COMMUNICATIONS STANDARDS REVIEW Volume 12, Number 16 May 3, 2001 REPORT OF Q4/15 RAPPORTEUR’S MEETING ON xDSL NETWORK ACCESS TRANSCIEVERS, APRIL 9 – 13, 2001, IRVINE, CA The following report represents the view of the reporter and is not the official, authorized minutes of the meeting. Q4/15, xDSL Network Access Transceivers, Rapporteurs Meeting, April 9 – 13, 2001, Irvine, CA..2 G.gen, including G.lite.bis and G.dmt.bis...................................................................................2 Performance Improvement Methods.....................................................................................2 Conformance Testing............................................................................................................3 Coding..................................................................................................................................4 Modulation...........................................................................................................................7 Initialization..........................................................................................................................7 Framing................................................................................................................................9 OAM ................................................................................................................................10 Q-Mode (Quiescent Mode)................................................................................................10 POTS Splitter.....................................................................................................................12 Phone Line Compatibility...................................................................................................12 G.shdsl.bis................................................................................................................................13 Test Measurements.............................................................................................................13 G.vdsl.......................................................................................................................................13 G.pnt.........................................................................................................................................14 G.hs.ter.....................................................................................................................................14 G.ploam.bis..............................................................................................................................16 G.test.ter...................................................................................................................................16 G.ref.ter....................................................................................................................................16 G.voice......................................................................................................................................16 Q4/15 Partial Attendance Roster, April 9 – 13, 2001, Irvine, CA...............................................20 Acronym Definitions......................................................................................................................22 Communications Standards Review Copyright Policy....................................................................24 May 3, 2001 Vol. 12.16 Copyright © CSR 2001 1 COMMUNICATIONS STANDARDS REVIEW REPORT OF Q4/15, XDSL NETWORK ACCESS TRANSCEIVERS, RAPPORTEURS MEETING, APRIL 9 – 13, 2001, IRVINE, CA, USA The scope of Q4/15 is the development of Recommendations related to digital subscriber line (DSL). R. Stuart (3Com) is the Q4/15 Rapporteur. IC-000 is the document list. IC-001 is the agenda. IC-002 is the report of the February 2001 Q4/15 Rapporteurs meeting at SG15 in Geneva, Switzerland. IC-004 provides the Q4/15 Rapporteur meeting electronic document submission guidelines. IC-005 is the Q4/15 attendance roster. Technical editor’s note: the issues addressed in the papers below remain open unless otherwise indicated. IC-068 is a notice that IBM has submitted a Patent Statement and Licensing Declaration (Annex 2 to TSB circular 245) to the TSB with Patent Policy Option 2 (paragraph 2.2) checked (nondiscriminatory basis and on reasonable terms and conditions). The database of statements is available at: <http://www.itu.int/ITU-T/patent/>. IC-089 (T. Starr, SBC; H. Sadjapour, AT&T; H. Taylor, BT) proposes to decide upon, specify, and prepare for ITU Determination in October 2001 as many improvements as possible for ADSL performance, robustness, diagnostics, power consumption, and backwards and forwards interoperability. The following improvements are urgently needed and will greatly assist service providers in making service available to a greater number of customers in a cost effective manner: • • • • • • • Higher ADSL bit-rates for loops up to 15 kft, both downstream and upstream Greater ADSL loop-reach for service rates of 384 kbit/s downstream and above Improved robustness against bridged taps Improved robustness against radio frequency ingress Expanded and standardized ADSL loop diagnostics such as reporting loop conditions from the far-end equipment Reduced ADSL power consumption to permit higher equipment density, especially for equipment at NGDLC-RTs Improved clarity of standards to facilitate improved ADSL interoperability G.gen, including G.lite.bis and G.dmt.bis T. Cole (AMD) is the G.lite.bis Associate Rapporteur. IC-U18 is the G.lite.bis issues list, based on contributions, discussion, and agreements as of the close of the February 2001 Geneva meeting. IC-R18 is the G.lite.bis draft Recommendation. IC-A18 is the G.lite.bis work program. F. Van der Putten (Alcatel) is the G.dmt.bis Associate Rapporteur. IC-U17r1 is the G.dmt.bis issues list, based on contributions uploaded and discussed as of the close of the February 2001 SG15 meeting. IC-R17 (F. Van der Putten, Alcatel, Editor) is the current draft of the G.dmt.bis Recommendation; revision marks reflect the changes agreed at the Clearwater interim meeting in January. At the August 2000 meeting, it was agreed that G.dmt.bis should reference other recommendations (e.g., G.lite.bis, existing, or to be created Recommendations) to avoid duplicating text in common with G.lite.bis. In a few places, this G.dmt.bis draft indicates differences with G.lite.bis. Those differences are only temporarily listed in this G.dmt.bis draft for clarification of the technical developments within Q4/15; they will be moved to G.lite.bis when appropriate. Performance Improvement Methods IC-069 (E. Eleftheriou, S. Ölçer, IBM; M. Sorbara, M. Eyvazkhani, GlobeSpan) considers the problem of bit-to-symbol mapping for LDPC coding in ADSL, and explores the merit of including 2 Vol. 12.16 Copyright © CSR 2001 May 3, 2001 COMMUNICATIONS STANDARDS REVIEW uncoded information bits, along with code bits, for the selection of transmit symbols. It first shows that, for bandwidth-efficient modulation, the inclusion of uncoded information bits allows higher net coding gains to be achieved as compared to the case where symbol mapping is obtained via LDPC code bits only. Second, it shows that for transmission over a channel with burst or impulse noise, the uncoded bits are not more vulnerable to detection errors than the code bits. It concludes that symbol mapping for large constellation sizes should employ code bits as well as uncoded bits. IC-071 (E. Eleftheriou, S. Ölçer, IBM; M. Sorbara, M. Eyvazkhani, GlobeSpan) presents a new reduced-complexity decoding algorithm for low-density parity-check codes that operates entirely in the log-likelihood domain. The computationally expensive check-node updates of the sum-product algorithm are simplified by using a difference-metric approach on a two-state trellis and by employing the dual-max approximation. The dual-max approximation is further improved by using a correction factor that allows the performance to approach that of full sum-product decoding. IC-084 (A. Carlson, Broadcom) reports on the results of several simple simulations involving lowPAR signals. It considers the effects of interpolation, filtering, and inclusion of the cyclic prefix. It was presented for information. It notes that when comparing absolute performance of various lowPAR signal proposals, results may vary by a few dB depending upon the detailed underlying assumptions that are made in each case. At the same time, as a general rule it seems that a change that decreases the PAR of a signal in a given case will decrease the PAR in most other cases as well. IC-084 suggests that relative statements are more trustworthy than are absolute statements concerning the PAR of a given class of signals. IC-081 (D. Van Bruyssel, Alcatel) proposes, based on the ad hoc requirements, an architecture for the advanced coding supporting a dual latency path configuration. It also proposes specific inner and outer codes. This contribution focuses on the dual latency path case. A very important supplementary requirement which is assumed, is that the first path is a low latency path (“Fast path”) and the second path is a high latency path which needs improved protection against