SONY CORPORATION P36S00EP
New frame and signalling pattern structure for multi-carrier systems
The present invention is directed to a new frame and signalling pattern structure for multi-carrier systems.
The present invention is hereby mainly directed (but not limited) to broadcast systems, 10 such as for example cable based or terrestrial digital broadcast systems, in which
content data, signalling data, pilot signals and so forth are mapped on to a plurality of frequency carriers, which are then transmitted in a given overall or complete transmission bandwidth. The receiver typically tunes to a partial channel (part of the overall transmission bandwidth) out of the complete transmission bandwidth 15 (sometimes called segmented reception) in order to receive only the content data which is necessary or wanted by the respective receiver. For example, in the ISDB-T standard, the overall channel bandwidth is hereby divided into 13 fixed segments of an
equal length (equal number of frequency carriers).
B
20 The object of the present invention is therefore to provide a transmission apparatus and method as well as a signal structure for a multi-carrier system, which allows a receiver to be flexibly tuned to any required part of the overall transmission bandwidth.
The above object is achieved by a transmitting apparatus according to claim 1. The
25 transmitting apparatus according to the present invention is adapted to transmit signals
in a multi-carrier system on'the basis of a frame structure, each frame comprising at
least two signalling patterns adjacent to each other in the frequency direction and at
i| *■ -
least two data patterns, said1 transmitting apparatus comprising
signalling mapping means adapted to map signalling data on frequency carriers of
30 each of said at least two signalling patterns in a frame, each signalling pattern having
the same length,
data mapping means adapted to map data on frequency carriers of said at least two data
patterns in a frame,
transforming means adapted to transform said signalling patterns and said data patterns
l!

2
from the frequency domain into the time domain in order to generate a time domain
transmission signal, and
transmitting means adapted to transmit said time domain transmission signal.
5 The above object is further achieved by a transmitting method according to claim 9. The transmitting method according to the present invention is adapted to transmit signals in a multi-earner system on the basis of a frame structure each frame
comprising at least two signalling patterns adjacent to each other in the frequency
I) direction and at least two data patterns, whereby the method comprises the steps of
i| 10 mapping signalling data on frequency carriers of each of said at least two signalling
patterns in a frame, each signalling pattern having the same length,
i| mapping data on frequency carriers of said at least two data patterns in a frame,
transforming said signalling patterns and said data patterns from the frequency domain
into the time domain in order to generate a time domain transmission signal, and
il 15 transmitting said time domain transmission signal.
The above object is further achieved by a frame partem according to claim 10. The frame pattern of the present invention is adapted for a multi-carrier system and comprises at least two signalling patterns adjacent to each other in the frequency 20 direction and at least two data patterns, wherein signalling data are mapped on
frequency carriers of each of said at least two signalling patterns in the frame, each signalling pattern having the same length, and wherein data are mapped on frequency carriers of said at least two data patterns in the frame.
25 The object of the present invention is further to provide a receiving apparatus and method, as well as a system and a method for transmitting and receiving signals in a multi-carrier system, which allow a flexible tuning to any required part of the transmission bandwidth.
30 The above object is achieved by a receiving apparatus according to claim 11. The .
receiving apparatus according to the present invention is adapted to receive signals in a multi-carrier system on the basis of a frame structure in a transmission bandwidth, each frame comprising at least two signalling patterns adjacent to each other in the frequency direction each with signalling data mapped on frequency carriers and at least

two data patterns with data mapped on frequency carriers, each of said at least two
signalling patterns having the same length, said receiving apparatus comprising
receiving means adapted to be tuned to and to receive a selected part of said
transmission bandwidth, said selected part of said transmission bandwidth having at
\ 5 least the length of one of said signalling patterns and covering at least one data pattern
^ to be received, and evaluation means adapted to evaluate the signalling data comprised
in a received signalling pattern in order to enable the receipt of said at least two data
patterns.

10 The above object is further achieved by a receiving method according to claim 22. The receiving method of the present invention is adapted to receive signals transmitted in a multi-carrier system on the basis of a frame structure in a transmission bandwidth, each frame comprising at least two signalling patterns adjacent to each other in the frequency direction each with signalling data mapped on frequency carriers and at least
15 two data patterns with data mapped on frequency carriers, each of said at least two signalling patterns having the same length, comprising the steps of receiving a selected part of said transmission bandwidth, said selected part of said transmission bandwidth having at least the length of one of said signalling patterns and
covering at least one data pattern to be received, and
t
20 evaluating the signalling data comprised in a received signalling pattern in order to
east two data patterns.
enable the receipt of said at.
The above object is further achieved by a system for transmitting and receiving signals
according to claim 23, comprising a transmitting apparatus for transmitting signals in a 25 multi carrier system on the basis of a frame structure, each frame comprising at least
two signalling patterns adjacent to each other in the frequency direction and at least
two data patterns, said transmitting apparatus comprising
signalling mapping means adapted to map signalling data on frequency carriers of
each of said at least two signalling patterns in a frame, each signalling partem having 30 . the same length,
data mapping means adapted to map data on frequency carriers of said at least two data
patterns in a frame,
transforming means adapted to transform said signalling patterns and said data patterns
from the frequency domain [into the time domain in order to generate a time domain

4
transmission signal, and
transmitting means adapted to transmit said time domain transmission signal, said system further comprising a receiving apparatus according to the present invention adapted to receive said time domain transmission signal from said transmitting 5 apparatus.
The above object is further achieved by a method for transmitting and receiving signals according to claim 24, comprising a transmitting method for transmitting signals in a multi carrier system on the basis of a frame structure, each frame comprising at least 0 two signalling patterns adjacent to each other in the frequency direction and at least two data patterns, said transmitting method comprising the steps of mapping signalling data on frequency carriers of each of said at least two signalling patterns in a frame, each signalling pattern having the same length, mapping data on frequency carriers of said at least two data patterns in a frame,
15 transforming said signalling patterns and said data patterns from the frequency domain into the time domain in order to generate a time domain transmission signal, and transmitting said time domain transmission signal,
said method further comprising a receiving method according to the present invention adapted to receive said time domain transmission signal.
20
The present invention therefore suggests a multi-carrier system which uses a frame structure or frame pattern in the frequency domain as well as in the time domain. In the frequency domain, each frame comprises at least two signalling patterns, which respectively carry signalling data or information on frequency carriers and respectively
25 have the same length (or bandwidth). After a conversion into the time domain, in the resulting time domain signal, each frame then comprises a respective signalling symbol as well as data symbols. Each frame pattern covers the entire or overall transmission bandwidth in the frequency direction, so that the overall transmission bandwidth is therefore equally divided by the signalling patterns having the same
30 respective length. The data patterns of each frame then follow the signalling patterns in time. The receiving apparatus can be freely and flexibly tuned to any wanted part of the transmission bandwidth, provided that the part of the transmission bandwidth to which the receiving apparatus can be tuned has at least the length of one of the signalling patterns. Hereby, the receiving apparatus is always able to receive the

