Presentation is loading. Please wait.

Presentation is loading. Please wait.

DSM 1 Dynamic Spectrum Management John M. Cioffi Prof. EE, Stanford U. And all the work by students: Taek Chung, George Ginis, Jeannie.

Similar presentations


Presentation on theme: "DSM 1 Dynamic Spectrum Management John M. Cioffi Prof. EE, Stanford U. And all the work by students: Taek Chung, George Ginis, Jeannie."— Presentation transcript:

1

2 DSM 1 Dynamic Spectrum Management John M. Cioffi Prof. EE, Stanford U. Cioffi@stanford.edu And all the work by students: Taek Chung, George Ginis, Jeannie Fang, Jungwon Lee, Dimitris Toumpakaris, & Wei Yu July 2002

3 DSM 2 Copper has more bw than fiber? l DSL will deliver 100 Mbps symmetric to all l 50 line bundle (500 meters of cat 3 tp) u 50 lines (100 Mbps/line) = 5 Gbps u FTTH shares 622 Mbps among 50 or more homes in PON architecture (even “future” systems are 2.5 Gbps) l So, where is DSL today? u Where is DSL going? fiber Cable of copper pair

4 DSM 3 DSL 200X- (V/M/ADSL) l Data, Voice, Video … 802.11 wireless (802.11,15) or other distribution possible within CPE, 11, 55, 100-500 Mbps VoDSL Interface 10/100 Mbps A/VDSLCloserDSLAMONU/RTFiber (from CO) Enet 1000BT router Enet. IEEE 802.X

5 DSM 4 DSM=Dynamic Spectrum Management l Standard in progress – USA (ANSI / T1E1.4) u Dynamic – loops, xtalk, rates very F Not limited by worst-case static situations u Extension to existing Spec Management F T1.417-2002 F Improve considerably reliability and performance F Cannot have spectrum compatibility at higher speeds without DSM F Crosstalk effect is too strong for static regulation F THIS IS A CRUCIAL turning point for DSL l Goals: 100 Mbps symmetric everywhere u & Improve the range of every rate < 100 Mbps SM (static) VDSL 10MDSL ADSL+ 100MDSL VDSL-2 DSM HDSL / ADSL 2000200120022003 … “EFM”

6 DSM 5 Outline l Crosstalk and Management l Autonomous (de-regulated) DSL l Bonded/Vectored DSL

7 DSM 6 DSL’s Challenge: Crosstalk l Same service provider l Different service provider l Management? u Spectrum – fixed or static F Ok, but performance and reliability severely compromised u Symmetric versus asymmetric PhoneCo=Base(DSLAM) user user NEXTFEXT

8 DSM 7 Some Crosstalk NEXT - Bellcore (attenuation) FEXT – British Telecom (gain) l Xtalk noise increases with frequency

9 DSM 8 Loop Topology l Many other configurations u symmetric/asymmetric mixtures u Data rate mixtures u Shorter lines u Bonding of lines CO DSLs CPE DSL CO DSL CPE DSL LT DSL CPE DSL

10 DSM 9 ADSL DMT Loading Basics (adapts to each line) Frequency Frequency AttenTWISTED-PAIR Bits/chanFrequency Bits/chanFrequency Frequency Atten TWISTED-PAIR with TAP, AM/RF, and XTALK AM xtalk Bits/chanFrequency Bits/chan 800 Million phone lines and growing fast! Bellcore tests, 1993 (ANSI Standard) GTE tests, 1996 ; NSTL, 1996

11 DSM 10 Evolution of Crosstalk / management time Network Central Office Present DSL 2002: autonomous (no coordination) SMC for provisioning, Fault isolation 1-20 Mbps (asymmetric/symmetric) SMC1 SMC2 CLEC DSLAM ILEC DSLAM 1 ILEC DSLAM 2 bibinnddeerrbibinnddeerrnder Network Central Office Content 1 Content 2 ILEC LT Switch router Fiber Or multi- channel DSL twisted pair DSLAMDSLAM 20-100 Mbps (symmetric) SMC Evolving DSL 2002+: BondingVectoring Common DSLAM (LT or RT) binder

12 DSM 11 Outline l Crosstalk and Management l Autonomous (de-regulated) DSL l Bonded/Vectored DSL Network Central Office Present DSL 2002: autonomous (no coordination) SMC for provisioning, Fault isolation 1-20 Mbps (asymmetric/symmetric) SMC1 SMC2 CLEC DSLAM ILEC DSLAM 1 ILEC DSLAM 2 bibinnddeerrbibinnddeerrnder

13 DSM 12 Model: Autonomous Case l Crosstalk between users is “matrix channel” u “Interference channel” in information theory l Can users be compatible without large loss? (SSM – no, DSM - yes)... Sharedbinder H(f) DSL 1 DSL K DSL 2 DSL K DSL 1 DSL 2

14 DSM 13 DSM Multi-user Rate REGIONS l Plot of all possible rates of users u ADSL & some VDSL use DSM – any point is possible l More than 2 users (vector of possible rate-tuples) l Region’s size (volume) can vary with: u each binder u transmission methods (i.e., DMT has much larger regions) R long R short Spectral pair 1 Spectral pair 2