impulse noise (“Slow path”). Possibly, the critical application supported on the fast path is interactive gaming and on the slow path is video streaming. It was agreed to open two questions: Shall the multi-level dual latency solution based on common inner coding be mandatory in transmitters and, for the higher latency path, shall the use of interleaving and Reed Solomon as in G.992 be mandatory as shown in Figure 3 of IC-081 (Proposed Architecture and Classes for Dual Latency Path Configuration)? Should the multi-level dual latency solution with common inter code based on multi-level SCCC (2D or 4D) G.992.1 trellis as shown in Figure 3 and described in Section 4 (Proposed Coding Schemes for the Inner Code) of IC-081 also be mandatory in transmitters? IC-070 (E. Eleftheriou, X. Hu, S. Ölçer, IBM; M. Sorbara, M. Eyvazkhani, GlobeSpan) presents an analytical study that allows comparison of the various schemes proposed for advanced coding in G.dmt.bis and G.lite.bis on the basis of achievable performance. Results show that the channel capacity of coded modulation schemes that employ Gray-labeling based signal mapping is higher than that of coded modulation schemes using a set-partitioning based labeling. IC-070 determines the impact of various Gray-labeling techniques on system performance, giving an indication of appropriate code rates to be used with each approach. It describes the system model on which computations of the performance limits are based, and provides numerical results allowing a comparison of different labeling strategies adopted for coded modulation. Conformance Testing IC-029 (A. Ginesi, Catena) propose definition of a set of test cases where the maximum transmit showtime PSD levels are specified as targets, given a desired data rate. This would allow testing of May 3, 2001 Vol. 12.16 Copyright © CSR 2001 3 COMMUNICATIONS STANDARDS REVIEW the fine gain-based power cut back for excess margin minimization. The objective is to enforce a more consistent application of this power cut back method, which has the benefits of lowering the unnecessary high level of interference into adjacent lines along with transmitter power consumption reduction. It was agreed that G.lite.bis shall specify a set of test cases where maximum transmit showtime PSD levels are specified as targets, given desired data rates. Coding IC-033 (Q. Jin, Mitel Semiconductor) proposes a turbo trellis-coded modulation method for DMT based xDSL system. The method uses turbo trellis coded modulation and Reed-Solomon interleaver to achieve less than 3 dB away from Shannon capacity. The turbo structure is simple and suitable for high data rate transmission such as VDSL. IC-034 (Q. Jin, Mitel Semiconductor) compares the multi-level turbo coding method and the fullturbo coding method for G.vdsl and G.dmt.bis, and G.lite.bis. It concludes that the multi-level coding scheme has performance advantages for high constellation and is less complex in general. On the other hand, the full-coding method is suitable for low constellation and has much smaller latency requirements, especially for low data rate transmission. IC-034 was presented for information. IC-024 (B. Li, A. Deczky, Catena) proposes the “multi-level” turbo TCM concatenated with RS codes as the advanced coding method for the G.dmt.bis and G.lite.bis standards. In this proposed advanced coding method, the RS encoder and “multi-level” turbo TCM encoder are separated by an interleaver. This interleaver randomizes any burst errors and then the randomized errors can be corrected by the RS decoder. See IC-025r1, below, for related agreements reached. IC-027 (B. Li, A. Deczky, A. Ginesi, Catena) compares the data rates of “full” turbo TCM with “multi-level” turbo TCM schemes in three real ADSL loops for both upstream and downstream. It reports that both schemes provide close to the same performance throughout the three test cases. But as the implementation of the “multi-level” turbo scheme is much less complex, it proposes that the “multi-level” coding technique be adopted for advanced coding in both G.dmt.bis and G.lite.bis. It also proposes that the number of bits coded be 2 for even constellations and 3 for odd constellations, as per BI-090r1A, Turbo TCM scheme with low decoding complexity (B. Li, A. Deczky, Catena, Q4/15 Bangalore, October 2000). IC-026r1 (B. Li, A. Deczky, Catena) illustrates the existence of a parity bit mismatch problem for the turbo TCM schemes using a fixed interleaver size. In the ADSL application, a DMT modulation scheme is used. The receiver measures the signal-to-noise ratio (SNR), and does the bit allocation for all the sub-carriers according to its SNR value. Therefore the bit loading is different from sub-carrier to sub-carrier, and data rates are different for different loops. From the point of view of the turbo code design, it is important and desirable to understand whether the interleaver size should be fixed or flexible to be aligned with DMT symbols. IC-026 analyzes the parity bit mismatch problem and shows that there will be some performance degradation for turbo TCM using fixed interleaver size. It proposes that the interleaver size for turbo TCM not be fixed, and that the interleaver size for turbo TCM be a multiple of DMT symbols. IC-041 (A. Torres, V. Demjanenko, F. Hirzel, Vocal Technologies) proposes that the size of the interleaver be fixed to an integer number of DMT symbols to take the maximum advantage of the turbo code principle and to avoid degradation of the high order constellations. 4 Vol. 12.16 Copyright © CSR 2001 May 3, 2001 COMMUNICATIONS STANDARDS REVIEW IC-042 (S. Abbas, Centillium; S. Heidari, Ikanos) proposes to independently enable trellis coding in G.992.1 in each direction. Currently, the trellis coding option is always enabled in both directions. Yet the benefit of trellis coding is not necessarily equal in upstream and downstream directions. Thus a mechanism to enable this option in each direction independently is desirable. IC-043© (S. Abbas, Centillium) proposes that trellis coding be mandatory for both G.dmt.bis and G.lite.bis in the downstream and upstream direction. Advanced coding to gain an extra 2-3 dB should consider the reuse of Reed Solomon and trellis coding as part of the complete scheme to help reduce the overall complexity of the ADSL modems. Finally, trellis coding should be improved to increase the net coding gain for long loops. It was agreed that if section 0 (advanced coding gain) is not accepted into the main body of G.lite.bis, Q4/15 shall mandate G.992.1 trellis coding up/down tx/rx, independently enabled by the receiver during initialization in the upstream and downstream direction. IC-061 (W. Farrell, iCODING Technology; D. Langston, SigmaTel) summarizes current options and issues affecting the selection of a convolutional-based turbo coding scheme as the advanced coding solution to G.dmt.bis and G.lite.bis. The most suitable coding scheme will depend on the relative importance of factors such as hardware and software complexity, coding gain at short and long loop channels, impulse noise protection, and simplicity of design. Given each of these considerations, IC-061 details the most appropriate selection for each, and concludes that the multilevel scheme appears to be the most versatile, especially if a triple code is used, which results in a very low complexity solution. The two main faults of the multi-level scheme are that it does not perform as well for long loop channels and it does not provide the same impulse noise protection. However its advantages of reduced complexity and coding gain performance over a large range of loop lengths and latencies mean it is likely to be the preferred solution. The most viable of full coding solutions is the triple code solution, which results in only 2.2 times the complexity of the multi-level double scheme (not including the RS decoder, required for the multi-level system). It also provides greater performance for long loops (approximately 1 dB, depending on the latency) to extend the reach of the service, and at the same time provides greater impulse noise protection. IC-023 (A. Ginesi, A. Deczky, Catena) presents a proposal for testing improved coding gain schemes with real loops. All the current tests required to assess and compare performance of new coding schemes are quite abstract as they just specify a fictitious SNR profile and ask for coding gain results. To quantitatively appreciate the real benefits of a new coding structure, it proposes to add a set of real loop tests where achievable data rate is reported instead of net coding gain. Two of the three loops proposed are part of agreed tests for performance targets: the 12 kft – 26 AWG and the 18 kft –24 AWG loops. The third test is a lower data rate test and has been added to fully cover the data rate range of interest, for high to low bit rates. It is a 17 kft – 26 AWG loop. All these loops are supposed to have a noise source of –140 dBm/Hz AWGN and 24 DSL cross-talk disturbers. IC-040 (A. Torres, V. Demjanenko, F. Hirzel, Vocal Technologies) provides details on the Euclidean distance, gain/loss, and PAR considerations of the non-square QAM constellations with independent I and Q, proposed in CF-038 (F. Hirzel, J. Torres, V. Demjanenko, Vocal Technologies, Q4/15 