5
signalling data of an entire signalling pattern, so that on the basis and using the signalling data comprising the physical layer information necessary for the receipt of the succeeding data patterns, the data patterns can be received in the receiving apparatus. 5
Advantageously each frame comprises at least two additional signalling patterns succeeding said at least two signalling patterns in the time dimension, each of said additional signalling patterns having the respective same length as the corresponding one of said at least two preceding signalling patterns. Hereby, even if the length (or
10 bandwidth) of each signalling pattern is not long enough to comprise all necessary
signalling data, it is possible to transmit the necessary remaining signalling data in the additional signalling data. Even if the receiving apparatus has a rather small (effective) receiving bandwidth, it is thus possible to transmit and receive all necessary signalling data.
15 Further advantageously, each frame comprises at least two training patterns, wherein pilot signals are mapped on frequency carriers of each training pattern in a frame, and wherein the signalling patterns are aligned to the training patterns in the frequency direction. Hereby, by means of the training patterns which precede the signalling patterns in the time direction, the receiving apparatus which first receives the training
20 partem is able to perform time, synchronisation, frequency offset calculation and/or channel estimation, where after the signalling data in the received signalling patterns can be received and used to receive the succeeding data patterns independent of the tuning position of the receiving apparatus. For example, every training pattern has the same length, and the length of each signalling pattern is the same as the length of each
25 of said training patterns. Alternatively, every training pattern has the same length, and the length of each signalling pattern is smaller than the length of each of said training patterns. Hereby, the length of each signalling partem may be half the length of each of said training patterns. Implementations may be possible in which the signalling patterns are not aligned with the training patterns.
30
Advantageously, each signalling pattern comprises at least one guard band. Hereby, it is ensured that the receiving apparatus can receive all signalling data in a signalling pattern even in the case that the effective receiving bandwidth is smaller that the tuning bandwidth, e.g. due to filter characteristics or the like. Hereby, each signalling pattern

6 may comprise a guard band at its beginning and a guard band at its end.
Advantageously, each signalling pattern of each frame comprises the location of the signalling pattern within the frame, which is extracted and evaluated on the receiving
5 side. In this case, further advantageously, each signalling pattern in each frame could comprise the identical signalling data except the location of the respective signalling pattern in the frame, which is different in at least some of the signalling patterns in a frame. Hereby, the receiving apparatus is able to determine its position within the overall transmission bandwidth (within each frame) for example during the
) initialisation period, in which the receiving apparatus is tuned to an arbitrary position within a frame, and then to tune to the bandwidth enabling the receipt of the wanted data on the basis of the signalling data in the received signalling pattern.
Further advantageously, the signalling patterns of each frame comprise signalling data
5 indicating the number of data patterns comprised in the frame. Further advantageously, the structure of the signalling data in the signalling patterns supports a limited maximum number of data patterns in the frequency direction of each frame, further, the signalling patterns of each frame may comprise individual signalling data for each data pattern comprised in the frame.
3
Further advantageously, the signalling data of the signalling patterns comprise an error detection and/or correction coding. Hereby, even if a receiving apparatus cannot receive an entire signalling pattern, the receiving apparatus may still be able to obtain the entire signalling information contained in the signalling pattern.
5
Although the receiver can be flexibly tuned to any wanted part of the transmission bandwidth, it is always possible to receive the signalling data of an entire signalling partem due to the new frame structure suggested by the present invention. Advantageously, the receiving apparatus comprises a reconstructing means adapted to
0 reconstruct the original signalling pattern from said received selected part of said transmission bandwidth. Hereby, said reconstructing means may be adapted to rearrange received signalling signals into the original signalling partem in case that the selected part of said transmission bandwidth to which the receiving means is tuned does not match with the signalling pattern structure. Thus, even if the selected part of

the transmission bandwidth to which the receiver is tuned does not completely and correctly match with one of the signalling patterns (in the frequency direction), the receiver will in such cases receive the last part of a (frequency wise) preceding signalling pattern and the first part of a (frequency wise) succeeding signalling pattern. For example, in case that the receiving apparatus knows its (frequency dimension) offset from the signalling pattern structure in each frame, said reconstructing means may be adapted to rearrange received signalling signals into the original signalling pattern. Alternatively, each frame comprises at least two additional signalling patterns succeeding said at least two signalling patterns in the time dimension, each of said additional signalling patterns having the respective same length as the corresponding one of said at least two preceding signalling patterns, wherein said reconstructing means is adapted to rearrange received two or more signalling patterns succeeding each other in the time dimension into the original signalling pattern. Thus, the preceding signalling pattern and the succeeding signalling pattern can together comprise the necessary signalling data even if the length of the signalling patterns in the frequency dimension is short that in the case where all necessary signalling data are comprised in a single signalling partem.
Alternatively or additionally, the signalling data of the signalling patterns comprise an error detection and/or correction coding, wherein said reconstructing means is adapted to perform an error detection and/or correction decoding on said received signalling signals in order to reconstruct the original signalling pattern. Hereby, the transmitted signalling patterns may comprise additional error coding, redundancies or the like enabling the receiver to reconstruct the original signalling pattern even if only a part of the signalling pattern can be received. Advantageously, each signalling pattern of each frame comprises the location of the signalling pattern within the frame, which is extracted and evaluated on the receiving side. In this case, further advantageously, each signalling pattern in each frame could comprise the identical signalling data except the location of the respective signalling pattern in the frame, which is different in at least some of the signalling patterns in a frame. Hereby, the receiving apparatus is able to determine its position within the overall transmission bandwidth (within each frame) for example during the initialisation period, in which the receiving apparatus is tuned to an arbitrary position

8
within a frame, and then to tune to the bandwidth enabling the receipt of the wanted data on the basis of the signalling data in the received signalling pattern. Advantageously, the signalling patterns of each frame comprise signalling data with the number of data patterns comprised in the frame, wherein said evaluation means is 5 adapted to extract said signalling data with the number of data patterns from a received signalling pattern. Further advantageously, the signalling patterns of each frame comprise individual signalling data with each data pattern comprised in the frame, wherein said evaluation means is adapted to extract said individual signalling data with each data pattern from a received signalling pattern.
10
Advantageously, the receiver is adapted to be tuned to and to receive a selected part of said transmission bandwidth so that an optimized receipt of a signalling partem in the selected part of the transmission bandwidth is enabled. Particularly if the frequency dimension structure of the data patterns and the signalling patterns in a frame do not
15 match, and if the selective part of the transmission bandwidth to be received in the receiver is larger (in frequency dimension) than the data pattern(s) to be received, it may be possible to optimize the tuning so that the best possible receipt of a signalling pattern is achieved, for example by adjusting the tuning so that the maximum part of one entire signalling pattern is received while still receiving the entire wanted data
20 pattem(s).
Generally! it may be advantageous to tune the receiver so that the selective part of the transmission bandwidth is received so that at least one data pattern to be received is centered in relation to the selective part of the transmission bandwidth. 25
Further advantageously, the receiver can be tuned to receive a selective part of said transmission bandwidth on the basis of signalling information received in a signalling pattern of a previous frame.
30 Further advantageously, each frame comprises additional data patterns succeeding said at least two data patterns in the time dimension, each of said additional data patterns having the respective same length as the corresponding one of said previous at least two data patterns. In other words, the structure of the data patterns in each frame is advantageously set up in a way that at least two data patterns are arranged next to each

