Probe Impairment Generator Level Detector Functional Description of the Set-up of the Performance Tests The balance about earth, observed at port Tx, at port Rx, and at the tips of the voltage probe should exhibit a value that is 10 db greater than the transceiver under test. This is to ensure that the impairment generator and monitor function does not appreciably deteriorate the balance about earth of the transceiver under test. The signal flow through the test set-up is from port Tx to port Rx, which means that measuring upstream and downstream performance requires an interchange of transceiver position and test cable end. The received signal level at port Rx is the level, measured between node A2 and B2, when port Tx as well as port Rx are terminated with the VDSL transceivers under test. The impairment generator is switched off during this measurement. The transmitted signal level at port Tx is the level, measured between node A1 and B1, under the same conditions.

28 The impairment noise should be a mix of random, impulsive and harmonic noise, as defined in the table. The level that is specified in the table is the level at port Rx, measure between node A2 and B2, while port Tx as well as port Rx are terminated with the design impedance, R v (100 Ohms). These impedance should be passive when the transceiver impedance in the switched-off mode is different from this value. The signal and noise levels are probed with a well balanced differential voltage probe, and the differential impedance between the tips of that probe should be higher than the shunt impedance of 100 k in parallel with 10 pf. Error! Reference source not found. shows the probe position when measuring the Rx signal level at the LT or NT receiver. Measuring the Tx signal level requires the connection of the tips to node pair [A1, B1]. NOTE: The various levels (or spectral masks) of signal and noise that are specified in this document are defined at the Tx or Rx side of this set-up. The various levels are defined while the set-up is terminated, as described above, with design impedance R V (100 Ohms) or with VDSL transceivers under test. Probing an RMS-voltage U RMS ( V ) in this set-up, over the full signal band, means a power level of P [dbm] that equals: 2 U 10log RMS P = dbm. Rv Probing an RMS-voltage U RMS ( V ) in this set-up, within a small frequency band of f (in Hertz), means an average spectral density level of P[dBm/Hz] within that filtered band that equals: 2 U log10 RMS P = dbm/hz. Rv f The bandwidth f identifies the noise bandwidth of the filter, and not the 3 db bandwidth. Next the crosstalk noise level of the impairment generator as defined in the table should be increased by adjusting the gain of an amplifier, equally over the full VDSL frequency band, until the bit error ratio is higher than This BER will be achieved at an increase of noise of x db, with a small uncertainty of x db. The noises will be added with a noise generator that functionally has seven sub-generators as shown in the figure above (this model is also used by ETSI). Some points are necessary to interpret the function of these generators: 1. The seven impairment subgenerators G1 to G7 synthesize noise as in Section 6.2, independent from test loops and bit rates. 2. The transfer function H1( f, d) measures the length- and frequencydependence of the NEXT coupling and is roughly proportional to 2ς d [ 1 e ] f.75 f, where ς is a constant representing roughly the slope of the attenuation constant for the skin effect, or α = ςf.

29 The transfer function H 2 ( f, d) measures the length- and frequencydependence of the FEXT coupling and is proportional to d ς f d f e. 4. Switches S1 S7 determine whether or not a apecific noise generator contributes to the total noise in any specific test. 5. Amplifier A1 allows scaling of the noise over the entire band for noise margin tests. NEXT FEXT G1 G2 ( f d ) H, 1 ( f d ) H, 2 S1 S2 A1 Background Noise (A&B), cable independent WGN G3 G4 S3 S4 Inject into channel output G5 AM RF S5 HAM RFI G6 S6 Impulse G7 S7 Noise generators and composition of noise for measurements. The table below enumerates the testing situations for evaluation and acceptance of VDSL transmission designs or modems. All tests are for a probability of error of 1e-7 as specified in Section 5.2 and noises are further described in detail in Sections 6.2 and 6.3. If more than one noise source is listed, all are injected simultaneously and margin is measured on the sum of the noises. Different transmission techniques may lead to different levels of VDSL self NEXT or FEXT and so the following tables should be construed in terms of the transmission technique being tested.

30 This table lists the tests. For each set of tests, the variable is not the amount of noise, but rather the length of test loop, where x is for TP1 and y is for TP2 and are in 150 foot increments.