Clearwater, January 2001), IC-025r1 (B. Li, A. Deczky, Catena) discusses the feasibility of using advanced coding without Reed-Solomon codes for impulse noise protection. To achieve the same impulse noise protection performance, advanced coding without RS code needs a frame size that is equivalent to the interleaver size of advanced coding with RS code. This will lead to a huge RAM requirement that is not feasible to implement. Using advanced coding concatenated with RS code is a good solution. May 3, 2001 Vol. 12.16 Copyright © CSR 2001 5 COMMUNICATIONS STANDARDS REVIEW IC-025 proposes that turbo codes replace the current TCM while the current RS codes are kept as outer codes. It was agreed that G.lite-bis advanced coding scheme shall consist of replacing the G.992.1 trellis code while keeping the current RS code as outer code as shown in IC-025r1 Fig. 2. IC-035 (Q. Jin, Mitel Semiconductor; H. Sadjadpour, AT&T) proposes a general turbo coding structure for a DMT based xDSL system; it includes both the multi-level coding method and the full-coding method. Using this structure, users can determine how many bits should be coded in each frequency subcarrier. IC-076 (K. Chu, Conexant) addresses programmable trellis codes for ADSL. There has been extensive discussion about incorporating advanced coding schemes into G.dmt.bis and G.lite.bis; much of the discussion has been focused on the technical details of the various Turbo codes. But there has been less discussion about what is needed for ADSL and how best to address this need. Easier and more reliable deployment are the main reasons why improved performance is desirable. Improving performance can make deployment of ADSL more robust and can improve data rate and reach relative to current ADSL systems. IC-076 proposes to use programmable trellis coding for the advanced coding scheme in G.dmt.bis and G.lite.bis. This programmable trellis coding scheme has many advantages including simple encoding structure, and allows trade-offs between performance and decoder complexity. The programmable trellis code fits directly into the existing ADSL coding scheme; the 16-state Wei code used in legacy ADSL becomes a special case. Further, it allows a trade-off between latency and performance, and increased flexibility without introducing interoperability problems. Programmable trellis codes have been used successfully for ANSI HDSL2, ITU G.991.2, and ETSI SDSL. IC-085 (B. Rezvani, S. Heidari, Ikanos) presents a new approach for turbo coding for G.gen.bis applications. The new approach addresses two important system parameters: 1) availability of fast and interleaved path, and 2) good coding gain with minimal complexity. Higher performance is reached by utilizing a channel permuter, and lower complexity is achieved by using a binary encoder. Furthermore, the latency issue associated with turbo codes is addressed by passing the fast data buffer. Due to inherent delay associated with the interleaver / deinterleaver of turbo coding schemes, in this proposal Turbo Trellis Coded modulation is solely used in the interleaved path. This method has the advantage of having similar structure to Trellis Coded modulation of ADSL and therefore from complexity point of view it is a known quantity. In order to reduce the complexity of decoding, IC085 proposes a binary convolutional coding which allows constructing a simpler trellis decoder. Furthermore, by using a simple channel permuter, an increase in the performance is reached. The Turbo Encoder interleaver’s size is proposed to be a programmable value between 1 to 4 kbits, where the user may trade-off the complexity and latency for performance. IC-088r1 (A. Deczky, Catena; M. Sorbara, GlobeSpan; S. Ölçer, IBM; W. Ferrell, iCODING; J. A. Torres, Vocal) proposes, to narrow down code features for advanced coding, that the features listed below be agreed for inclusion as mandatory for advanced coding in G.dmt.bis and G.lite.bis: • • • • • Symbol mapping is based on per-dimension Gray labeling Symbol mapping includes uncoded information bits in addition to coded bits [for small QAMconstellation size (e.g., 16-QAM and smaller), symbol mapping may include code bits only] RS coding according to G.992.1 and G.992.2 is supported Advanced coding scheme represents an inner code when concatenated with outer RS code The block size of the advanced code is variable 6 Vol. 12.16 Copyright © CSR 2001 May 3, 2001 COMMUNICATIONS STANDARDS REVIEW Modulation IC-028 (A. Ginesi, A. Deczky, Catena) presents a proposal for exchanging information on the presence of the first transmit signal image above the Nyquist frequency. Some current G.992.1 and G.992.2 modems implement the transmitter (either the U/S or the D/S) with an IFFT size greater than the one specified by the standard. If the images of the transmit signal are correctly regenerated with period equal to the clock frequency of the standard IFFTs (276 kHz for U/S and 2208 kHz for D/S), there is no difference; the transmit signal is exactly the same as the signal generated by a standard-size IFFT. But this does not occur in some of the G.992.x modems. Even though this does not compromise interoperability, if the receiver is not made aware of the way the transmit signal is generated, inter-performance may be affected, particularly on short loops. IC-028 proposes to have some information exchanged between ATU-R and ATU-C in G.hs to let the receivers know how the transmit signals are being generated. This way the receiver can adapt its signal processing algorithms to adequately process the received signal. In particular, the transmitter should tell the receiver whether or not the transmit signal images with periodicity equal to the clock of the Standard–size IFFT are being generated (through a Standard-size IFFT or with bigger size IFFT plus image restoration). It also proposes that the spectrum beyond the Nyquist frequency be either zero filled or contain the complex conjugate image of the base-band signal. It further proposes that this information be exchanged well before the receiver equalizer is trained, preferably during G.hs. To this end, it suggests, new G.hs code-points need to be defined; it shows a way to implement the required code-points in the standard information field at SPar(2) and NPar(3) levels. It was agreed to exchange information about the presence or not of the first (complex conjugate) image of the transmit signal above the Nyquist frequency between the ATU-R and ATU-C, prior to training the receiver equalizer. Initialization IC-050 (M. T. Arvind, A. Verma, Sasken) proposes a simple procedure for deciding the U/S and D/S transmit power levels during initialization. The procedure can be used to implement politeness cutback, spectral shaping, and crosstalk control, as well as attenuation of particular carriers. It is in line with the current set of agreements on initialization. IC-049 (A. Ginesi, S. McClennon, Catena) proposes to apply the loop dependent power control mechanism described in G.991.1 (G.shdsl) to G.lite.bis and G.dmt.bis modems. It is recognized that a power/PSD limitation of the transmit signal in G.lite.bis and G.dmt.bis provides the benefit of lower power consumption in the transmitters (both ATU-C and ATU-R) and lower cross-talk into adjacent pairs. In a DLC environment, this can be beneficial for the downstream channel, as ADSL from the cabinet may be running in parallel with ADSL from the CO. In a self-FEXT limited environment, a loop length dependent power control for upstream does provide significant benefits as it improves and stabilizes the performance across the subscriber base (See MA-073, Short loop upstream power control in FEXT-limited ADSL Systems, Nortel Networks, Q4/15 Melbourne, Australia, March-April 1999). IC-051 (M. T. Arvind, A. Verma, Sasken) proposes modifications to the initialization proposal (section 0.4.2 of IC-R17) to make the proposal consistent with the current set of agreements on initialization. The modifications also address the editor’s notes. IC-064 (M. T. Arvind, A. Verma, Sasken) presents a comparative study of the following current proposals for G.dmt.bis and G.lite.bis initialization: Section 0.4.2 of draft G.dmt.bis (IC-051) (Sasken), D.28 (Aware), BI-091 (Nortel) and BI-099 (Samsung). It takes the agreed set of requirements in Section 0.4 as criteria for comparison. It compares power cutback, training, and May 3, 2001 Vol. 12.16 Copyright © CSR 2001 7 COMMUNICATIONS STANDARDS REVIEW exchange stages of initialization, and considers overall features, like flexibility, robustness, and length of initialization. This comparison was presented for information. IC-054 (K. Harris, M. Wingrove, Nortel) contrasts and analyzes a number of different initialization proposals, and makes suggestions for each. Due to the complexity of initialization it is difficult to analyze the sequence as a whole, but most of the innovations proposed can be analyzed individually. IC-054 contrasts the positions found in: • • • • • • • • BI-091, New initialization sequence for G.lite.bis and G.gen.bis (K. Harris, Nortel) FI-130R2, A shortened full initialization approach for fast retrain (A. Ginesi, Nortel) CF-070, Initialization proposal for G.dmt.bis and G.lite.bis (M. Arvind, A. Verma, Sasken) BI-099, Basic structure of adaptive length