9
other in the frequency dimension so that the entire transmission bandwidth is covered. Additional data patterns are then arranged in the same frame but following the at least two data patterns in the time direction, whereby each additional or following data pattern has the same length (in the frequency dimension or direction) as the previous 5 data pattern in the same frequency position. Thus, if a receiving apparatus is tuned to a specific part of the transmission bandwidth, at least two data patterns per frame are received, each of said data patterns having the same length but following each other in the time dimension. Hereby, the length of each of the data patterns in the transmitting apparatus could be adjusted dynamically. Alternatively or additionally, the number of
10 additional data patterns in the time dimension could be adjusted dynamically. Any
dynamic changes in respect to the data patterns will then be signalled in the signalling patterns. The multi-carrier system with the frame structure as suggested by the present invention thus enables a very flexible transmission of data content in which the length of data patterns, and thus the amount of data per data pattern can be dynamically
15 changed, for example from frame to frame or in any other required way. Alternatively, the length and/or the number of the data patterns may be fixed or permanent.
It has to be understood that the present invention can be applied to any kind of multi-carrier system in which a transmitting apparatus is adapted to transmit data in an entire
20 transmission bandwidth and a receiving apparatus is adapted to selectively receive only a part of said entire transmission bandwidth. Non limiting examples for such systems may be existing or future uni-directional or bi-directional broadcast systems, such as wired or wireless (for example cable based, terrestrial etc.) digital video broadcast systems. The non limiting example for a multi-carrier system would be an orthogonal
25 frequency division multiplex (OFDM) system, however, any other suitable system
could be used in which signalling data, pilot signals and other kind of data are mapped on a plurality of frequency carriers. The frequency carriers may hereby be equidistant and respectively have the same length (bandwidth). However, the present invention may also be used in multi-carrier systems in which the frequency carriers are not
30 equidistant and/or do not have the respectively same length. Further, it should be
understood that the present invention is not limited to any kind of specific frequency range neither in the overall transmission bandwidth applied on the transmitting side nor on the selected part of the transmission bandwidth to which the receiving side is tuned. However, in some applications it might be advantageous to use a receiving

10
bandwidth on the receiving side, i.e. a bandwidth for the part of the transmission bandwidth to which the receiver can be tuned, which corresponds to the bandwidth of receiving devices of existing (digital video broadcast or other) systems. A non limiting example for a receiver bandwidth may be 8 MHz, i.e. the receiving side can be tuned
5 to any wanted 8 MHz bandwidth from the overall transmission bandwidth. Hereby, the overall transmission bandwidth could be a multiple of 8 MHz, for example 8 MHz, 16 MHz, 24 MHz, 32 MHz etc' so that the segmentation of the overall transmission bandwidth, i.e. length of each training pattern could be 8 MHz. However, other segmentations are possible, ,'e.g. (but not limited to) a length of each training pattern of
10 6 MHz.
Generally, in case of the non limiting example of 8 MHz for the receiver bandwidth, the length of each of the signalling patterns used in the frame structure of the present invention could be 8 MHz, 4MHz (or less).
15 The present invention is explained in more detail in the following description of preferred embodiments in relation to the enclosed drawings, in which
Fig. 1 shows a schematic diagram of an entire transmission bandwidth from
which a selected part can be selectively and flexibly received by a receiver,
20 Fig. 2 shows an example for a segmentation of the overall transmission
bandwidth,
Fig. 3 shows a schematic time domain representation of a frame structure
according to the present invention,
Fig. 4A shows a frequency domain example of a training pattern,
25 Fig. 4B shows a time domain representation of the training pattern of Fig. 4A,
Fig. 5A shows a frequency domain representation of a further example of a
training pattern,
Fig. 5B shows a time domain representation of the training pattern of Fig. 5 A,
Fig. 6 shows a schematic frequency domain representation of an overall
30 transmission bandwidth with repetitive training patterns according to the
present invention.
Fig. 7 shows a simulation result of an auto-correlation of multi-carrier system
in which the transmission bandwidth is equal to the reception bandwidth,

11
Fig. 8 shows a simulation result for an auto-correlation in which the receiving
bandwidth coincides with a training pattern according to the present invention,
Fig. 9 shows a simulation result of an auto-correlation in case that the receiving
bandwidth does not coincide with a training pattern according to the present
5 invention, I
^j Fig. 10 shows a schematic example of a frame structure or pattern according to
the present invention,
Fig. 11 shows a part' of the frame structure of Fig. 10 with an explanation of a reconstruction of a signalling pattern,
0 Fig. 12 shows a schematic example of a receiver filter characteristic,
ji
Fig. 13 shows a further example of a frame structure of pattern according to the present invention,
Fig. 14 shows a part of a further example of a frame structure or pattern
■i
according to the present invention,
5 Fig. 15 shows a schematic representation of signalling patterns with guard
bands,
Fig. 16 schematically shows an example of a frame structure of the present
invention in the time dimension,
Fig. 17 shows a schematic block diagram of an example of a transmitting
!0 apparatus according to the present invention, and
Fig. 18 shows a schematic block diagram of an example of a receiving
apparatus according to the present invention
Fig. 1 shows a schematic representation of an entire transmission bandwidth I, in 25 which a transmitting apparatus according to the present invention, as for example the transmitting apparatus 54 schematically shown in Fig. 17, transmits signals in a multi-carrier system in line with the present invention. Fig. 1 further schematically shows a block diagram of a receiving apparatus 3 of the present invention, which is adapted to be tuned to and selectively receive a selected part 2 of the transmission bandwidth 1. 30 Hereby, the receiving apparatus 3 comprises a tuner 4 which is adapted to be tuned to and selectively receive the wanted part 2 of the transmission bandwidth 1 as well as further processing means 5 which perform the further necessary processing of the received signals in line with the respective communication system, such as a demodulation, channel decoding and the like. A more elaborate example of a receiving