15 DSM 14 ANSI DSM ADSL “Blueprint” Example (T1E1.4/2001- 273, 278) ADSLCOADSLrcvr 10 + X kft ADSLRT 10 kft Line 1 Line 2 5 kft CO Same binder FEXT l CLASSIC field problem with ADSL present-day deployments u Long-line often does not work and field personal have to be dispatched l Improve the performance of line 1 AUTONOMOUSLY (no coordination) using DSM-mode in ADSL modems u Eliminate problem automatically u Greatly extend ADSL range ADSLrcvr fiber

16 DSM 15 Initial Short-line spectra, Long-line turns on DSM ADSL basic spectrum result Short-line spectra LongLinespectra f Downstream Spectra (2 lines) l l Short line yields to long u u No “DSL hogging” l l Enormous improvement on long line u u At expense of reduced rate on short line l l Do so autonomously Intermediate Short-line spectra, Intermediate

17 DSM 16 Verizon Experimental Loop Configuration (T1E1.4/2002-069 – “Bell Atlantic” Test Lab, Maryland) Mainframe Cable Vault Remote Side Mainframe Cable Vault CO side Remote Side Pair 1 of 14 kft, 24 gauge 25-pair binder Pair 16 of 6 kft, 26 gauge 25-pair binder (Simulates RT portion of loop) CO side ATU-C ATU-R Verizon Broadband Integration Lab Simulation of CO-ADSL fed through RT binder with RT-ADSL ATU-C ATU-R Pair 17 of 6 kft, 26 gauge 25-pair binder (Simulates RT-ADSL loop) Remote Side CO ADSL RT ADSL 14 kft 6 kft

18 DSM 17 Crosstalk into pair 17 Pair Noise (dB) Pair Noise (dB) 1 -82.5 13 -84.4 2 -78.1 14 -80.7 3 -74.9 15 -77.7 4 -82.3 16 -63.2 5 -83.0 18 -67.7 6 -84.2 19 -67.5 7 -78.0 20 -76.9 8 -81.0 21 -80.8 9 -75.3 22 -78.7 10 -73.3 23 -78.0 11 -79.6 24 -73.4 12 -82.7 25 -80.7 l Used worst-case FEXT coupling pairs for experiment

19 DSM 18 DSM-ADSL Rate Region (Verizon loops) l Rate-Adaptive produces (9,.1) u Mode used by Verizon nominally in lab l Instead, DSM-mode provides (Short, Long) = u (2, 1.8), (4, 1.4), (6,.9), and (8,.6)

20 DSM 19 Telcordia (Bellcore) Results “Blueprint” Test (T1E1.4/2002-063) ADSL CO ADSL rcvr 15 kft ADSL RT 10 kft (fiber) ADSL rcvr Line 1 Line 2 ADSL RT 10 kft (fiber) ADSL rcvr 5 kft 15 kft CO Same binder FEXT Line 3 l Improve the performance of lines 1 and 3 (which is bad under RA operation) l Line 1 only operates at 300 kbps in “normal” mode ADSLCOADSLrcvr ADSLRT ADSLrcvr ADSLrcvrADSLRT

21 DSM 20 Telcordia DSM-ADSL Rate Regions t1e1.4/2002-063 lines 1 & 3 (2 held at 1.6 Mbps) lines 1 & 2 (3 removed) – 063R1 l No coordination (line 2 held at 6 dB margin, 1.6 Mbps l 4x improvement on Line 1 (was at 300 kbps if line 2 does RA or margin-max training) – like “Verizon” expt. u 4x because 15kft, not 20 kft like Verizon

22 DSM 21 DSM Algorithm (Iterative-Water-filling) NSR(f) S(f) l Water-filling is known optimum on single-user channel l Use DMT, and MINIMIZE POWER for given rate/margin u Best autonomous solution to “interference problem” in binder l The more lines that do DSM, the better they all work u NEVER any worse than existing fixed/static Spectrum Management l AUTONOMOUS – no coorination necessary, procedure assures overall best mutual spectra (least harmful xtalk) of all DSLS in distributed fashion

23 DSM 22 3-steps to Broadband DSL l DSM u ADSL -Software option in current ADSL modems u ADSL+ & 10M-DSL (DMT only) – iterative wf u 100M-DSL (DMT only) ADSL to D: 7 Mbps U:.3 Mbps 1 line 0-6 km AMDSL to D: 25 Mbps U: 10 Mbps 1-2 lines CO/LT 0-4 km BDSL to D: 100 Mbps U: 100 Mbps 1-4 lines/LT 0-2 km … 200220032004 Dynamic Spectrum Management

24 DSM 23 10MDSL (EFM) Range/Rate Goals [1] ratelines Per line Length(kft)Goal(kft) 10 M 1102.5>3.5 10M253.5>4.5 10M33.335>6 10M42.57>8 10M528>9 4M4112>12 2M4.515>15