initialization sequence for G.dmt.bis and G.lite.bis (M. Park, Samsung AIT) D.28, Proposals for initialization (M. Tzannes, Aware, SG15, February 2001) BI-096, A simplified initialization procedure for G.lite.bis and G.dmt.bis (B. Barazesh, R. Banerjea, Lucent) BI-111, Report of the Ad hoc on Initialization (T. Cole, AMD; F. Van der Putten, Alcatel) CF-034, Interpretation of spectrum shaping parameters exchanged in handshake to support operation with a notched transmit PSD (I. Sharfer, Tioga) It focuses on the following areas: • • • • • • • Timing recovery Adaptive length equalization and capacity estimation Determination of upstream and downstream cutback Messaging of Bi and Gi Information Rate adaptation Fast retrain Communication of reduction of REVERB tones In the January 2001, Clearwater meeting, it was agreed to improve the randomness of MEDLEY. This new PRBS (Pseudo Random Binary Sequence) generator proposed in IC-030 (A. Ginesi, A. Deczky, Catena) re-uses the structure of the data path scrambler used in G.lite and G.dmt modems. The 23-bit shift register would be used for both R-MEDLEY and C-MEDLEY, and would generate a bit stream with period 223 -1=8388607, i.e., ~16384 DMT signal periods for C-MEDLEY and ~131072 DMT signal periods for R-MEDLEY. IC-094 (Alcatel, Catena, Nortel, Sasken) presents a multi-company consensus proposal for G.dmt.bis and G.lite.bis initialization. It was agreed to adopt IC-094 as working text (and remove the proposed text from section 0). IC-022 (A. Ginesi, A. Deczky, Catena) discusses bandwidth requirements of the REVERB signals used for equalizer training. It is well known that the choice of bandwidth of the REVERB signals (both for U/S and D/S) may have a great impact on the performance of DMT modems. In G.992.x standards the set of tones to be used during REVERB is transmitter-discretionary, within the recommendation that “…it should be sufficiently wide-band in order to probe the transmission channel in such a way that the REVERB signal does not sufficiently lengthen the estimated channel impulse response….” IC-022 proposes that: • • 8 The set of transmitted tones to be used during REVERB shall be determined by the receiver, for each direction. The transmitter shall also be able to transmit the Nyquist bin. A TBD real number shall be used to modulate the Nyquist bin. Vol. 12.16 Copyright © CSR 2001 May 3, 2001 COMMUNICATIONS STANDARDS REVIEW • The cut back that can be applied to any tone during REVERB shall be less than TBD dB. The max cut back shall be set to allow proper channel identification. Framing IC-038 (J. Carlo, B. Wiseman, K. Kratochwil, Texas Instruments) lists current agreements on mandatory and optional framing parameters and proposes that the G.dmt.bis and G.lite.bis standards agree to mandatory and optional framing parameters. Specifically, it proposes the following: • • • • • • • • Mandatory FEC redundancy values on non-interleaved latency path shall be zero. Mandatory FEC redundancy values on interleaved latency path shall be 0, 2, 4, 6, 8, 10, 12, 14, 16. Mandatory Mp values shall be 1, 2, 4, 8, and 16. Mandatory Bp,n values shall be integer and satisfy: 1 ≤ ∑ p,n Bp,n ≤ 255 Mandatory Tp values shall be all positive integers less than or equal to 68. Mandatory support of two latency paths is TBD. In multiple latency mode, the interleaved latency path shall have the highest latency path index p = (N LP –1). Do not restrict the interleaver length to be the same value as the length of the codeword; instead, define the interleaving length according to G.992.1. It was agreed that G.lite.bis framing requires support of R= 0, 2, 4, 6, 8, 10, 12, 14, 16 for the mandatory latency path in each direction (tx/rx). IC-044© (S. Abbas, Centillium) proposes that the mandatory interleaver memory requirements be reduced in each direction for the second generation ADSL modems. Due to interactive gaming and similar applications that customers are running on the ADSL modems, undesirable delay due to interleaving is being experienced. Therefore, a number of service providers are not using the required (mandatory) long interleaver depths. Current ADSL modems support up to D=64 in the downstream and up to D=16 in the upstream directions. For a maximum interleaving block size of 256 bytes, this corresponds to a significant amount of memory, which requirement, in turn, translates into a significant portion of the ADSL chipset die size (for one chip memory implementations) and power consumption of the ADSL chip sets. This requirement appears to be excessive and should be reduced to a reasonable number to bring down the power consumption and complexity of the ADSL modems. Reduction in memory requirement may impact the impulse noise immunity for short-loop high bit rate connections, where largest values of D and RS codeword length are practically used, and consequently require largest memory. The exact number for reduced memory should be studied further. IC-052 (F. Van der Putten, K. Demuynck, Alcatel) proposes a mapping of overhead bytes into the sync-byte of each MUX data-frame, defining two additional parameters. It also proposes a layout for the indicator bits. Until now, it has been agreed that the CRC is sent in the same latency path it relates to (BI-106, Framing and overhead ad hoc report, Q4/15 Bangalore, October 2000). The other overhead information is carried in a bit-oriented structure or by messages. Some proposals have already tried to define a mapping of the overhead into the framing structure. But as they all still refer to “Fast” and “Interleaved” path (especially when defining the indicator bits), they cannot be used for the second generation ADSL modem as these paths are no longer explicitly defined. IC-051 proposes an approach that can be used when a maximum of four latency paths are defined. IC-037 (D. Mahesh, Sasken Communication Technologies) considers the issue of carrying packets over ADSL service, based on the generic ADSL TC architecture. It proposes that the TPS-TC May 3, 2001 Vol. 12.16 Copyright © CSR 2001 9 COMMUNICATIONS STANDARDS REVIEW layers provide a method to carry packet-based traffic over the ADSL physical layer. IETF RFC1662 (PPP in HDLC-like Framing, July 1994) recommends specifications to carry packets over bit-oriented and byte-oriented synchronous links. RFC-1662 recommends the usage of an HDLClike framing structure to transport PPP packets; this method can be extended to carry packets over the ADSL link. It was agreed that G.lite.bis should (as a goal) have shared text for packet TPS-TC with G.vdsl. IC-039 (J. Carlo, B. Wiseman, F. Chow, K. Kratochwil, Texas Instruments) proposes a framing scheme for the overhead channel, which combines the advantages of proposals BI-094, ADSL overhead structure with a single and with multiple PMS-TC functions (F. Van der Putten, Alcatel), and BI-048, Proposal for a multi-latency framing for support of simultaneous transport of data and channelized voice over ADSL bis modems (ADC, Aware, Centillium, Cisco, Ikanos, Legerity, Siemens, SigmaTel, Texas Instruments), from the Q4/15 October 2000 meeting. The proposed framing scheme decouples the insertion of overhead bytes from the Mux Data Frame and thus achieves that no frame overhead rate is wasted on latency paths that only carry CRC and path-related Indicator Bits in the overhead channel. It proposes that if the ADSL modem operates in multiple latency mode and one latency path is used to carry CVoDSL, then the HDLC-framed overhead data is carried in the latency path that carries CVoDSL. It proposes that support of all positive integers less than or equal to 68 for parameter Tp is mandatory. (See also companion proposal IC-038, above.) This proposed framing of the overhead channel satisfies all the requirements outlined in BI-106, the Framing and Overhead ad hoc report. It was agreed to amend the Q4/15 framing structure to include the control parameter Tp on each latency path as shown in Figure 1 of IC-039 (Proposed framing of the overhead channel for dual latency mode and two frame bearers). Valid range of Tp is ≥1. OAM IC-053r1 (K. Harris, Nortel) proposes the addition of the physical layer protocol for ADSL and ADSL.lite to retrieve from ATU-R during showtime the in-band downstream channel frequency response H(f), the noise N(f) measured at initialization, and the signal to noise ratio SNR(f) measured during showtime on a per bin basis. It also proposes to allow for retrieval of similar inband information in the upstream direction from ATU-C to enhance service maintenance and diagnostics. The diagnostic parameters currently defined in G.992.1 and G.992.2 are not specific enough for accurately determining line quality under certain circumstances. Two agreements were reached: A mechanism shall be provided to transfer information from which the channel transfer function per bin and noise PSD per bin can be derived into the ATU-C MIB. A mechanism shall be provided to transfer information from which the ATU-C and ATU-R showtime SNR per bin can be derived into the ATU-C MIB upon request. Q-Mode (Quiescent Mode) IC-036 (M. Tzannes, A. Friedmann, Aware) describes a receiver-transparent Q-mode that allows the transmitter to enter and exit Q-mode as desired while the receiver is unaware of the Q-mode state, and proposes that it be