12
apparatus according to the present invention is shown in the schematic block diagram of Fig. 18, which shows a receiving apparatus 63 comprising a receiving interface 64, which can for example be an antenna, an antenna pattern, a wired or cable-based receiving interface or any other suitable interface adapted to receive signals in the 5 respective transmission system or communication system. The receiving interface 64 of the receiving apparatus 63 is connected to a receiving means 65 which comprises a tuning means, such as the tuning means 4 shown in Fig. 1 as well as further necessary processing elements depending on the respective transmission or communication system, such as down conversion means adapted to down convert the received signal 10 to an intermediate frequency or the base band.
As stated above, the present invention enables a flexible and changing reception of a wanted part 2 of the transmission bandwidth 1 in a receiver by providing a specific and new frame structure for a multi-carrier system. Fig. 2 shows a schematic representation
15 of an overall transmission bandwidth 1, within which a transmitting apparatus 54 of the present invention is adapted to transmit data content, such as video data, audio data or any other kind of data, in different segments or parts 6,1, 8, 9 and 10. For example, the parts 6, 7, 8, 9 and 10 could be used by the transmitting apparatus 54 to transmit different kinds of data, data from different sources, data intended for different
20 recipients and so forth. The parts 6 and 9 have for example a maximum bandwidth, i.e. the maximum bandwidth which can be received by a corresponding receiving apparatus 63. The parts 7, 8 and 10 have smaller bandwidths. The present invention now suggests to apply a frame structure or pattern to the entire transmission bandwidth 1 whereby each frame comprises at least two training patterns adjacent to each other in
25 the frequency direction and a number of data patterns. Each training pattern of a frame will have the same length and the identical pilot signals. In other words, the overall transmission bandwidth 1 is divided into equal parts for the training patterns, whereby the maximum bandwidth to which a receiver can be tuned, for example the bandwidth shown for parts 6 and 9 in Fig. 2, has to be equal or larger than the length of each
30 training pattern. Hereby, by properly receiving an entire training pattern, a receiving apparatus 63 according to the present invention can correctly synchronize to the transmitting apparatus 54 and tune to and receive the wanted data in a flexible and non limiting way. Additionally, a frequency offset calculation and/or a channel estimation is/are possible in the receiving apparatus 63 on the basis of such a received training

13
pattern. It is further clear that the length of the various data parts in the transmission bandwidth cannot exceed the length (number of frequency carriers) of the training patterns in the respective frame as will be explained in more detail further below.
5 Fig. 3 shows a schematic representation of a time domain structure of frames 11, 11', 11" according to the present invention. Each frame 11, 11', 11" comprises a preamble symbol (or training symbol) 12, 12', 12", one or more signalling symbols 13, 13'and several data symbols 14,14'. Hereby, in the time domain, the preamble symbols or training symbols are preceding the signalling symbols which are preceding the data
10 symbols. Each frame 11, 11', 11" may have a plurality of data symbols, wherein
systems are possible in which the number of data symbols in each frame 11, 11', 11" varies. The preamble symbols are used in a receiving apparatus 63 to perform time synchronisation and eventually additional tasks, such as channel estimation and/or frequency offset calculation. The signalling symbols 13, 13', contain signalling
15 information, for example all physical layer information that is needed by the receiving apparatus 63 to decode the received signals, such as but not limited to LI signalling data. The signalling data may for example comprise the allocation of data content to the various data patterns, i.e. for example which services, data streams, modulation, error correction settings etc. are located on which frequency carriers, so that the
20 receiving apparatus 63 can obtain information to which part of the entire transmission bandwidth it shall be tuned. Further, the signalling symbols may contain signalling data indicating the offset of the respective data pattern from the preamble or training pattern and/or the signalling pattern so that the receiving apparatus 63 may optimize the tuning to the wanted part of the transmission frequency in a way that the receipt of
25 the training patterns and/or the signalling patterns is optimized. The use of the frame structure according to the present invention has the further advantage that by dividing the data stream into logical blocks, changes of the frame structure can be signalled from frame to frame, whereby a preceding frame signals the changed frame structure of the or one of the succeeding frames. For example, the frame structure allows a 30 seamless change of modulation parameters without creating errors.
Figures 4A, 4B, 5A and 5B show non limiting examples of preamble structures which could be used in the present invention. It has to be understood, however, that other possible preamble structures could also be used. Fig. 4A shows a frequency domain

14
representation of a preamble or training pattern 15 in which a plurality of frequency carriers 16 (in the shown example 2048 carriers), respectively carry a pilot signal. In other words, all frequency carriers of the training pattern 15 carry a pilot signal. Fig. 4B shows the training pattern of Fig. 4A after the transformation in the time domain. 5 The time domain training symbol comprises a plurality of time domain samples 17 (in the shown example 2048 samples) in a single repetition. In other words, the time domain training symbol does not have any repetitions in the time domain samples. Fig. 5A shows a further non limiting example of a frequency domain preamble pattern 18, comprising a plurality of frequency carriers (in the shown example 512 carriers). In the
10 shown example, only every fourth sub-carrier carries a pilot signal 19, all other sub-carriers 20 do not carry pilot signals. After transformation into the time domain, the time domain preamble or the training symbol 21 shown in Fig. 5B shows four repetitions 22, each repetition 22 having the identical samples 23 (same value and number). In the shown example, the time domain training symbol has a length of 2048
15 time samples and each repetition 22 comprises 512 samples. The general rule is that the number of repetitions in the time domain corresponds to the repetition rate of the pilot signals in the frequency domain. In case that the distance of the pilot signals in the frequency domain is higher, the number of repetitions in the time domain increases. The repetitions in the time domain preamble or training symbol are sometimes called
20 'shortened' training symbols. In the example of Fig. 5B, the time domain symbol thus comprises four shortened training symbols. In some applications it may be advantageous to use pseudo noise pilot signal sequences in order to obtain pseudo noise like signal patterns in the time domain. Also, a so called CAZAC (constant amplitude zero auto correlation) sequence could be used for the pilot signals, or any
25 other suitable sequence resulting in pseudo noise like signal patterns and having good correlation properties both in the frequency as well as in the time domain. Such sequences allow a time synchronisation in a receiving apparatus 63 of the present invention. In addition hereto, such sequences allow a reliable channel estimation in the receiving apparatus 63 in case that the Nyquist criterion is fulfilled in the frequency
30 dimension. Further, such sequences allow a frequency offset calculation and/or a channel estimation in the receiving apparatus 63.
As mentioned above, the present invention suggests a frequency domain frame structure or frame partem for the entire transmission bandwidth of the transmitting

