Pulse or digital communications – Systems using alternating or pulsating current – Plural channels for transmission of a single pulse train
Reexamination Certificate
1999-11-02
2004-03-02
Chin, Stephen (Department: 2634)
Pulse or digital communications
Systems using alternating or pulsating current
Plural channels for transmission of a single pulse train
C375S222000, C370S210000
Reexamination Certificate
active
06700936
ABSTRACT:
BACKGROUND OF THE INVENTION
This invention relates to receivers for many-carrier signals, for example OFDM (orthogonal frequency-division multiplex) signals, such as used in digital audio broadcasting (DAB) or digital sound broadcasting, and also in digital television.
The digital audio broadcasting system used in the United Kingdom, now known as mode 1 DAB, requires seven DAB ensembles (channels) each occupying 1.536 MHz within the overall frequency range 217.5 MHz to 230 MHz, a total bandwidth of only 12.5 MHz. The spacing between ensemble centres is 1712 KHz, of which 1536 kHz is taken up by the signal, so that the spacing between the top of one ensemble and the bottom of the next is only 176 kHz. The signal bandwidth of 1536 kHz arises from the use of 1536 separate carriers spaced at spacings of 1 kHz. The number 1536 is chosen as three-quarters of 2048, which equals 2 to the power of eleven, that being the minimum power of 2 which is greater than 1536.
A DAB receiver must be able to receive the desired DAB ensemble to which it is tuned in the presence of a number of interfering DAB signals occupying the adjacent spectrum. Of the adjacent signals, it is the nearest neighbour which is most difficult to reject because of the high ratio of signal bandwidth to edge spacing; the edge spacing is less than one-eighth of the signal bandwidth.
A typical DAB receiver of this type is shown in schematic block diagram form in
FIG. 1
of the drawings. The receiver
100
has an antenna
112
feeding a radio frequency (RF) stage shown as an RF amplifier
114
. The output of the RF amplifier is applied to a mixer
116
which receives a first local oscillator signal LO
1
at a terminal
118
. The mixer reduces the frequency of the received signal, typically at around 225 MHz, to an intermediate frequency of typically about 36 MHz. The output of the mixer
116
is applied to an IF bandpass filter
120
which passes the desired intermediate frequency in the region of 36 MHz. The output of the IF filter
120
is then applied to an I/Q demodulator circuit
122
, which receives a second local oscillator signal LO
2
at a terminal
124
. The I/Q demodulator circuit reduces the signals to baseband frequency and also separates the in-phase (I) and quadrature phase (Q) components of the signal. The output
126
of the I/Q demodulator thus in fact comprises two signals, as indicated on
FIG. 1
, and the subsequent circuitry is duplicated for the two signals, as is well known.
The output
126
of the I/Q demodulator
122
is applied to an anti-alias bandpass filter
128
, the characteristics of which are described in more detail below, and from the anti-alias filter
128
are applied to a sampler or analog-to-digital converter
130
. The sampler
130
operates at 2.048 Ms/s (mega-samples per second), which is of course the same as the sample rate of the digital signals which were used to form the transmitted signal at the transmitter. In the sampler
130
the signals are now converted from analog form to digital form, and are then applied to a fast Fourier transform (FFT) circuit
132
. The FFT generates a signal in the form of a sequence or series of symbol periods. The FFT has 2048 points which corresponds to the theoretical number of carriers with a sampler operating at 2.048 Ms/s, a carrier spacing of 1 kHz, and an active symbol period of 1 ms. In fact as noted above, only 1536 carriers are used, the remainder having a theoretical amplitude of zero.
To achieve this places considerable demands on the filter
128
. This filter should have a pass-band extending to ±768 kHz (half of 1536 kHz) but a cut-off frequency of ±1024 kHz (half of 2048 kHz). This is a sharp cut-off and is difficult to achieve.
The output
134
of the FFT is a time-based signal which is then processed using conventional receiver circuitry (not shown).