specified for G.dmt.bis and G.lite.bis. The benefit of this proposal is that it provides all the power saving advantages of the current Q-mode proposals without requiring the receiver to implement the complexities of detecting the entry and exit Q-mode symbols. Furthermore, it is compatible with the current Q-mode proposals on several issues, including allowing the transmitter to define the Q-mode signal and allowing a receiver to choose to operate as 10 Vol. 12.16 Copyright © CSR 2001 May 3, 2001 COMMUNICATIONS STANDARDS REVIEW described in the current Q-mode proposal in G.dmt.bis (i.e., detecting the entry and exit Q-mode symbols). It was agreed that if G.lite-bis specifies Q-mode, the scrambler should be removed from the PMSTC layer and be replaced by an equivalent function in the PMD layer IC-048 (P. Reusens, D. Van Bruyssel, M. Beck, Alcatel) describes a method to obtain a Q-mode “filler” symbol resembling REVERB, with a reduced peak-to-average ratio (PAR). This method optimizes the PAR of the Q-mode signal while conserving pure 4-QAM constellation points on all carriers, thus allowing simple detection of the Q-mode and simple phase-locking at the receiver side. Through the reduced PAR, the power dissipation at the transmitter side will be considerably lower during Q-mode than during regular showtime. It was agreed that the definition of a tolerance (TBD) on the amplitude only for the stationary and non-stationary signals shall be added to the proposal as an additional proposal to that agreed in IC062r1. IC-062r1 (I. Sharfer, Tioga) proposes a detector for the flexible semi-stationary Q-mode signal, previously proposed as a means of reducing the non-stationary interference effects while allowing significant power saving at the transmitter. IC-062r1 also shows several results of performance analysis; these results indicate that the proposed detector can reliably distinguish between the Qmode signal and the normal showtime signal. It also makes some observations on the required number of subcarriers and the distortion level. It was agreed that the proposal be updated with the tolerance on the Q-mode signal to be a specification on the maximum allowed signal-to-distortion ratio as defined in IC-062 section 2.2 (False alarm probabilities for a fixed miss probability). IC-063 (R. Verbin, Tioga) presents a Q-mode signal example that combines the two approaches to design a Q-mode signal: using a general 4QAM signal, and adding deviations to the 4QAM Reverb constellation points. The resulting signal has an extremely low PAR of 4.5 dB and “distortion” low enough to allow reliable detection (as presented in IC-062r1). IC-080 (G. Long, Centillium) considers that the most suitable Q-mode filler symbol should be receiver transparent, should have low par that is achievable at flexible complexity, and should keep the same average TX PSD as showtime to ensure stationary characteristics of the crosstalk it generates. A number of other contributions address these considerations but none of them achieve all of the above features simultaneously. IC-080 proposes the definition of a Q-mode filler signal definition that exhibits all of the above features simultaneously. IC-065© (B. Wiseman, A. Redfern, Texas Instruments) collects a series of issues that, it suggests, must be addressed by proposed Q-mode protocol mechanisms. It was agreed to include the items listed in IC-065© as goals for information to provide on Q-mode proposals. IC-092r1 (A. Carlson, Broadcom) outlines Q-mode issues to facilitate agreement to various parts of the Q-mode proposal currently recorded in Section 0 of G.dmt.bis, and makes suggestions for their resolution. It also proposes that: • • False alarm and miss probabilities be specified for successful transitions into and out of Qmode. While receiver is in Q-mode, each receive TPS-TC passes idle data appropriate to that TPS-TC across its interface. These proposals were added to the open issues list. May 3, 2001 Vol. 12.16 Copyright © CSR 2001 11 COMMUNICATIONS STANDARDS REVIEW IC-096r1 is the report of the Q-mode framing ad hoc group that met to discuss the TPS-TC and PMS-TC changes needed to support the Q-mode proposal; it includes recommended text to be added to Section 0 of G.lite.bis. The following companies attended: Texas Instruments, Broadcom, Tioga, Lucent, Legerity, Alcatel, Aware, Intel, Centillium, and Analog Devices. The agreements reached are: 1. TPS-TC requirements for Q-mode proposal • • • • The transmit TPS-TC shall generate an all zero bearer channel if no user data are available for transmission at the V reference point. For a packet mode TPS-TC investigate “HDLC suspend/abort mode.” For the general STM TPS-TC, it is not clear how to do Q-mode. It is appears to be possible for the CVoDSL STM TPS-TC but the group did not recommend a method. 2. PMS-TC requirements for Q-mode proposal • • The PMS-TC shall pass all zero symbol data to the PMD when all zero data are received from the TPS-TC (no user data at the V reference point) and no overhead information is available for transmission. Input is needed for transmission of NTR during Q-mode. POTS Splitter IC-021 (E. Eckert, Nortel) requests a change to the POTS splitter signature resistance defined in G.992.1 to align it with the definition contained in T1.413; it represents a consensus view of T1E1.4. It was agreed that G.992.1 shall be modified through a corrigendum to show the resistance included in the signature test termination of the CO splitter with a value of 33 kohm (instead of 110 kohm). Phone Line Compatibility IC-067 (M. Maranhao, M. Russel, Cisco) discusses the issue of sealing/wetting current, which has received attention recently within a number of standards groups. The purported benefits of sealing current are based on conventional wisdom from deployments of older technologies; they are not based on real evidence of its impact on current xDSL technologies. Furthermore, the current requirements are inconsistent across different sets of standards/draft standards. IC-067 reports on the results found in the technical literature regarding cable splices, splice deterioration due to exposure to contaminants, contact resistance, the impact of contact resistance variation in data transmission, and sealing/wetting current. It presents evidence, based on experimental results in a sample of phone companies’ loops, that only a relatively small percentage of loops are actually affected by any form of sealing current. Also, 75% of junctions remain unaffected when exposed to accelerated corrosive conditions. Results reported in the technical literature show that sealing currents of the order of milliamperes still result in varying contact resistances of a level that causes unreliable performance of analog modems, ISDN transceivers, and DSL transceivers operating at 800 kbit/s. While no experiments have been done on the effects of changes in contact resistance on other technologies like G.shdsl and ADSL, similar results can be expected, based on: 12 Vol. 12.16 Copyright © CSR 2001 May 3, 2001 COMMUNICATIONS STANDARDS REVIEW • • Evidence remained consistent across three different technologies: analog modems, ISDN, and DSL operating at 800 kbit/s The higher bandwidth and bandwidth efficiency of these newer technologies would make them even more susceptible to changes in contact resistance IC-067 concludes that that further study is needed before the appropriate requirements, if any, for wetting current for current xDSL technologies under consideration can be quantified. G.shdsl.bis S. Blackwell (Centillium) is the G.shdsl Associate Rapporteur. IC-A15r1 is the G.shdsl agenda and work plan. IC-R15© is the final draft text of Recommendation G.991.2 - Single-pair highspeed digital subscriber line (SHDSL) transceivers. IC-U15r2 is the updated issues list for G.shdsl.bis; it reflects contributions to this meeting. Test Measurements IC-032 (R. Goodson, Adtran) proposes additions to G.991.2 §11.4.3 (Total Transmit Power Requirement), allowing for measurement of transmit power in either data mode or activation mode. It also includes text that clarifies the transmit power levels required during activation mode, and proposes that this text be included in the G.shdsl Implementers Guide. IC-057 (R. Goodson, Adtran) proposes additions to the test circuit calibration. The G.shdsl text specifying the test circuit calibration specifies a very strict return loss, which is very difficult to satisfy simultaneously at low and high frequencies. IC-057 proposes that a test circuit return loss which obeys the specification of 11.4.2 (Test Circuit Calibration) in the 3 kHz to 3 MHz range be acceptable. IC-058 (R. Goodson, Adtran) discusses some typographical errors in G.991.2 paragraph B.3.5.6 (Measurement of Crosstalk Noise Margin), and proposes corrections required to reconcile G.shdsl with ETSI SDSL. IC-083 (S. Blackwell, Centillium) proposes to align Annex B (Regional Requirements - Region 2) with Annex A (Regional Requirements - Region 1) by defining calibration in Annex B the same way as it is defined in Annex A; Annex B currently does not define noise margin calibration. IC-090 (S. Blackwell, Centillium) is the report of the ad hoc meeting held during this Q4/15 meeting to discuss plans for document publications related to G.shdsl (G.991.2) and G.shdsl.bis. The ad hoc was attended by representatives of Adtran, Aware, Broadcom, Centillium, Conexant, Intel, and Virata. Discussion centered on the mechanism for publishing corrections and changes to the approved G.991.2 text. The ad hoc proposed development of a G.991.2 Implementers Guide , with papers proposing content delivered by August, and Q4/15 agreement on the Guide at the SG15 meeting in October 2001. G.vdsl S. Palm (Broadcom) is the G.vdsl Associate Rapporteur. IC-R11 is the draft G.vdsl Recommendation. IC-U11 is the updated G.vdsl issues list, as of April 9, 2001. IC-A11 is the G.vdsl agenda and work program. IC-072 (T. Pollet, Alcatel) discusses the interpretation of the reference PSD for VDSL PBO. It was previously agreed to express power back-off in VDSL in terms of a reference PSD, but the current formulation in the standard contains an ambiguity that could lead to problems. In particular, May 3, 2001 Vol. 12.16 Copyright © CSR 2001 13 COMMUNICATIONS STANDARDS REVIEW there is no defense or robustness against overestimations of the electrical length. IC-072 proposes to eliminate possible excessive crosstalk due to this by requiring that the receive PSD on any line does not exceed the reference PSD. G.pnt J. Magill (Agere, formerly Lucent) is the G.pnt Associate Rapporteur. IC-U12 is the G.pnt updated issues list; it includes the agreements from the February SG15 meeting. IC-R12 is the draft Recommendation G.pnt.f, Phone line networking transceivers – Foundation. IC-075 is a liaison from SG9 informing that they have initiated work under their Question 14 on home networks in a broadband environment. The current draft of their developing requirements Recommendation may be found at <http://ties.itu.int/u/tsg9/sg9/td/Geneva-06-032001/TD028r1.doc> (ITU password required). SG9 is aware that Q4/15 is also working on some aspects of home networking. In the interest of harmonization of standardization efforts, SG9 invites Q4/15 to share the status of their work effort and to comment on the SG9 work. IC075attachment1© includes a copy of TD-028r1©, Draft Recommendation J.hnwr Home Networking Requirements for Cable Based Services. IC-091 is the proposed reply to the Q14/9 liaison in IC-075. Q4/15 informs that the SG15 work on home networking was initiated in 1999 and is limited to home networks employing telephone wiring. SG15 initiated this work partly in response to a concern over interference between home networking equipment and DSL transceivers that may use the same wiring or pairs in the same telephone cable. The result of SG15 work to date is Recommendation G.989.1, Home phone line networking transceivers – Foundation. This Recommendation was approved by SG15 in February 2001, and is now available in pre-published form from the ITU: <http://www.itu.int/itudoc/itut/approved/g/g989-1.html>. Q4/15 notes that in the SG9 draft Recommendation, Section 5.2, SG9 includes the use of telephone wiring in the overall scope of their work. Q4/15 asks that SG9 make reference to G.989.1 when they discuss the use of telephone wiring in their Recommendations, such that the specific requirements of this application are understood. Q4/15 notes they are continuing their work and currently have work items on full definition of physical, MAC, and link layers, and requirements for isolation filters for use between the home network and the access network. G.hs.ter L. Brown (Conexant) is the G.hs.ter Associate Rapporteur. IC-U16 is the G.hs.ter updated issues list. IC-R16 is the draft revised Recommendation G.994.1, Handshake procedures for digital subscriber line (DSL) transceivers. IC-020 (Nortel, Adtran, Lucent, Mindspeed, ADC, SBC, Alcatel, GlobeSpan, 3Com) proposes clarification of G.994.1 to specify that the vendor ID field contained in the identification field (I) represents the equipment vendor, not the chip set provider. It also proposes to add additional code points to identify the provider of the chip set or sub-system implementing the physical layer functionality; both identifications are necessary. IC-020 also represents T1E1.4 consensus. IC-082 (S. Palm, Broadcom) analyzes the situation described in IC-020 and proposes a technical solution that does not require a revision of G.994.1. The physical layer chipset vendor ID information would continue in Table 7. A new SPar(1) parameter would be added that supplements the vendor ID information in Table 7/G.994.1. Table 9/G.994.1 contains the SPar(1) coding for 14 Vol. 12.16 Copyright © CSR 2001 May 3, 2001 COMMUNICATIONS STANDARDS REVIEW the identification field; it would be the location of the high order bit for the supplemental information. Table 9.12/G.994.1 would contain supplemental vendor ID information SPar(2) coding. Table 9.12.1.X/G.994.1 would contain the eight octets of vendor ID information encoded into eleven NPar(3) octets. IC-031 is a liaison from the DSL Forum on the issue of the use of the vendor ID in G.994.1 (G.hs). The DSL Forum has identified a number of issues affecting interoperability raised by the vendor ID; they ask that Q4/15 consider extending the vendor ID information that can be sent in G.994.1 and work with the DSL Forum to regularize the use of vendor IDs in G.994.1. It is their hope that once these issues are addressed by the ITU-T, the DSL Forum Testing and Interoperability Group can develop appropriate tests and interoperability recommendations based on the ITU-T’s recommendations for improvements in the definition and use of the vendor ID. The next DSL Forum meeting will take place in Oslo, Norway, June 19 - 22, 2001. IC-055 (K. Harris, Nortel) proposes expansion of the vendor ID. Both G.hs and the EOC register of G.lite and G.dmt define a vendor ID field, but neither specifies the meaning of those fields. In currently deployed modems, these fields have been inconsistently filled with either the chipset vendor, the modem vendor, or left empty. Both the chipset vendor and the modem vendor are desirable pieces of information, as is additional information on software and hardware versions. IC-055 proposes a method of coding the complete set of desired parameters by the addition of G.hs code points. This will allow a timely solution to the existing confusion and allow currently deployed modems to continue to provide their level of performance and interoperability. IC-056 (K. Harris, Nortel) proposes draft text for inclusion into G.hs to clarify and expand the use of vendor ID information; it is based on the principles proposed in IC-055. IC-059 (J. Mueller, 3Com) proposes codepoints for achieving the ITU clarification of the interpretation of the vendor ID field for G.hs requested in IC-020r1. It requests that the vendor ID information be signaled via the identification field and that the chipset vendor ID information be signaled in the standard information field. IC-093 (L. Brown, Conexant) is the report of the ad hoc group that met to discuss and resolve issues associated with the G.994.1 vendor ID fields; it includes the reply to the communication from the DSL Forum (IC-031). The ad hoc recognized that there are three aspects to the issue: clarifications to the text of G.994.1, enhancing interoperability performance, and communication of OA&M information. They came to the following conclusions: Clarification of the G.994.1 text: • • It was agreed to add the following text to § 9.3.3 of G.994.1: – Note: This field should typically identify the vendor of the G.994.1 functionality, whether implemented in hardware or software. It is not intended to indicate the system integrator. – The only purpose of the country code is to identify the country of registry of the provider code. This text should be prepared as a corrigendum with the goal of application of the AAP (Alternative Approval Procedures) at the October 2001 meeting of SG15. Enhancing interoperability performance: • • The ad hoc did not believe that expanding the vendor ID fields will help. It was agreed that, where implementation flexibility exists that may affect interoperability performance, a better course of action is to include code points within an xDSL branch of the G.994.1 code tree (e.g., information relating to FFT length). May 3, 2001 Vol. 12.16 Copyright © CSR 2001 15 COMMUNICATIONS STANDARDS REVIEW • Negotiation/selection of features that are beyond the scope of the xDSL Recommendations are handled by the G.994.1 non-standard information field (§ 9.5). OA&M information: • It was agreed that information such as equipment vendor name, model number, serial number, and CLEI (Common Language Equipment Identification) code is important but falls within the category of OA&M information, and is beyond the scope of G.994.1. This kind of information should be handled within G.ploam and possibly exchanged over an embedded operations channel. G.ploam.bis C. Storry (Alcatel Canada) is the G.ploam.bis Associate Rapporteur. IC-U19 is the updated issues list for G.ploam.bis as of April 12, 2001. G.test.ter M. Tzannes (Aware) is the G.test.ter Associate Rapporteur. IC-R13 is the current draft of the G.test Recommendation. G.test.bis (G.test 2001) was approved at the February 2001 ITU meeting. IC-R13 contains the working text for the next version of G.test; at this point it only contains the new working text that is to be added to G.test 