15
apparatus 54, in which identical training patterns are repeated over the entire transmission bandwidth, i.e. immediately adjacent to each other in the frequency direction. Fig. 6 visualizes schematically such a sequence of identical and adjacent training patterns 25, 26, 27, 28 in an entire transmission bandwidth 24. In other words,
5 the same sequence of pilot signals is mapped onto the frequency carrier of each training partem 25,26, 27, 28, so that each training pattern has the same length (or bandwidth) and the same number of frequency carriers (assumed that the frequency sub-carriers are equidistant and respectively have the same length or bandwidth). Advantageously, as shown in Fig. 6, the overall transmission bandwidth 24 is equally
10 divided into the training patterns 25, 26, 27, 28 having respectively the same length. The length of the training patterns 25, 26, 27 and 28 also corresponds to the minimum tuning bandwidth to which the receiving apparatus 63 of the present invention can be tuned in order to receive signals, in order to ensure that the receiving apparatus 63 is always able to receive an entire training pattern for synchronisation (and channel
15 estimation, and /or frequency offset calculation).
The present invention therefore enables a receiving apparatus 63 to be tuned to any position within the overall channel bandwidth 24 in a very flexible manner while still being able to perform a reliable synchronisation by correlating the received pilot
20 signals for example in a correlation means 67 of the receiving apparatus 63 as shown in Fig. 18. Again, the invention suggests to divide the entire transmission frequency bandwidth 24 into adjacent sub-blocks or segments each having a training pattern containing a repetition of the identical pilot signal sequence and thus having the same length. The length of each of the training pattern thus corresponds advantageously to
25 the bandwidth to which the receiving apparatus 63 can be tuned. For example, as
shown in Fig. 18, the receiving apparatus 63 comprises a receiving interface 64, such as an antenna, a wired receiving interface or the like, to which signals are received in a receiving means 65, which comprises a tuner. If the receiving apparatus 63 is tuned to a part of the transmission bandwidth which matches or coincides to one of the training
30 patterns, the pilot signal sequence is received in the original order. If the receiving
apparatus 63 is tuned to an arbitrary part of the transmission bandwidth or for example between two training patterns, still all pilot signals of the training pattern are received, however, not in the original sequence. However, due to the cyclic behaviour of the pilot sequence sequences, very good correlation properties are still present particularly

16
if pseudo noise sequences are used for the pilot signals in each training pattern and the correlation means 67 of the receiving apparatus 63 of the present invention still delivers good results when performing an auto-correlation i.e. a correlation of the received pilot signals with themselves. Specifically, in wired systems, such as cable 5 systems, auto-correlation is expected to deliver good results because of the high signal to noise ratio. Also, such sequences enable a frequency offset calculation and/or a channel estimation in the receiving apparatus 63.
Fig. 7 shows an example of a simulation result for 64 sample pseudo noise sequence
10 for a multi-carrier system without segmentation of the training pattern, i.e. in which the transmission bandwidth is identical to the receiving bandwidth. The correlation peak is clearly visible. Fig. 8 shows a further example of a simulation result for a system according to the present invention, in which the entire transmission bandwidth comprises identical training patterns and the receiver is tuned to a part of the
15 transmission bandwidth. In the simulation shown in Fig. 8, the receiver was tuned and identically matched to the first segment, i.e. the first training pattern of the entire transmission bandwidth. In other words, the simulation shows an auto-correlation result for the situation in which the receiver receives the pilot signals of a training pattern in the original sequence. Again, the correlation peak is clearly visible. Fig. 9
20 now shows a simulation result for the system of Fig. 8, whereby the receiver was tuned to a position between two training patterns so that the receiver did not receive the pilot signals in the original sequence, but received the last part of a preceding training pattern before the first part of the succeeding training pattern. However, due to the cyclic behaviour of the pilot sequences and the training patterns, it is still possible to
25 obtain an auto-correlation peak, which is shown in Fig. 9.
In case that the receiving apparatus 63 knows its tuning position, i.e. knows the offset from the start of a frame or from the respective start of each training pattern, an optionally provided rearranging means 66 could rearrange the received pilot signals 30 into the original sequence and to perform a cross-correlation on the basis of a
comparison with a stored version of the expected training pattern in order to obtain a cross-correlation result. Such a cross-correlation result will normally have a better quality then an auto-correlation result since it is less effected by noise. Thus, for systems with low signal to noise ratios, cross correlation would be the better choice.

17
Fig. 10 shows a schematic example of a frequency domain representation of a frame structure or pattern 29 according to the present invention. The frame structure 29 covers the entire transmission bandwidth 24 in the frequency direction and comprises 5 at least two training patterns 30 adjacent to each other in the frequency direction, each carrying the identical sequence of pilot signals on respective frequency carriers and having the same length. In the example shown in Fig. 4, the entire transmission bandwidth 24 is sub-divided into four training patterns 30, but any other higher or lower number of training patterns might be suitable. In the transmitting apparatus 54 of
10 the present invention as shown in Fig. 17, a pilot mapping means 55 is adapted to map the pilot signals onto the frequency carriers of each training pattern. Advantageously, a pseudo noise sequence or a CAZAC sequence is used for the pilot signals, but any other sequence with good pseudo noise and/or correlation properties might be suitable. Also, the pilot mapping means 55 may be adapted to map a pilot signal onto every
15 frequency carrier in the training patterns, as explained in relation to Fig. 4.
Alternatively, the pilot mapping means 55 might be adapted to map a pilot signal onto every m-th frequency carrier (m being a natural number larger than 1) as for example explained in relation to Fig. 5. The length or bandwidth 39 of every training pattern 30 is the same as the bandwidth 38 to which the tuner of the receiving apparatus 63 can be
20 tuned. However, the part of the transmission bandwidth to which the tuner of the
receiving apparatus 63 can be tuned, may be larger than the length of a training partem 30. Besides for the correlation performed in the correlation means 67 in the receiving apparatus 63, the received pilots can further (after transformation into the frequency domain in the transformation means 68) be used for a channel estimation for the
25 frequency carriers in the frame in a channel estimation means 69, which provides a de-mapping means 70 with the necessary channel estimation information enabling a correct de-mapping of the data in the received data signals. Also, the received pilots can be used in the receiving apparatus 63 for a frequency offset calculation in a corresponding means which is not shown in Fig. 18.
30
The frame structure or pattern 29 further comprises at least two signalling patterns 31 adjacent to each other in the frequency direction which follow the training patterns 30 in the time direction. Each signalling pattern 31 has the same length and bandwidth as the respectively preceding training pattern 30, and the beginning and the end of each

