Pulse or digital communications – Receivers
Reexamination Certificate
1998-12-04
2002-02-19
Bocure, Tesfaldet (Department: 2631)
Pulse or digital communications
Receivers
C375S340000
Reexamination Certificate
active
06349120
ABSTRACT:
The present invention relates to methods of spectral sampling in communication systems using Discrete Fourier Transforms. More particularly, it relates to spectral sampling where there are large propagation delays or large frequency or time uncertainties for detection. Even more particularly, the present invention relates to spectral sampling of a continuous wave burst from a mobile satellite station during synchronization.
In mobile satellite communications, mobile satellite stations, satellites and gateway stations must be synchronized in order to diminish the effects of large propagation delays from the satellite due to the large distances from each satellite station to a base station. The propagation and other delays cause large frequency and time uncertainty at the base station. Typically, synchronization is achieved during a call set up, wherein a user terminal sends a random access burst (RAB) to the base station. A continuously wave (CW) signal is part of a random access burst (RAB) and is used to resolve initial large frequency and time uncertainty at the base station. The RAB is random and the acquisition of the RAB is based on the detection of the CW burst.
A conventional way to detect the CW burst, is to perform a Discrete Fourier Transform (DFT) on the random access burst. Discrete Fourier Transforms are well studied and relatively easy to implement for the skilled artisan.
Additionally, some unique words (UW), or packet identification bit sequences, are used in the random access burst (RAB) to resolve the large time uncertainty. In this way, the satellite can use a DFT algorithm for CW signal detection in the random access burst and, after achieving frequency estimation, further achieve time estimation based upon correlation of the unique words from the random access burst, as received by the satellite, with a reference random access burst.
However, conventional use of a Discrete Fourier Transform (DFT) scheme for a continuous wave burst detection suffers a spectral sampling effect. Conventionally the Discrete Fourier Transform scheme involves performing a finite number of band-pass filter calculations, or bin calculations, with filter calculations defined according to uncertainty in the frequency. The Discrete Fourier Transform starts to search around a nominal frequency (plus or minus some error ellipse) of the communication system, which can be C-band, K-band, L-band, or any other frequency band.
In the case of the CW signal, the frequency is unknown at the receiver, and thus the Discrete Fourier Transform spectral lines (which are a function of the sampling frequency, and not the frequency of the CW signal) do not necessarily coincide with the frequency of received CW signal. Thus, the Discrete Fourier Transform output cannot truly represent a real signal spectrum of the CW signal, thus causing defective detection performance to vary with frequency of the CW signal. Thus, the detection performance will vary according to the CW signal frequency, and, unless the CW frequency exactly equals one of the Discrete Fourier Transform's bin frequencies, will be degraded according to how far the CW signal's frequency varies from bin frequency of the Discrete Fourier Transform.
Conventionally, a communication receiver or the satellite will coherently combine all the digital samples from the CW signal. This requires the communication receiver to track the signal phase, requiring a more complicated receiver in general. This conventional method achieves the highest theoretical performance in the case where the frequencies of the CW signal are always aligned with one of the DFT bin frequencies. When the frequency of the CW signal is not always aligned with zone of the Discrete Fourier Transforms bin frequencies, overall performance suffers, unless complicated and expensive circuitry is involved in achieving such performance using coherent signal combination.
As shown in a block diagram in
FIG. 2
, in a conventional CW detector using a Discrete Fourier Transform, a first stage shifts an RF frequency at
212
down to an intermediate frequency, or IF frequency at
214
. Next, after sampling the analog signals of in-phase (I) and quadrature phase (Q) components, two identical digital-low pass filters
220
,
221
smooth the I and Q components samples. Next, filtered I and Q samples are first decimated and then fed into the Discrete Fourier Transform processor. The Discrete Fourier Transform processor
222
computes the signal power of those bin frequencies covering an initial frequency uncertainty, and a maximum power is chosen from amongst the bin frequencies by a maximum power selector
224
. The chosen maximum power is compared with a pre-set threshold by a comparator
226
to determine if a continuous wave burst has arrived.
Since many frequency bins must be computed, it can become increasingly time consuming to perform the Discrete Fourier Transform computation for a random access burst with very large frequency and time uncertainty. The number of frequency bins is therefore kept to a minimum by increasing the size (i.e., filter bandwidth) of each bin. Once a preliminary detection is achieved, a finer frequency estimation and time estimation is performed using narrower filtering bins.
Prior methods to combat the effect of spectral sampling have included: 1) zero-padding plus interpolation, and 2) windowing the received CW signal samples.
Each of the two known prior methods of zero-padding and windowing are computationally intensive and while improving the overall performance of the conventional Discrete Fourier Transform algorithm increases the performance time and complexity of the receiver within the satellite.
The known method of zero-padding for CW signal detection requires more frequency bins with the same frequency uncertainty. Thus, the method of zero-padding is not acceptable for modern mobile satellite communication systems.
The prior method of windowing weights the samples for a non-linear window and consequently reduces side lobe effects of the signal. This requires a large amount of computation because a window weighting must be redone upon each new burst. Thus, the method of windowing is not acceptable for modern mobile satellite communication systems.
There is therefore a need in the field of mobile satellite communications for an improved method of continuous wave detection (cw detection) using a Discrete Fourier Transform processor which does not degrade performance or increase complexity of computations.
The present invention advantageously addresses the above and other needs.
SUMMARY OF THE INVENTION
The present invention advantageously addresses the needs above as well as other needs by providing a method for improving spectral sampling in continuous wave burst detection by using a non-coherent combination of sub-burst Discrete Fourier Transforms in a processor.
The advantage of using sub-bursts to more closely align a signal frequency to a bin frequency out weighs the theoretical degradation of using non-coherent combining of the sub-bursts instead of coherent combining. The optimum sub-bursts are determined with this invention.
The present approach involves, in one embodiment, the steps of: dividing the digital samples of a continuous wave burst into smaller sub-bursts, each having a smaller number of samples M than the total number of samples N in the continuous wave burst; performing a Discrete Fourier Transform on each sub-burst to produce associated sub-burst Transform data; and non-coherently combining the associated sub-burst Transform data for each sub-burst to produce a non-coherent digital signal.
In another embodiment, the invention can be characterized as a method for improving spectral sampling in a communication system using a Discrete Fourier Transform processor to detect a continuous wave signal of large uncertainty, further comprising steps of: receiving a continuous wave burst comprising a signal; digitizing the signal into N samples; selecting an optimal number L of sub-bursts based upon system performance including data ra
Antia Yezdi
Hammons, Jr. A. Roger
Shi Zhen-Liang
Bocure Tesfaldet
Hughes Electronics Corporation
Sales Michael W.
Whelan John T.
LandOfFree
Method for improving spectral sampling using sub-burst... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for improving spectral sampling using sub-burst..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for improving spectral sampling using sub-burst... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2934016