2001. IC-U13 is the G.test issues list. G.ref.ter S. Abbas (Centillium) is the G.ref.ter Associate Rapporteur. IC-R14 is the draft G.ref.ter Recommendation. IC-U14r1© is the G.ref.ter updated issues list, based on contributions and agreements at the end of this meeting. IC-097© is a draft liaison to the DSL Forum requesting permission to use material from approved DSL Forum technical reports in the development of ITU-T Recommendations. G.voice C. Hansen (Intel) is the G.voice Associate Rapporteur. IC-U10© is the G.voice updated issues list. IC-A10 is the G.voice agenda and work program. IC-R10 is the draft G.voice Recommendation. IC-086 (C. Hansen, Intel) presents the minutes of the March 29, 2001, conference call on ATM and G.voice. Attendees to this meeting were: Centillium, Intel, GlobeSpan, Nortel, Infineon, AMD, Alcatel, Aware, Legerity, BT, Cisco, Qwest, Siemens, and Lucent. The focus of the meeting was to consider how and what ATM functions to incorporate in G.voice. The group discussed the general requirements for voice over ATM, which can be found in the DSL Forum document TR-039, Requirements for VoDSL. It was noted that the ATM Forum document AF-VMOA-0145.000, Voice and multimedia over ATM - Loop emulation service using AAL2, also defines a mechanism to transport voice over ATM, but is not DSL-specific and does not employ some of the new ADSL features Q4/15 is currently considering for G.adsl.bis. There are also ETSI documents that relate to this topic (which a meeting attendee agreed to post to the reflector). It was noted that the FS-VDSL group is interested in/considering such proposals. Specifically, the group recognized that VoDSL in terms of ATM and PHY layer interaction may support one of the following three architectures: 16 Vol. 12.16 Copyright © CSR 2001 May 3, 2001 COMMUNICATIONS STANDARDS REVIEW 1) “Big fat pipe”: The ADSL PHY layer is configured during initialization as one pipe. This could be either as a fast or interleaved channel. This is a simple approach and can be easily supported with current G.992.x. Q4/15 may need to describe this scenario better to ensure interoperability. 2) “N fat pipes”: In this mode the G.adsl.bis modem is configured during initialization to simultaneously support both a fast and an interleaved channel. The concept is to use the fast channel for the voice ATM traffic and the interleaved channel for data, video, and other. 3) “Dynamic allocation of bandwidth”: In this system architecture the DSL transceiver would simultaneously support both fast and interleaved channels. But the bandwidth would be dynamically allocated between the channels as required by the voice and other applications. This may require the use of features considered for ADSL.bis, such as DRR and SRA. It was agreed that the first action would be to accomplish baseline interoperability in architecture 1. Once that is completed, work would follow on defining architectures 2 and 3. A concern was voiced that any work beyond 1 would entail considerable effort, as interaction with other standards would be necessary. The point was also raised that as DSL is increasingly being used for media intensive applications, the potentially inefficient mode of architecture 1 will become wasteful. It was also noted that schemes to optimize the PHY layer functions for voice should be considered for all DSLs. IC-045 (Centillium, Texas Instruments, ADC, Aware, Legerity) proposes updates to the text for a common out-of-band mechanism to transport either CAS or CCS signaling messages across an xDSL link for channelized voice, which was agreed for Section 0 of the G.voice in Geneva. An HDLC-based frame structure is proposed to encapsulate both CAS and CCS signaling messages associated with channelized voice. Use of an HDLC frame format provides CRC, multiplexing and variable length framing benefits. IC-095 (C. Hansen, Intel) is the report of the ad hoc on contribution IC-045. The ad hoc was uncertain that the transport capacity with the IC-045 encapsulation is sufficient to support GR-303 and V.5 (remote terminal) requirements due to the HDLC overhead and transport capacity allocated to STM signaling. Much discussion surrounded the applicability of mechanisms to support various voice protocols already present in G.SHDSL to G.voice. It was noted by the group that residential voice and business/SOHO voice applications may differ significantly in their requirements and hence separate mechanisms may be required for G.voice and G.SHDSL. The ad hoc recommended two agreements below and noted the one open issue: Agreed: CCS signaling shall be carried over an octet oriented HDLC frame. Agreed: Update section 0 in G.voice. Open: Should CAS signaling be carried over an octet-oriented HDLC frame for ADSL? IC-046 (S. Abbas, Centillium) proposes an update to the introductory working text for G.voice that was agreed in the January 2001 Clearwater meeting (based on CF-044 [Texas Instruments, ADC, Cisco, Centillium, Ikanos, Siemens]). The update delineates the three voice transport mechanisms commonly known as Channelized Voice over DSL (CVoDSL), Voice over ATM (VoATM), and Voice over IP (VoIP). IC-047 (S. Abbas, Centillium) discusses the G.voice document structure and proposes updates to the currently agreed table of contents. Currently, the main body is proposed to contain a general introduction, classes of voice service functional interface, voice service elements definitions, signaling mechanisms, supported TPS-TCs, and network interactions sections and annexes containing TPS-TC specific descriptions. IC-047 proposes to reorganize the document structure so that the main body contains the truly common information across all DSL Recommendations May 3, 2001 Vol. 12.16 Copyright © CSR 2001 17 COMMUNICATIONS STANDARDS REVIEW and VoDSL scenarios. Annexes for different flavors of VoDSL will describe the normative way to support these across different DSL Recommendations. IC-060r1 (P. Kelliher, Aware; I. Volkening, Infineon; V. Davar, Legerity; B. Wiseman, Texas Instruments) defines a set of basic parameters (based on DSL Forum TR-039) that CVoDSL G.voice must support in an xDSL modem to interoperate with a similarly defined xDSL modem. IC-066© (S. Abbas, Centillium; B. Wiseman, Texas Instruments; R. Kroninger, ADC) proposes a mechanism to transport channelized voice signaling and data across the ADSL STM-TC. The ADSL STM-TC performs cross-layer communications by passing primitives with certain parameters to the layers above and below. ADSL STM-TC configuration for channelized voice should accommodate transport of both signaling and voice data. IC-066© proposes a fixed bandwidth of 32 kbit/s for signaling transport. This capacity has been selected to minimize complexity during rate adjustments and to provide sufficient throughput for both CAS and CCS signaling messages. IC-073 (L. Haster, Ericsson) addresses voice in the next generation network. It gives a brief summary on the direction of future networks and how they will move from today’s TDM/circuitbased into more packet/cell-oriented networks. Such networks can be ATM or IP based. The new packet-oriented network also needs new access types to work more efficiently. The author’s impression of the G.voice terms of reference is to create a framework document that leads to adoption of the exiting access types to the new network. The author notes this as fine, as long as the transmission media is kept compatible. IC-073 also notes that methods for existing PSTN/ISDN to interwork over packetized networks already exist, e.g. VoDSL according to DSL Forum TR-039 (Version 1.1, March 2001): Requirements for Voice over DSL and ATM Forum af-vmoa-0145.000 (July 2000): Voice and Multimedia Over ATM-Loop Emulation Service Using AAL2. IC-074 (submitted by Alcatel) is a copy of the publicly available TR-039 (DSL Forum); it was provided to Q4/15 for information. TR-039 specifies an interoperable end-to-end architecture to support broadband voice and data service over DSL systems operating in either packet-mode or ATM-mode. IC-077 (C. Hansen, Intel) presents the text from a selected number of ITU-T Recommendations (G.114, G.174, G.176, and G.131) that consider the effect of delay introduced during voice transmission that may bear upon the definition of G.voice; it was presented for information. IC-078 (T. Cole, AMD; C. Hansen Intel) proposes a general system architecture as a basis for further definition of the functionality contained in G.voice. The system model makes use of a control and data interface into the TPS layer of DSL as well as new functions of the transceiver layer that are under consideration for G.adsl.bis. IC-078 proposes this system model as a basis for further discussion. It also presents three system modes of operation of increasing complexity (similar to the three architectures offered under IC-086 above). IC-079 (J. Carlo, B. Wiseman, G. Wunsch, Texas Instruments) proposes an updated mechanism for dynamic rate repartitioning (DRR) for channelized voice over DSL (CVoDSL) systems; it was proposed for information text to be added to the G.voice ADSL-specific section. VoDSL may benefit from the re-allocation of bandwidth between voice and data streams at the ADSL physical layer. Applications combining voice over DSL and data traffic benefit from the ability