signalling pattern 31 in the frequency direction are identical to the beginning and the end of the respective (time wise) preceding training partem 30, so that the frequency structure of the signalling patterns 31 is identical to the frequency structure of the training patterns 30. In other words, the signalling patterns 31 are aligned to the 5 training patterns 30. The transmitting apparatus 54 of the present invention shown in Fig. 17 comprises a signalling data mapping means 57 which is adapted to map signalling data onto the frequency carriers of each signalling pattern 31. Hereby, each signalling pattern 31 comprises for example the location of the signalling pattern 31 within the frame. For example each signalling partem 31 in each frame has and carries
10 the identical signalling data, except the location of the respective signalling pattern in the frame, which is different in each signalling pattern 31 in a frame. The signalling data are for example LI signalling data which contain all physical layer information that is needed by the receiving apparatus 63 to decode received signals. However, any other suitable signallmg data may be comprised in the signalling patterns 31. The
15 signalling patterns 31 might for example comprise the location of the respective data segments 32, 33, 34, 35, 36 so that a receiving apparatus 63 knows where the wanted data segments are located so that the tuner of the receiving apparatus 63 can tune to the respective location in order to receive the wanted data segments. As shown in Fig. 18, the receiving apparatus 63, after the receiving means 65 with the tuner, comprises a
20 transformation means 68 for transforming the received time domain signals into the frequency domain, where after the signalling data (after an optional reconstruction in a reconstruction means 71), are de-mapped in a de-mapping means 72 and then evaluated in an evaluation means 73. The evaluation means 73 is adapted to extract the necessary and required signalling information from the received signalling data. If
25 necessary, additional signalling patterns could be provided in the time direction immediately succeeding the signalling patterns 31.
The frame structure or pattern 29 further comprises at least two data segments extending over the entire frequency bandwidth 24 in the frequency direction and 30 following the signalling patterns 31 in the time direction. In the time slot immediately following the time slot in which the signalling patterns 31 are located, the frame structure 29 shows several data segments 32, 33, 34, 35, 36 and 37 with different lengths, i.e. a different number of respective frequency carriers onto which data are mapped. The frame structure 29 further comprises additional data segments in

19
succeeding time slots, whereby the additional data patterns respectively have the same length and number of frequency carriers as the respectively preceding data pattern. For example, the data pattern 32', 32" and 32'" have the same length as the first data pattern 32. The data patterns 33', 33" and 33'" have the same length as the data 5 segment 33. In other words, the additional data patterns have the same frequency dimension structure as the several data patterns 32, 33, 34, 35, 36 and 37 in the first time slot after the signalling patterns 31. Thus, if the receiving apparatus 63 for example tunes to a part 38 of the transmission bandwidth in order to receive the data pattern 35, all time wise succeeding data patterns 35', 35" and 35'" which have the 10 same length as the data pattern 35 can be properly received.
The flexible and variable data pattern structure of the frame structure or pattern 29 as suggested by the present invention can for example be implemented in the transmitting apparatus 54 of the present invention as shown in Fig. 17 by mapping of various
15 different data streams, for example with different kinds of data and/or data from
different sources, as visualized by the branches data 1, data 2 and data 3 in Fig. 17. The respective data are then mapped onto frequency carriers in respective data patterns by the respective data mapping means 58, 58' and 58". As stated, at least some of the various data patterns may have different lengths, i.e. different numbers of frequency
20 carriers in case that the frequency carriers are equidistant and have the same
bandwidth, respectively. Alternatively, the number of data patterns in the frequency direction may be the same as the number of training patterns, wherein the length (or bandwidth) of each data patterns may be identical to the length of each training patterns and they may be aligned to each other (have the same frequency direction
25 structure). Alternatively, each data pattern might have the same length and the number of the data patterns might be a multiple of the number of training patterns, while still having the same frequency structure and alignment. Thus for example, 2, 3, 4 or more data patterns would be aligned to each of the training patterns. Generally, the length of the data patterns needs to be smaller or at maximum equal to the effective receiver
30 bandwidth so that the data patterns can be received in the receiving apparatus 63. Further, the transmitting apparatus 54 may be adapted to change the data pattern structure, e.g. the length and/or the number of the data patterns dynamically. Alternatively, the structure of the data patterns could be fixed or permanent.

20
Further, it is to be noted that'the data patterns could advantageously comprise pilot signals mapped on some of the frequency carriers in order to enable a fine channel estimation on the receiving side. Hereby, the pilot signal could be scattered among the carriers with the data in a regular or an irregular pattern depending.
5
i
In the transmitting apparatus 54, the frequency carriers with the pilots from the pilot mapping means 55, the frequency carriers with the signalling data from the signalling mapping means 57 and the frequency carriers with the data from the various data mapping means 58, 58', 58" are then combined to a frame pattern or structure 29 10 according to the present invention in a frame forming means 59.
Generally, the frame structure of the present invention could be fixed or permanent, i.e. the overall bandwidth as well as the extension of each frame in the time direction could be fixed and always the same. Alternatively, the frame structure can also be flexible, 15 i.e. the overall bandwidth and/or the extension of each frame in the time direction
could be flexible and changed from time to time depending on the desired application.
i!
For example, the number of time slots with data patterns could be flexibly changed. Hereby, the changes could be signalled to a receiving apparatus in the signalling data of the signalling patterns.
20 ;: - ■
It can be seen in Figure 10, that the part 38 to which the receiving apparatus 63 is tuned, does not match with the frequency structure of the training patterns 30 and signalling patterns 31. However, is explained above, due to the cyclic nature of the pilot signal sequences in the training patterns 30, the correlation means 67 of the
.25 receiving apparatus 63 is still able to perform an auto-(or cross-)correlation. Further, in this situation shown in Figure 10, the receiving apparatus 63 needs knowledge about the offset of the part 38 in relation to the frequency structure of the frame pattern 29 in order to be able to re-arrange the receive signalling carriers into the original signalling sequence of the signalling patterns 31 which is done in a reconstruction means 71. This
30 is due to the fact that the signalling patterns 31 have the same length and frequency
i) structure as the training patterns 30.
During the start-up phase or initialization phase of the receiving apparatus 63, the receiving apparatus 63 tunes to an arbitrary frequency part of the overall frequency

21
bandwidth. In the non-limiting example of a cable broadcast system, the training pattern 30 could for example have a 8 MHz bandwidth. Thus, during the start-up phase, the receiving apparatus 63 is able to receive an entire training pattern 30 in the original or re-ordered sequence as well as an entire signalling pattern 31 in the original 5 or re-ordered sequence from the received training pattern 30. The receiving apparatus 63 is able to perform a correlation in the correlation means 67 in order to obtain a time synchronisation, as well as perform a channel estimation (usually a coarse channel estimation) in a channel estimation means 69 and/or a frequency offset calculation after a transformation of the received time domain signals into the frequency domain in
10 the transformation means 68. In the evaluation means 73 of the receiving apparatus 63, the received signalling data are evaluated, for example the location of the received signalling pattern in the frame is obtained so that the receiver can freely and flexibly tune to the respectively wanted frequency position, such as the part 38 is shown in Figure 10. In the new tuning position, which will usually not necessarily match with
15 the frequency structureof the training patterns 30 and the signalling patterns 31, the receiving apparatus 63 is still able to perform synchronisation, channel estimation and frequency offset calculation on the basis of the pilot signals of the training patterns 30 due to their cyclic nature. However, in order to be able to properly evaluate the signalling data of the signalling patterns 31, the received signalling signals have to be
20 re-ordered which is performed in a re-constructing means 71 as described. Fig. 11 shows this reordering in a schematic example. The last part 31' of a previous signalling pattern is received before the first part 31" of a succeeding signalling pattern, where after the reconstructions means 71 places the part 31' after the part 31" in order to reconstruct the original sequence of the signalling data, where after the
25 reordered signalling pattern is evaluated in the evaluation means 73 after a
corresponding de-mapping of the signalling data from the frequency carriers in the de-mapping means 72. It is to be remembered that the content of each signalling pattern 31 is the same, so that this reordering is possible.
30 Often, a receiving apparatus does not provide a flat frequency response over the complete receiving bandwidth to which the receiver is tuned. In addition, a transmission system usually faces increasing attenuation at the boarder of the receiving bandwidth window. Fig. 12 shows a schematic representation of a typical filter shape example. It can be seen that the filter is not rectangular, so that e.g. instead of 8 MHz