25 DSM 24 25 10MDSLs in same binder (same length) l No coordination u 10 Mbps single-line range is 2.5x SSM result (SHDSL) u 10 Mbps >1.5 km on 2 lines VerifiedByTelcordia VerifiedByVoyan

26 DSM 25 Example PSD for 4000’ l Overlap (EC) to 800 kHz l Each has 3 basic bands

27 DSM 26 MDSL with ADSL present l Range of ADSL and of MDSL unaffected if DSM is used u Not true for Static SM (or SHDSL)

28 DSM 27 ADSL at CO, MDSL at RT l Cabinet is 10 kft from CO – last 2 to 6 kft in same binder l ADSL at CO “swamped” by 64 PAM l ADSL at CO largely unaffected by IW MM AA Fiber, 10 kft Copper, 4 kft Copper, L kft 12, 14, 16

29 DSM 28 VDSL vs MDSL rate regions l Fixed is 998 on V and 1 MHz on M l Still exceeds all targets of M and V

30 DSM 29 Table Revisted (extra column) l Allows mixture of ADSL with MDSL, and VDSL with MDSL l Rate regions do reduce in size for VDSL u But all VDSL and MDSL goals still exceeded Table 2 – Augmentation of Table 1 with IW results Aggregate bit rate (sym payload) # of twisted pairs Average bit rate per pair Objective loop length Desired loop length DSM Result 10 Mb/s 1 2.5 kft >3.5 kft 5 kft 10 Mb/s 2 5 Mb/s 4 kft >5 kft 8 kft 10 Mb/s 3 3.33 Mb/s 5.5 kft >6.5 kft 9 kft 10 Mb/s 4 2.5 Mb/s 7 kft >8 kft 10 kft 10 Mb/s 5 2 Mb/s 8 kft >9 kft achieved 4 Mb/s 4 1 Mb/s 12 kft >12 kft achieved 2 Mb/s 4 512 kb/s 15 kft >15 kft achieved

31 DSM 30 998 Plan with Iter-water (VDSL) l Rates exceed VDSL objectives l Note spectra converges to 998 when 998 is present (4000’ MDSL)

32 DSM 31 Basic Principles - Summary l No DSL modem should use more transmit power than it needs l No DSL modem should use more bandwidth than it needs l Can be solved adaptively and autonomously u Stop proliferating DSL standards and incompatible spectrum management

33 DSM 32 Basic Solution l Use/Extend ADSL DMT Frequency Grid u Already on 24 million lines (> 95% of market) l ADSL+ = 512 tones down / 128 up l 10MDSL= 1024 tones down/1024 up l VDSL = 2048 up/down l 100MDSL 4096 up/down

34 DSM 33 Outline l Crosstalk and Management l Autonomous (de-regulated) DSL l Bonded/Vectored DSL Network Central Office Content 1 Content 2 ILEC LT Switch router Fiber Or multi- channel DSL twisted pair DSLAMDSLAM time 20-100 Mbps (symmetric) SMC Evolving DSL 2002+: BondingVectoring Common DSLAM (LT or RT) binder

35 DSM 34 Sharedchannel Coordinated 2-sided Signals (vector bonding) l Full Vectoring Problem l Highest data rates Bondedlines Controller Bondedlines Controller

36 DSM 35 Sharedchannel DSM Level 2: Coordinated 1-sided Signals l “broadcast” and “multiple-access” probs in IT u One-sided vectoring, FDM of up/down l Bonding (vector broadcast and vector MA problems in IT) LTDSLAM User 1 User 2 User K... Controller

37 DSM 36 998 with/without Vectoring l Enormous gain, especially upstream at shorter lengths where FEXT is large

38 DSM 37 Full Vectoring (sum up/down) 100 Mbps range – 500 meters, single line u 1 km on 2 lines u >1.5 km on 4 lines l Even when partially vectored, 100 Mbps on 2 lines at 500 m, 1 km on 4 lines

39 DSM 38 Broadband DSL l Enables video, data, voice services and packaging Bit rate # of lines Rate/lineRange 100 Mbps 1 100 M 300-500 m 100 Mbps 2 50 M 1 – 1.2 km 100 Mbps 4 25 M 1.5-2 km

40 DSM 39 3 steps l ANSI 2001-200R5, 2002-018, 057, 059, 127, 129 l http://isl.stanford.edu/~cioffi/dsm/index.html http://isl.stanford.edu/~cioffi/dsm/index.html l Looking to collaborate with service/content providers, equipment/semi companies ADSL to D: 7 Mbps U:.3 Mbps 1 line 0-6 km 2002 … AMDSL to D: 25 Mbps U: 10 Mbps 1-2 lines CO/LT 0-4 km BDSL to D: 100 Mbps U: 100 Mbps 1-4 lines/LT 0-2 km 20032004 Dynamic Spectrum Management


Download ppt "DSM 1 Dynamic Spectrum Management John M. Cioffi Prof. EE, Stanford U. And all the work by students: Taek Chung, George Ginis, Jeannie."

Similar presentations


Ads by Google