The circuit of
FIG. 1
will be known to those skilled in the art, and further description thereof is not necessary.
Likewise, a corresponding transmitter will be known to those skilled in the art, and includes a 2048-point inverse FFT operating in the digital domain corresponding to the FFT
132
at the receiver. The inverse FFT receives a conventional time-based signal and converts it into a many-carrier signal for transmission.
FIG. 2
is a spectrum diagram showing three adjacent ensembles in the frequency spectrum. The numerical values are those appropriate to the DAB system described above, and are referred to the centre frequency of the central ensemble E which is taken to be zero. One ensemble E+1 is shown above this with positive values and another ensemble E−1 is shown below it with negative values. The values are in kilohertz, but as the individual carriers are spaced by 1 kHz, they can equally be treated as a count of carriers. The amplitudes of the signals shown are purely arbitrary; they are shown for convenience of illustration with a slight peak at the centre of each ensemble but in theory the amplitudes should be flat.
It will be seen that each ensemble extends over 1536 carriers, and that the spacing between corresponding points on the ensembles is 1712 carriers.
FIG. 2
also shows, for the central ensemble, the positions where the sampling frequency and the inverse appear. These fall at ±2048 carriers. The values of half the sampling frequency, fs/2, which fall at ±1024 carriers, are also shown. The value of half the sampling frequency is, as is well known, the Nyquist limit. Frequencies which appear above half the sampling frequency can not be correctly represented by the sampling process.
These frequencies above half the sampling frequency, when subjected to sampling, give rise to spurious components in the sampled signal known as aliased components. The aliased components are commonly thought of as through the signals in the range fs/2 to fs were “reflected” about the frequency fs/2. Thus, a frequency which is a g Hz below the sampling frequency fs, that is a signal of frequency (fs−g) Hz, gives rise to an alias component of frequency of g Hz. This is correct for basebands signals, but for signals above baseband, correctly what happens is that the signals above fs/2 are translated downwards by a frequency shift equal to the sampling frequency fs. Such shift occurs in fact for all integral multiples of the sampling frequency, but only the first and most powerful need be considered in practice.
This is illustrated in
FIG. 3
, which shows just the central ensemble E of FIG.
2
and the ensemble E+1 above it. It also shows the aliased components which arise by down-shifting the upper ensemble E+1 by the sampling frequency. Those frequencies which arise in the range from above the Nyquist frequency, or half the sampling frequency, namely 1024 kHz, up to the top of the upper ensemble E+1, namely 2480 kHz, are moved downwards by 2048 kHz. The aliased components E−1 now span the frequency range +1024 to +432. Of these the frequencies in the range of −768 to +432 fall within the band of the wanted ensemble E. These can not be rejected by simple frequency-selective filtering. The shifted frequencies are marked in the figure by cross-hatching.
The FFT circuit
132
expects only 1536 carriers out of a possible 2048, and thus inherently rejects energy in the frequency range 768 kHz to 1280 kHz. This upper limit equals the sampling frequency 2048 kHz minus the expected upper carrier frequency limit of 768 kHz. Within the wanted band, this includes the rejection regions marked R on FIG.
3
.
We have thus appreciated that to cut out the interference components requires strong IF and anti-alias filtering in the filters
120
and
128
of
FIG. 1
, in order to reject the adjacent channel energy from ensemble E+1 before it reaches the analog-to-digital converter or sampler
130
. To produce a sufficiently sharp cut-off may require for example a surface acoustic wave (SAW) filter for use as the filter
128
. Such filters are expensive and lossy and may result in the partial loss of a num
British Broadcasting Corporation
Chin Stephen
Williams Lawrence
Wolf Greenfield & Sacks P.C.
LandOfFree
Many-carrier transmission system and a receiver therefor does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Many-carrier transmission system and a receiver therefor, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Many-carrier transmission system and a receiver therefor will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3265927