to share bandwidth based on call demand. VoDSL solutions based on ATM or IP multiplexing perform this bandwidth sharing at higher layers. CVoDSL solutions use the DSL physical layer to partition voice and data traffic. To enable similar sharing of bandwidth between voice and data using CVoDSL, DRR has been previously proposed (BI-071, B. Wiseman, Texas Instruments; Cisco, 18 Vol. 12.16 Copyright © CSR 2001 May 3, 2001 COMMUNICATIONS STANDARDS REVIEW PairGain, Q4/15 Bangalore, October 2000), whereby the physical layer bandwidth is repartitioned between voice and data streams based on call demand. The total physical layer bandwidth is not changed. CSR CDs Provide Additional Resources CSR CDs are machine searchable (Adobe Acrobat) and very useful for reviewing past technical issues as well as providing a base of information for prior-art searches. Your company’s patent or legal departments may also find these CDs useful. Quarterly CDs: 3 months of CSR reports on each CD, in an annual subscription $695 to non-subscribers but only $200 as an add-on to current subscribers Annual CDs: 12 months of CSR reports on a CD for each calendar year 1990 to present $695 to non-subscribers, $200 to current subscribers First Decade CD: all CSR reports from 1990 through 1999 on one CD $2,000 to non-subscribers; subscribers please take a $200 discount for each year of subscription during 1990 – 1999 May 3, 2001 Vol. 12.16 Copyright © CSR 2001 19 COMMUNICATIONS STANDARDS REVIEW Q4/15 Partial Attendance Roster, April 9 – 13, 2001, Irvine, CA Dick Stuart, 3Com Q4/15 Rapporteur 3Com 3Com ADC Telecom ADC Telecom ADC Telecom ADC Telecom Adtran Aethra Telecommunicaziom Agere Systems Alcatel Alcatel Alcatel Alcatel Alcatel AMD Analog Devices AT&T Aware Aware Aware Broadcom Broadcom Broadcom Home Networking BT Catena Networks Catena Networks Centillium Communications Centillium Communications Cisco Systems Cisco Systems Conexant Systems Conexant Systems Conexant Systems Conexant Systems Copper Mountain Networks Ericsson Telecom AB Ericsson Telecom AB Ericsson Telecom AB ESS Technology Inc. France Telecom R&D GlobeSpan Hitachi IBM Zurich Res. Lab ICoding Technology Ikanos Communications Infineon Intel Corp. Intel Corp. Legerity Inc. 20 Joe Mueller Richard Stuart Chihchigng Tsai Max Yang Nelson Zagalsky George Zimmerman Richard Goodson Roberto Flaiani John Magill Thierry Pollet Chuck Storry Danny Van Bruyssel Frank Van der Putten Ivo Van Gelder Terry Cole Vladimir Friedman Hamid R. Sadjadpour Arnon Friedman Patrick Kellihor Marcos Tzannes Art Carlson Miguel Peeters Stephen Palm Henry Taylor Andrew Deczky Alberto Ginesi Syed Abbas Steve Blackwell Marcus Maranhao Mark Russell Rammy Bahalul Les Brown Keith Chu George Eisler Rick O’Connell Stefan Allevad Jan Bostrom Mattias Hyll Jordan Cookman Hubert Mariotte Massimo Sorbara Yoichi Tsukioka Sedat Ölçer Wade Farrell Behrooz Rezuani Neal King Chris Hansen Kanna Krishnan Vijay Davar Vol. 12.16 Copyright © CSR 2001 May 3, 2001 COMMUNICATIONS STANDARDS REVIEW Lucent Technologies Lucent Technologies Matsushita Graphic Communication Systems, Inc Mitel Semiconductor NEC Next Level Communications Nortel Networks (Canada) NTT PC-Tel Sasken Communication Tech. Sasken Communication Tech. SBC ST Microelectronics Sumitomo Electric Sumitumo Electric USA Texas Instruments Texas Instruments Texas Instruments Texas Instruments Texas Instruments Tioga Technologies Ltd. Toshiba Virata Corp. Vocal Technologies Ltd. Harry Mildonian Carl Posthuma Keiichi Tomita Gary Jin Hiroshi Okado Todd Pett Kate Harris Koji Kikushima Khashayar Mirfakhraei M. T. Arvind Amit Verma Tom Starr Stephania Boiocchi Masami Ueda Toshio Ooka Steven Bieser Jim Carlo Konrad Kratochwil Arthur Redfern Grant Wunsch Reuven Jlan Yasumasa Kikunaga Bijit Halder Alberto Torres Visit the CSR Web Pages: http://www.csrstds.com The Web Pages include an on-line store (order subscriptions and reports), an updated Telecom Acronym Definitions list, updated meeting schedules, background material on telecom standards and CSR (the company), data sheets on both CSR technical journals, and more. May 3, 2001 Vol. 12.16 Copyright © CSR 2001 21 COMMUNICATIONS STANDARDS REVIEW ACRONYM DEFINITIONS AAL ADSL ANSI ATM ATU-C ATU-R AWG AWGN CAS CCS CLEI CO CRC CVoDSL D/S DLC DMT DRR DSL EOC ETSI FEC FEXT FFT FS VDSL HDLC HDSL ID IETF IFFT IP ISDN ITU ITU-T LDPC MAC MIB MUX NGDLC NTR OA&M OAM PAR PBO PHY PMD PMS PMS-TC POTS PPP PSD 22 ATM Adaptation Layer Asymmetric Digital Subscriber Line American National Standards Institute Asynchronous Transfer Mode ADSL Transceiver Unit - Central Office End ADSL Transceiver Unit - Remote Terminal End American Wire Gauge Additive White Gaussian Noise Channel Associated Signaling Common Channel Signaling Common Language Equipment Identification Central Office Cyclic Redundancy Code Channnelized VoDSL Downstream Digital Loop Carrier Discrete MultiTone Dynamic Rate Repartitioning Digital Subscriber Line Embedded Operations Channel European Telecommunications Standards Institute Frame Erasure Concealment Far End Cross Talk Fast Fourier Transform Full Service VDSL (FSAN Committee) High level Data Link Control High-rate Digital Subscriber Line Identification Internet Engineering Task Force Inverse Fast Fourier Transform Internet Protocol Integrated Services Digital Network International Telecommunication Union ITU Telecommunications Sector Low Density Parity Check Media Access Control Management Information Base Multiplexer Next Generation Digital Loop Carrier Network Timing Reference Operations, Administration, and Maintenance Operations, Administration, and Maintenance Peak to Average Ratio Power Back Off Physical Layer Physical Media Dependent layer Physical Media Specific Physical media Specific - Transmission Convergence Plain Old Telephone Service Point-to-Point Protocol Power Spectral Density Vol. 12.16 Copyright © CSR 2001 May 3, 2001 COMMUNICATIONS STANDARDS REVIEW PSTN QAM RAM RFC RS RT RX SDSL SNR SOHO SRA STM TBD TCM TDM TPS TPS-TC TSB TX U/S VDSL VoDSL Public Switched Telephone Network Quadrature Amplitude Modulation Random Access Memory Designation for an IETF Standard Reed-Solomon (code) Remote Terminal Receive Symmetrical high bit rate Digital Subscriber Line Signal to Noise Ratio Small Office Home Office Seamless Rate Adaptation Synchronous Transmission Mode To be Determined Trellis Coded Modulation Time Division Multiplex Transport Protocol Specific Transport Protocol Specific-Transmission Convergence Telecommunications Standardization Board (ITU) Transmit Upstream Very high speed DSL Voice over DSL The CSR Library Subscribers may order copies of documents shown in boldface type from Communications Standards Review, where not controlled. $50.00 for the first document in any order, $40.00 for the second, and $25.00 for each additional document in any order. Volume discounts available. Please contact CSR. Documents listed with © are controlled documents. These documents are not for sale, but we can provide you with the author’s contact information. ITU and ETSI meeting documents are also not for sale, but we can provide you with the author’s contact information. We have a large library of standards work in process and can help you locate other information you may need. CSR recommends that you obtain published standards from Global Engineering Documents. Tel: 800 854-7179, +1 303 792-2181, Fax : +1 303 397-7935, http://global.ihs.com May 3, 2001 Vol. 12.16 Copyright © CSR 2001 23 COMMUNICATIONS STANDARDS REVIEW COMMUNICATIONS STANDARDS REVIEW COPYRIGHT POLICY Copying of individual articles/reports for distribution within an organization is not permitted, unless the user holds a multiple copy license from CSR. The single user electronic version may be mounted on a server whose access is restricted both to a single organization and to one user at a time. You are welcome to forward your single user electronic copy (deleting it on your system) to another user in your organization. CSR offers an Intranet subscription which permits unlimited copies to the subscribing organization. Year 2001 Standards Committee Meeting Schedules Please see the updated calendar at http://www.csrstds.com/mtgs.html. Communications Standards Review regularly covers the following committee meetings: TIA TR-30 TR-41 Data Transmission Systems & Equipment User Premises Telephone Equipment Requirements ITU-T SG15 WP1 SG15 WP2 SG16 Network Access Network Signal Processing Multimedia ETSI AT TIPHON TM6 Access and Terminals Voice over Internet Transmission & Multiplexing DSL Forum xDSL, Access Technologies Communications Standards Review (ISSN 1064-3907) reports are published within days after the related standards meetings. Publisher: Elaine J. Baskin, Ph.D. Technical Editor: Ken Krechmer. Subscription Manager: Denise Hylen Lai. Copyright © 2001, Communications Standards Review. All rights reserved. Subscriptions: $795.00 per year worldwide, electronic format; $995.00 paper format. Corporate Intranet subscriptions (Corporate license for unlimited copies) are $2,150.00. Submit articles for consideration to: Communications Standards Review, 757 Greer Road, Palo Alto, CA 94303-3024 USA. Tel: +1-650-856-9018. Fax: +1-650-856-6591. e-mail: ebaskin@csrstds.com. Web: http://www.csrstds.com. 32122 24 Vol. 12.16 Copyright © CSR 2001 May 3, 2001