IEEE S.F. Bay Area Council e-GRID's
BayAreaTech

Technology & Networking in Silicon Valley & the SF Bay Area: Upcoming Meetings, Courses and Conferences

November 17, 2013

Mftg: Sign-based Zero-Forcing Adaptive Equalizer Control for High-Speed I/O

by @ 8:20 pm. Filed under ALL, Electronics Design, Semiconductors
 

MONDAY November 18, 2013
SCV Circuits and Systems Chapter
Speaker: Dr. Yasuo Hidaka, Fujitsu Laboratories of America
Time: Networking/Light Dinner at 6:30 PM; Presentation at 7:00 PM
Cost: $2 donation accepted for food
Place: QualComm, Building B, 3165 Kifer Road, Santa Clara
RSVP: from website
Web: sites.ieee.org/scv-cas

Equalizers are often used for wireline transceivers to cancel Inter-Symbol Interference (ISI) caused by frequency-dependent channel loss.  Adaptation of equalizer parameters is desired, because channel characteristics are generally unknown.  Least Mean Square (LMS) and Sign-Sign-LMS (SS-LMS) are adaptation algorithms widely used in digital signal processing, because it is easy to achieve minimum mean square error (MMSE) using these algorithms.  However, LMS and SS-LMS are generally not applicable to mixed-signal circuits such as continuous-time linear equalizer (CTLE) or transmitter pre-emphasis in High-Speed I/O, because key reference information required for LMS and SS-LMS are not necessarily available for those mixed-signal circuits.  On the other hand, Sign-based Zero-Forcing (S-ZF) is an adaptation algorithm that is always applicable to mixed-signal circuits including CTLE and transmitter pre-emphasis, because it does not require any additional reference information in analog circuits other than error information.  Unlike conventional ZF, LMS, or SS-LMS, S-ZF does not assume random data sequence for measurement of correlation between error and reference information.  Instead, S-ZF uses filter patterns to accurately de-convolve error information to residual ISI information. S-ZF works robustly for non-scrambled data such as 8B10B coding which may include a monotone sequence such as continuous 1010.  Adaptation performance of S-ZF can be similar to LMS and SS-LMS, if used with proper parameters, because S-ZF can also achieve the MMSE condition for the worst channel.  For decision feedback equalizer (DFE), S-ZF converges to the same state as SS-LMS, because they are based on the same formula.  We have applied the S-ZF scheme for adaptation of equalizers such as CTLE, speculative or non-speculative DFE, and transmitter pre-emphasis in several generations of high-speed I/O from 1.25Gb/s up to 32Gb/s.  Recently, we have also extended the S-ZF scheme to be applicable to low-frequency equalizer for which measurement of long-term residual ISI such as for 50 unit intervals (UIs) is required.

20131118scv

Comments are closed.

[IEEE S.F. Bay Area Council - www.e-grid.net] [powered by WordPress .]

SF Bay Area Council

Categories

  • ALL (4,557)
  • Antennas & Propagation (4)
  • BioEngineering (663)
  • Blogroll (33)
  • Circuits (11)
  • Communications (1,853)
  • Computers/Software (1,744)
  • Consumer Electronics (97)
  • Control Systems (18)
  • Electrical/Power (1,263)
  • Electronics Design (2,768)
  • Employment (4)
  • Employment opportunities (5)
  • Engineering Mgmt (1,805)
  • Green energy (40)
  • History (3)
  • Industrial Applications (73)
  • Information Theory (10)
  • Magnetics (22)
  • Microwave (8)
  • NanoEngineering (869)
  • Optics/Displays (968)
  • Photonics (24)
  • PhotoVoltaics (3)
  • Product Safety (21)
  • Reliability (34)
  • Robotics and Automation (15)
  • Semiconductors (1,827)
  • Signal Processing (132)
  • Vehicular Technology (9)
  • Women in Engineering (7)
  • Young Professionals (2)
  •  

    Support our advertisers:

    Visit our
    GRID MARKETPLACE

    Enabling Javascript allows us to show you upcoming conferences in this column.

    For the Firefox browser, select Tools/Options/Content and select "Enable Javascript".

    If you are using Microsoft Internet Explorer you may need to click on the yellow bar above and select 'Allow Blocked Content'. You must then click 'Yes' on the following security warning.

    archives:

    October 2017
    S M T W T F S
    « Sep    
    1234567
    891011121314
    15161718192021
    22232425262728
    293031  

    View in Google Calendar

    search blog:


    SUBSCRIBE: Get the e-GRID twice a month by email - upcoming IEEE SF Bay Area meetings, conferences.

    RSS Feed Subscribe to our RSS Feed.

    PUBLICIZE your event to IEEE's membership.


    general links:

    22 queries. 0.585 seconds