Timing and automatic frequency control of digital receiver using the cyclic properties of a non-linear operation

• US 5,282,228 A
• Filed: 12/09/1991
• Issued: 01/25/1994
• Est. Priority Date: 12/09/1991
• Status: Expired due to Term

• Current Assignee: Aviat Networks Canada ULC (Aviat Networks Incorporated)
• Original Assignee: Novatel Communications Ltd (Hexagon AB)
• Inventors: Kaube, Michael, Anvari, Kiomars, Scott, Kenneth E.
• Primary Examiner(s): Kuntz, Curtis
• Assistant Examiner(s): Bocure, Tesfaldet
• Application Number:
  US07/804,424
• Time in Patent Office: 778 Days
• Field of Search: 375/97, 375/94, 375/80, 375/81, 375/93, 329/304, 329/345, 329/346, 329/306, 329/307, 331/1 R, 455/164.1
• US Class Current: 375/344
• CPC Class Codes: H04L 2027/003 at baseband only
• CPC Class Codes: H04L 2027/0055 single phase
• CPC Class Codes: H04L 27/2332 using a non-coherent carrier
• CPC Class Codes: H04L 7/0054 Detection of the synchronisation error by features other than the received signal transition by means of signal transition H04L7/033

Abstract

A technique for correcting the sampling time and carrier frequency error in a receiver for digitally modulated signals. Discrete-time, complex-valued samples of the incoming signal are fed to a pair of non-linear operators such as correlators. The first correlator provides a signal having a first phase component related to the symbol timing error and a second phase component related to the carrier frequency error. The second correlator provides a signal having a first phase component related to the symbol timing error and a second phase component related to the negative of the carrier frequency error. The phase components are then separated and detected to extract an estimate of the symbol timing error and the carrier frequency error. In the preferred embodiment, the complex-valued samples are frequency-shifted, before being fed to the correlators, so that the phase components of interest appear at zero frequency. The invention satisfactorily demodulates transmitted digital signals in applications, such as cellular time-division multiple access (TDMA), where they are susceptible to multipath fading.

Claims

1. A method for correcting the symbol sampling time and carrier frequency error in a received modulated digital signal, comprising the steps of:

demodulating the modulated digital signal, by mixing it with a local reference oscillator to provide a baseband signal;

sampling the baseband signal to provide discrete-time, in-phase (I) and quadrature (Q) components thereof;

performing a first non-linear operation on the I and Q components, to produce a first composite signal having a first phase component related to a symbol timing error and a second phase component related to a carrier frequency error;

performing a second non-linear operation on the I and Q components, to produce a second composite signal having a first phase component related to the symbol timing error and a second phase component related to the negative of the carrier frequency error;

detecting the symbol timing error by comparing the first and second composite signals;

detecting the carrier frequency error by comparing the first and second composite signals;

adjusting the time of subsequent sampling steps according to the detected symbol timing error; and

adjusting the frequency of the I and Q components according to the detected carrier frequency error.

2. A method as in claim 1 wherein the first non-linear operation is a cyclic correlation.

3. A method as in claim 1 wherein the first non-linear operation is multiplication of a present I and Q component and a complex conjugate of an I and Q component time-delayed by a symbol period.

4. A method as in claim 1 wherein the second non-linear operation is a cyclic correlation.

5. A method as in claim 1 wherein the first non-linear operation is multiplication of a present I and Q component and a complex conjugate of an I and Q component time-advanced by a symbol period.

6. A method as in claim 1 additionally comprising the step of, after the sampling step:

frequency shifting the I and Q components so that they appear at a frequency of zero.

7. A method as in claim 1 wherein the step of adjusting the frequency of the I and Q components is performed after the step of sampling the baseband signal.

8. A method as in claim 1 wherein the step of adjusting the frequency of the I and Q components is performed by using a discrete-time multipler to multiply the I and Q components by a sinusoid having a frequency equal to the detected carrier frequency error.

9. A method as in claim 1 wherein the local oscillator is an adjustable oscillator and the step of adjusting the frequency of the I and Q components is performed by adjusting the local oscillator frequency by an amount indicated by the detected carrier frequency error.

10. A method for correcting the symbol sampling time and carrier frequency error in a received modulated digital signal, comprising the steps of:

taking samples of the modulated digital signal in synchronism with a local clock reference signal, to provide discrete time, in-phase (I) and quadrature (Q) components of the modulated digital signal;

performing a first non-linear operation on the I and Q components, to produce a first composite signal having a first phase component related to a symbol timing error and a second phase component related to a carrier frequency error;

performing a second non-linear operation on the I and Q components, to produce a second composite signal having a first phase component related to the symbol timing error and a second phase component related to the negative of the carrier frequency error;

detecting the symbol timing error by comparing the first and second composite signals;

detecting the carrier frequency error by comparing the first and second composite signals; and

adjusting subsequent I and Q components by the steps of;

adjusting the I and Q components by adjusting the timing of the local reference clock signal by an amount indicated by the symbol timing error; and

adjusting the frequency of the I and Q components by multiplying the I and Q components by a sinusoid having a frequency equal to the detected carrier frequency error.

11. A method for correcting the symbol sampling time and carrier frequency error in a received modulated digital signal, comprising the steps of:

generating a digital clock reference signal;

taking samples of the modulated digital signal in synchronism with the local clock reference signal, to provide discrete time, in-phase (I) and quadrature (Q) components of the modulated digital signal;

forming complex-valued signal components (S) by combining the I and Q components, with the I component being the real part of the complex-valued sample S, and the Q component being the imaginary part of the complex-valued sample S;

introducing a phase shift in the complex-valued sample S to compensate for a frequency error, f_o, in the digital clock reference signal, the phase shift being introduced by discrete-time multiplying the complex-valued sample S by a digital complex sinusoid having a complex frequency e^-j(2π

fe tn), where t_n indicates discrete time;

introducing a timing correction in the digital clock reference signal, by adjusting the time at which the samples are taken by an amount equal to a timing error, τ

;

estimating the frequency error, f_o, and the timing error, τ

, in the digital clock signal, by performing the steps of;

multiplying each of the complex-valued samples, S, by a time-shifted and conjugated of itself, to provide a first correction signal, C₊, having a first phase component related to the frequency error f_o divided by a nominal interval T between successive samples, and also having a second phase component related to the negative ratio of the timing error τ and

the sample interval T;

multiplying each of the complex-valued samples, S, by a time-shifted and conjugated of itself, to provide a second correction signal, C_-, having a first phase component related to the negative of the frequency error f_o divided by the nominal interval T, and also having a second phase component related to the negative ratio of the timing error τ and

the interval T between successive samples;

detecting the timing error, τ

, by multiplying the first correction signal C₊ by the second correction signal C_- and determining the arc tangent thereof; and

detecting the frequency error, f_o, by multiplying the conjugate of the first correction signal C₊ by the second correction signal C_- and determining the arc tangent thereof.

12. A method as in claim 11 wherein the step of detecting the frequency error f_o additionally comprises the steps of;

determining a conjugate-square-root of the product of the first correction signal C₊ and the second correction signal C_- ;

multiplying the conjugate-square-root by the first correction signal C₊, to obtain a first multiplied signal;

multiplying the conjugate-square-root by the second correction signal C_-, to obtain a second multiplied signal;

summing the first and second multiplied signals; and

determining the arc tangent of the first and second multiplied signals.