Multiplex communications – Communication techniques for information carried in plural... – Combining or distributing information via frequency channels
Reexamination Certificate
1998-10-05
2002-11-26
Olms, Douglas (Department: 2661)
Multiplex communications
Communication techniques for information carried in plural...
Combining or distributing information via frequency channels
C370S480000
Reexamination Certificate
active
06487221
ABSTRACT:
BACKGROUND OF THE INVENTION
The field of the invention is that of data transmission, in particular of the frequency division multiple access (FDMA) type, consisting in transmitting a frequency multiplex of channels.
In general, each channel of the multiplex is associated with a carrier frequency and with a data rate. In the present case, the carrier frequency and the data rate of each channel can be completely arbitrary.
More precisely, the invention relates to a digital method and to digital apparatus for filtering, decimating, and transposing into baseband a particular channel of a frequency multiplex. In other words, the present invention relates to a method and to apparatus serving to extract a particular channel referred to as the “channel of interest” from a frequency multiplex of channels, and to transpose it into baseband, and to do so while reducing the oversampling factor of said channel of interest.
In conventional manner, the operations of filtering, decimation, and transposition are implemented in a demodulator capable of receiving a frequency multiplex, and they are performed prior to other operations, for example operations such as filtering, reception, interpolation, synchronization, decoding, etc.
Also in conventional manner, it is assumed below in the present description that the frequency multiplex has already been transposed to an intermediate frequency or indeed is already in baseband. It is recalled that when a multiplex is in baseband that does not necessarily mean in any way that the channel of interest within said multiplex is itself in baseband. Consequently, it is clear that the invention (which seeks to bring the channel of interest into baseband) is applicable to both of the above cases.
The present invention relates solely to digital type processing (filtering, decimating, and transposing). In order to take maximum advantage associated with digital technology (reproducibility, precision, flexibility of design rules, . . . ), it is desirable for certain tasks that have traditionally been performed in the analog domain to be performed in completely digital manner.
This means that the frequency multiplex as a whole has already been digitized (sampled and subjected to analog-to-digital conversion). In other words, the entire frequency multiplex has been sampled at a rate that is relatively high because its band is still complete (for example the sampling rate may lie in the range 40 MHz to 60 MHz).
Since the channel of interest (the channel that is to be extracted and transposed into baseband) is of narrow width compared with the sampling frequency, it is desirable to reduce the oversampling on said channel of interest to a moderate value (4 to 8 samples per symbol, for example). Typically, the data rates processed lie in the range 10 kbit/s to 500 kbit/s, for example.
This reduction of oversampling is performed by an operation known as “decimation” which consists in selecting one out of every n samples. The factor n is referred to as the “decimation” factor.
In conventional manner, the operation of decimation is preceded by a filtering operation. Decimation produces aliasing. Efficient filtering purges the frequency band in those locations where information-carrying spectrum components are going to be aliased.
It is also desired to transpose the channel of interest into baseband so as to make subsequent processing possible (filtering, reception, interpolation, synchronization, decoding, etc.). The channel of interest is located at an arbitrary frequency within the multiplex and indeed there is no guarantee that the multiplex is itself centered on zero frequency.
A digital demodulator can solve the problem of frequency divided channels by using a decimating filter that has a tree structure. Each stage of the structure has two branches, each branch performing filtering on a sub-band equal to half the total band it receives, followed by decimation by two. The tree structure makes it possible to extract simultaneously as many distinct channels as it possesses outputs, i.e. 2
E
distinct channels where E is the number of stages. It is also important to observe that the tree structure does not require any specific transposition means, since the filtering and decimation performed at each stage bring the carriers of the various channels progressively towards zero frequency.
Unfortunately, such a tree structure suffers from the major drawback of being unsuitable for use with an arbitrary multiplex. A tree structure requires the channel carriers to be distributed in a very specific manner within the multiplex. More precisely, each channel carrier must occupy an initial position within the multiplex such that at each stage it is to be found exactly within one of the two filtering sub-bands. It will be understood that if at the input to a given stage (i.e. at the output from the decimator of the preceding stage) a carrier lies at the boundary between the sub-bands associated with the two filters of the stage (overlap of transition zones), then it runs the risk of being conserved by neither of the two filters.
Furthermore, the tree structure is designed to extract all of the channels of the multiplex and such a structure is therefore excessive when only one channel is to be extracted from a plurality of channels.
In the state of the art, solutions are also known relating solely to the pair of operations comprising filtering and decimation. Those prior solutions recommend avoiding only one step of filtering for high rates of decimation. Such an approach would require a large fraction of the frequency band to be filtered so as to leave room for the large amount of aliasing produced by high rates of decimation. Unfortunately, such a constraint means that the transition zone of the filter performing the rejection operation is narrow, which can only be achieved by a digital filter that is defined by a large number of coefficients, with each of the coefficients also needing to be quantified with precision, i.e. it can only be achieved at the cost of a large number of elementary operations on bits.
Known filtering and decimation solutions tend rather to subdivide the filtering function into two or three stages. This provides advantages in terms of overall complexity, since the number of operations performed in each stage is significantly reduced. However, in those known solutions, the configuration of the various successive stages is specific to the carrier and to the data rate of the channel that is to be extracted. In other words, at each stage, filtering is adapted to the characteristics (carrier and data rate) of a particular channel. Consequently, as a general rule, two successive stages are not identical. In addition, a given filtering and decimation apparatus is not simple to use for two distinct channels of interest. Under such circumstances, all of the filtering and decimation parameters in each stage need to be updated individually.
In other words, those known solutions for filtering and decimation suffer from the major drawback of requiring large capacity storage for storing several sets of filtering and decimation coefficients, and also of requiring a mechanism for selecting between pertinent sets of coefficients. In addition, since the stages are completely different from one another, there is no possibility of implementing them by sharing a common computation resource.
Elsewhere, for the problem of transposition, a commonly adopted solution consists in transposing the entire multiplex once so as to center the channel of interest on zero frequency. In this way, the above-mentioned known solutions for filtering and decimation in which filtering is centered likewise on zero frequency, can be used to extract the channel of interest and to decimate it.
Such transposition is not very advantageous. It needs to be performed immediately after the analog-to-digital converter, and therefore at the high sampling frequency that is imposed thereby. Unfortunately it would be more advantageous to minimize the sampling frequency before proceeding with transposition into b
Bertrand Cyril
Sehier Philippe
Alcatel
Olms Douglas
Sam Phirin
LandOfFree
Digital device and method for filtering, decimation and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Digital device and method for filtering, decimation and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Digital device and method for filtering, decimation and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2948177