Browse Prior Art Database

Blockwise Updating in Decision Aided Timing Recovery

IP.com Disclosure Number: IPCOM000088323D
Original Publication Date: 1977-May-01
Included in the Prior Art Database: 2005-Mar-04
Document File: 3 page(s) / 42K

Publishing Venue

IBM

Related People

Maiwald, D: AUTHOR

Abstract

In the receiver of a synchronous data transmission system, correct timing recovery is crucial to the performance of the system.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 3

Blockwise Updating in Decision Aided Timing Recovery

In the receiver of a synchronous data transmission system, correct timing recovery is crucial to the performance of the system.

Application of decision-aided timing recovery in all-digital receiver implementations is impeded by the fact that sampling is performed by an A/D converter at the very input of the receiver. Due to filtering operations required in such receivers (matched filter, Hilbert filter, equalizer) a sizeable feedback delay is introduced in the closed-loop timing-recovery circuit.

The drawing shows the realization of a simple timing-recovery algorithm in a digital receiver. F symbolizes the lumped signal processing (e.g., transversal filtering) which must be performed before a decision on the transmitted symbol can be made. Its effect on the control of the sampling phase can be approximated by a fixed delay dT, where T is the symbol interval.

A straightforward application of the stochastic-gradient algorithm to adjusting the sampler phase Theta yields the recursive relation Theta(n + d + 1) = Theta(n) + Gamma Re(e(n)/*/ c(n - 1) n = 0, 1, 2, ...... where c(n - 1) is the previous
symbol decision, e(n)/*/ = (z(n) - c(n))/*/ denotes the complex conjugate of the decision error at symbol time n, and Gamma ) 0 is the loop gain. Operation (1), and thus the updating of the sampling phase, is performed "every symbol time".

This updating method has two major consequences for the overall operation:

(i) Every signal sample y[(n + d)T + Theta] generally has a different phase shift Theta(n); thus, sampling of the input signal is nonuniform.

(ii) "Reaction" of the decision error e(n) to a certain phase adjustment can be observed only dT time units later.

Circumstance (i) makes it virtually impossible to rigorously analyze the dynamics of the recovery circuit. Since the proposition of equally-spaced sampling is violated, the transfer characteristic of the (generalized) filter F is distorted in a stochastic manner, and signal processing may differ considerably from the intended design.

Condition (ii) implies that the timing-control circuit is inherently unstable. Only very small values of the loop gain Gamma (approximately inversely proportional to the feedback delay d) can be tolerated, and their choice critically affects the system dynamics. In a digital implementation using fixed-point arithmetic, the accuracy problems are aggravated becau...