22
bandwidth, the receiving apparatus is only able to effectively receive 7.4 MHz bandwidth. The consequence is that the receiving apparatus 63 may not be able to perform the reordering of the signalling data as described in relation to Fig. 11 in case that the signalling patterns 31 have the same length and bandwidth as the receiving 5 bandwidth of the receiving apparatus 63, so that some signals are lost and cannot be received at the border of the receiving bandwidth. In order to overcome this problem, and other problems and in order to ensure that the receiving apparatus 63 is always able to receive one complete signalling patterns in the original sequence and does not have to reorder or rearrange the received signalling signals, the present invention
10 alternatively or additionally suggests to use signalling patterns 31 a which have a
reduced length as compared to the training patterns 30. The example shown in Fig. 13, it is suggested to use signalling patterns 31a which have exactly half the length of a training pattern 30, but still the same frequency structure as the training patterns 30. In other words, respective two (i.e. pairs) of the half length signalling patterns 31a are
15 matched and aligned with each one of the training patterns 30 as shown in Fig. 13. Hereby, each pair of signalling patterns 31a would have the identical signalling data including the location of the signalling patterns 31a in the respective frame. However, in relation to the other pairs of signalling patterns, in these other pairs, since they have a respective different location within the frame, the signalling data would be identical
20 except the location information. In the above example of each training pattern 30 having a bandwidth or length of 8 MHz, the signalling pattern 31 a would then each have a length or bandwidth of 4 MHz. Hereby, in order to ensure that the same amount of signalling data as before can be transmitted, it might be necessary to add additional half length signalling patterns 31b in the time slot succeeding the signalling patterns
25 31a and before the data patterns 32, 34, 35, 36 and 37. The additional signalling patterns 31b have the same time and frequency arrangement/alignment as the signalling patterns 31a, but comprise additional and different signalling information as the signalling information contained in the signalling patterns 31a. In this way, the receiving apparatus 63 will be able to receive the signalling patterns 31a and 31b
30 completely and the reconstruction means 71 of the receiving apparatus is adapted to combine the signalling data of the signalling patterns 31a and 31 b to the original sequence. In this case, the reconstruction means 71 in the receiving apparatus 63 can be omitted. It is also possible to only provide one time slot with half length signalling patterns 31a if all necessary signalling data can be transmitted in the half length and

23
the additional signalling patterns 31 b are not necessary. Alternatively, even more half length signalling patterns could be used in the succeeding time slot after the signalling patterns 31b
It should be generally (for all embodiments of the present invention) noted that the 5 length (or bandwidth) of the training patterns, the data patterns and/or the signalling patterns could be adapted to, e.g. could be smaller than or at maximum equal to, the effective receiving bandwidth of the receiving apparatus 63, for example to the output bandwidth of the receiving band pass filter, as described above.
10 Further, it should be generally noted that the training patterns, the signalling patterns and/or the data patterns of the frame structure described by the present invention could comprise additional guard bands, i.e. unused carriers at the beginning and/or the end of the respective pattern or frame. For example, each training pattern could comprise a guard band at the beginning and the end of each pattern. Alternatively, in some
15 applications it might be advantageous if only the first training pattern in each frame, in the example of Figure 10 the training pattern at position 39, could comprise a guard band only at the beginning of the pattern, and the last training pattern in each frame could comprise a guard band only at the end of the pattern. Alternatively, in some applications only the first training pattern in each frame, in the example of Figures 10
20 the training pattern at position 39, could comprise a guard band at the beginning as well as at the end of the pattern, and the last training pattern in each frame could comprise a guard band at the beginning as well as at end of the pattern. The length of the guard band comprised in some or all of the training patterns could for example be smaller or at maximum equal to the maximum frequency offset the receiving apparatus
25 can cope with. In the mentioned example of a bandwidth of 8MHz for each training
pattern, the guard band could for example have a length of 250 to 500 kHz or any other suitable length. Also, the length of each of the guard bands comprised in the training patterns could be at least the length of the carriers which are not received in the receiving apparatus due to the filter characteristics as described in relation to Fig. 12.
30 Also, in case that the signalling patterns have guard bands, the length of each of the guard bands comprised in the training patterns could be at least the length of each of the signalling pattern guard bands.
Additionally or alternatively, each signalling pattern, i.e. the signalling patterns 30, 3 la

24
and/or 31b, could comprise a guard band with unused carriers at the beginning and the end of each pattern. An example for this situation is shown in Fig. 15, which schematically shows several signalling patterns 3la arranged succeeding each other in the frequency dimension each having a guard band 31a' at its beginning and a further 5 guard band 31 a" at its end. For example, in an OFDM system in which the overall transmission bandwidth is a multiple of the training pattern length of 8 MHz (4nk mode: k is the Fourier window size of 1024 carriers/samples, n= 1, 2, 3,4 ....) and each signalling pattern has a length of 4 MHz, a suggestion for the length of each guard band at the beginning and the end of each signalling pattern would be 343 frequency
10 carriers (which is the number of not used carriers in the data patterns at the beginning and end of each frame in each 4nk mode). The resulting number for usable carriers in each signalling pattern would be 3584/2 - 2x343 = 1106 carriers. It has to be understood, however, that these numbers are only used as examples and are not meant to be limiting in any sense. Hereby, the length of each of the guard bands comprised in
15 the signalling patterns could be at least the length of the carriers which are not received in the receiving apparatus due to the filter characteristics as described in relation to Fig. 12, so that the length of the signalling data in each signalling pattern is equal to (or may be smaller than) the effective receiver bandwidth. It should be noted that if additional signalling patterns 31b are present, as explained in relation to Fig. 13, they
20 will have the same guard bands 31 a' and 31 a" as the signalling patterns 31 a. Also, the signalling patterns 30 as described in relation to Fig. 13 could have the guard bands 31a' and 31a" as described.
Additionally or alternatively, each data pattern could comprise a guard band with 25 unused carriers at the beginning and the end of each partem. Alternatively, in some applications only the respective first data patterns in each frame in the frequency direction, in the example of Figures 10 and 13 the data patterns 32, 32', 32", 32'" could comprise a guard band only at the beginning of the data pattern, and the last data patterns in each frame in the frequency direction, in the example of Figures 10 and 13 30 the data patterns 37, 37', 37", 37'" could comprise a guard band at the end of the data pattern. Hereby, the length of the guard bands of the data patterns could for example be the same as the length of the guard bands of the signalling patterns if the signalling patterns comprise guard bands, and/or the guard bands of the training patterns if the training patterns comprise guard bands.