31 VDSL Performance Test Summary TEST NAME Loop no. Downstream rate Upstream rate Noise(s) 0. NULL VDSL0 modem maximum 1. RANGE TESTS 1.1 Short Asymmetric Medium Asymmetric Long Asymmetric Short Symmetric Medium Symmetric RADIO TESTS 2.1 Short Asymmetric Medium Asymmetric Long Asymmetric Short Symmetric Medium Symmetric 13 from Table 1 modem maximum from Table 1 none 52 Mbps 6.4 Mbps WGN 26 Mbps 3.2 Mbps WGN 13 Mbps 1.6 Mbps WGN 26 Mbps 26 Mbps WGN 13 Mbps 13 Mbps WGN 20 VDSL FEXT 20 VDSL FEXT 20 VDSL FEXT 20 VDSL FEXT 20 VDSL FEXT 52 Mbps 6.4 Mbps WGN, RFI 20 VDSL FEXT 26 Mbps 3.2 Mbps WGN, RFI 20 VDSL FEXT 13 Mbps 1.6 Mbps WGN, RFI 20 VDSL FEXT 26 Mbps 26 Mbps WGN, RFI 20 VDSL FEXT 13 Mbps 13 Mbps WGN, RFI FEXT

32 XTALK NOISE TESTS 3.1 Short Asymmetric Medium Asymmetric Long Asymmetric Short Symmetric Medium Symmetric 13 4.OPTIONAL Bridged-TAP tests 4.1 Short Asymmetric Medium Asymmetric Long Asymmetric Short Symmetric Medium Symmetric 13 With 50 BT, CPE With 50 BT, CPE or With 50 BT, CPE With 50 BT, CPE With 50 BT, CPE 52 Mbps 6.4 Mbps Noise A & B20 VDSL FEXT 26 Mbps 3.2 Mbps Noise A & B20 VDSL FEXT 13 Mbps 1.6 Mbps Noise A & B20 VDSL FEXT 26 Mbps 26 Mbps Noise A & B20 VDSL FEXT 13 Mbps 13 Mbps Noise A & B 52 Mbps 6.4 Mbps AWGN 26 Mbps 3.2 Mbps AWGN 20 VDSL FEXT 20 VDSL FEXT (from Test 1.1) 20 VDSL FEXT (from Test 1.2) (ETSI) 13 Mbps 1.6 Mbps AWGN 26 Mbps 26 Mbps AWGN 13 Mbps 13 Mbps AWGN 20 VDSL FEXT (from Test 1.3) 20 VDSL FEXT (from Test 1.4) 20 VDSL FEXT (from Test 1.5) end of testing proposal

33 ITU VDSL items OAMP The entire text of Section 7 of [2] is proposed for issue 10.1 in G.vdsl issues list. (text not repeated here, but available on web). 3. New Issues for G.vdsl from [2] 3.1 Performance requirements The following text from Section 5.1 on Performance requirements is proposed for a new issue in G.vdsl: beginning of proposal Performance measures (A) The probability of error should be lower than Pe=10-7 measured according to measurement methods described in Section 6.5 with 6 db of margin on the loops and in the presence of the noises specified in Table 6.5. The measurement period shall be at least 30 minutes and the amateur radio interferer shall visit each amateur band at least twice (at different frequencies within the band) during the test period. A long term performance test shall be performed for a period of not less than 24 hours to ensure long-term temporal stability. Errored seconds, cell loss, and cell delay are deferred to the VDSL standard. In the presence of multiple noises (except RFI), performance margin will be measured with all summed noises increasing by a common factor. FEC specification was removed from this document and instead relegated to PMD and/or TC standards documents, but the removal of detailed specification does not change the requirements elsewhere specifying architecturally where it should be used. The maximum "fast" path delay approximately 1.00 ms and maximum interleave-path delay ms milliseconds, with ms delay optional. The latency is measured between the hypothetical (logical)

34 interfaces in the system reference diagrams of Section 4, currently called the α and β interfaces. Dual latency is optional, but single latency should have programmed delay in 1 ms increments. Implementations shall provide the means to verify delay between the α and β interfaces for the purposes of laboratory design qualification testing, although this may require additional external hardware and software not provided for normal use. The latency measurement method may be extended to the TPS-TC layer as well as to the γ interfaces, where the elements may for instance be ATM cells. Timing and synchronization requirements should be specified in the standards document because they are dependent on transmission technique. Use of the 20 ms latency choice should allow correction of noise bursts up to 500 s in length, while use of the 10 ms latency choice should allow correction of noise bursts up to 250 s New Issue on Activation and power control The entire section 5.4 of [2] is proposed for G.vdsl requirement. 4. Summary: This VDSL contribution makes specific consensus, American VDSL System Requirements proposals to the ITU. The intent is for all to be discussed and made requirements or goals as appropriate. Some new items are also proposed for addition to the VDSL issues list. 5. References [1] "G.vdsl:Updated 'Issues List' for G.vdsl," ITU Temporary Document MA-021R1, SG15/Q4, Melbourne Australia, April 2, [2] ANSI T1E1.4 Contribution R8, "VDSL System Requirements Document," December 1998 (Editor J. Cioffi).