Browse Prior Art Database

Method for Fast Convergence Decision Feedback Equalizer

IP.com Disclosure Number: IPCOM000012794D
Publication Date: 2003-May-28
Document File: 2 page(s) / 95K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method for fast convergence of the Decision Feedback Equalizer (DFE) that calculates errors by directly using the output of feed forward (FF) without any synthesis. Benefits include real-time fast convergence.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 55% of the total text.

Method for Fast Convergence Decision Feedback Equalizer

Disclosed is a method for fast convergence of the Decision Feedback Equalizer (DFE) that calculates errors by directly using the output of feed forward (FF) without any synthesis. Benefits include real-time fast convergence.

Background

Currently, in all wireless or wire line communication systems, channels add noise and distort the signal. So to compensate for the effect of channel distortion, communication receivers use equalizers. In time varying channels, receivers use adaptive equalizers based on Least Mean Square (LMS), or Recursive Least Square (RLS) algorithms. 

Figure 1 shows the LMS in the current DFE, where e(n) is error between Reference Symbol (d(n)) and Filter output (y(n)). The filter minimizes the error e(n). Channels spread the symbols in time domain. As a result, symbols interfere with each other and this interference is called Inter Symbol Interference (ISI). The two types of ISI are:

 

  • Precursor ISI: Interference of current symbol to future symbols
  • Postcursor ISI: Interference of current symbol to past symbols

Figure 2 shows the current DFE, where FF removes Precursor ISI and FB measures Postcursor ISI. From the current symbol (output of FF) the ISI from previously detected symbols is subtracted, which is given by:

                                        z(n)=y(n)-x(n) and e(n)=d(n)-z(n)

In this output, FF is first synthesized by subtracting ISI by past symbols, and used for calculating error e(n)...