25
As stated above the signalling data comprised in the signalling patterns 31, 31 a and or 31 b (or other signalling patterns according to the present invention) comprise the physical layer information, which enables a receiving apparatus 63 according to the 5 present invention to obtain knowledge about the frame structure and to receive and decode the wanted data patterns. As a non limiting example, the signalling data could comprise parameters such as the overall or entire transmission bandwidth, the guard band length for the training patterns, the location of the respective signalling pattern within the frame, the guard band length for the signalling patterns, the guard band
10 length for the data patterns, the number of frames which build a super frame, the
number of the present frame within a super frame, the number of data patterns in the frequency dimension of the overall frame bandwidth, the number of additional data patterns in the time dimension of a frame and/or individual signalling data for each data pattern in each frame. Hereby, the location of the respective signalling pattern
15 within a frame can e.g. indicate the position of the signalling pattern in relation to the training patterns or in relation to the segmentation of the overall bandwidth. For example, in the case of Fig. 10 in which the signalling patterns have the same length as and are aligned to the training patterns, the signalling data comprise indication if the signalling pattern is located in the first segment (e.g. the first 8 MHz segment), or the
20 second segment etc.. In case of the signalling patterns having half the length of the training patterns, as e.g. explained in relation to Fig. 13, each pair of adjacent signalling patterns then has the same location information. In any case, the receiving apparatus will be able to tune to the wanted frequency band in the succeeding frame using this location information. The individual signalling data are a separate block of
25 data individually provided for each data pattern present in the frame and may comprise parameters such as the first frequency carrier of the data pattern, the number of frequency carriers allocated to the data pattern, the modulation used for the data pattern, the error protection code used for the data pattern, the usage of a time interleaver for the data pattern, the number of frequency notches (frequency carriers
30 which are not used for data transmission in data pattern) in the data pattern, the position of the frequency notches and/or the width of the frequency notches. The signalling mapping means 57 of the transmitting apparatus 54 is adapted to map the corresponding signalling data on the frequency carriers of each signalling pattern. The evaluation means 67 of the receiving apparatus 63 is adapted to evaluate the received

26
signalling data and to use or forward the information comprised in the signalling data for further processing within the receiving apparatus 63.
In case that the signalling data comprise the mentioned individual signalling 5 information for each data pattern present in a frame, the structure of the signalling
patterns support a maximum limited number of data patterns in the frequency direction per frame in order to restrict the size of each signalling pattern to a maximum size. Thus, although the number of data patterns in the frequency direction of each frame could be dynamically and flexible changed, this would then be true only within a
10 certain maximum number of data patterns. The additional data patterns in the time direction of each frame are respectively aligned with the preceding data patterns, as explained above. Thus, each additional succeeding data pattern has the same position, length, modulation etc. as the preceding data pattern so that the signalling data for the preceding data pattern are also valid for the succeeding data pattern. Hereby, the
15 number of additional data patterns in the time direction of each frame could be fixed or flexible and this information could also be comprised in the signalling data. Similarly, the structure of the signalling patterns could support only a maximum limited number of frequency notches in each data pattern.
20 Alternatively or additionally, in order to overcome the problem that parts of the signalling patterns 31 may not be receivable in the receiving apparatus 63, the transmitting apparatus 54 could optionally comprise an error coding means 56 adapted to add some kind of error coding, redundancy, such as repetition coding, cyclic redundancy coding, or the like to the signalling data which are mapped onto the
25 frequency carriers of a signalling pattern by the signalling mapping means 57. The additional error coding would enable the transmitting apparatus 54 to use signalling patterns 31 in the same length as the training patterns 30, as shown in Fig. 10, since the receiving apparatus 63 is able, for example, by means of the reconstructing means 71, to perform some kind of error detection and/or correction in order to reconstruct the
30 original signalling pattern.
For the mentioned example of the signalling patterns having a length of 4 MHz and are aligned to training patterns (segments) of 8 MHz in an OFDM system, the following table shows a specific (non-limiting) example of a signalling structure: 35

MHz (four times the training pattern length of 8 MHz), each data partem has a minimum length of 1 MHz. The resulting maximum size of a signalling pattern is (48 + 32 + 32(36 + 4*24)) =j 48 + 32 + 4224 = 4304 Bits. An appropriate shortened Reed Salomon coding could[be applied to the signalling data. The encoded data could 5 for example be mapped onto'two consecutive QPSK symbols, or any other suitable modulation could be used.
Alternatively, the frame structure can have a maximum of 64 data patterns per frame in the frequency dimension, so that in a system with an overall bandwidth of 32 MHz
10 (four times the training pattern length of 8 MHz), each data pattern has a minimum length of 0.5 MHz. The resulting maximum size of a signalling pattern is (48 + 32 + 64(36 + 4*24)) = 48 + 32 + 8448 = 8528 Bits. An appropriate shortened Reed Salomon coding could be applied to the signalling data. The encoded data could for example be mapped onto two consecutive 16-QAM symbols, or any other suitable modulation
15 could be used.
Alternatively, the frame structure can have a maximum of 16 data patterns per frame in the frequency dimension, so that in a system with an overall bandwidth of 32 MHz (four times the training pattern length of 8 MHz), each data pattern has a minimum 20 length of 2 MHz. The resulting maximum size of a signalling pattern is (48 + 32 +
16(36 + 4*24)) - 48 + 32 + 2112 = 2192 Bits. An appropriate shortened Reed Salomon coding could be applied to the signalling data. The encoded data could for example be mapped onto one QPSK symbol, or any other suitable modulation could be used.
5 In the following, the parameters of the signalling data mentioned in the above table 1 are described in more detail:'
a) n of n4k:
n=l
30 n=2
n=3
Defines the overall transmission bandwidth of the ! proposed 4nk system as a multiple of 8 MHz 8 MHz 16 MHz 24 MHz n=4: 32 MHz

31
frequency bandwidth of the data patterns to be received. Alternatively, the receiving apparatus 63 could be tuned so that the reception of the signalling pattern 31 is optimized by placing the part 38 so that a maximum part of a signalling pattern 31 is received while the wanted data pattern is still fully received. Alternatively, the present
5 invention suggests that the length of the respective data patterns should not be different from the length of the respective preamble patterns 30 and signalling patterns 31 by more than a certain percentage for example 10%. An example for this solution can be found in Fig. 14. The borders between the data patterns 42,43, 44 and 45 are (in the frequency direction) not deviating from the borders between preamble patterns 30 and
0 the signalling patterns 31 by more than a certain percentage, such as (but not limited to) 10%. This small percentage can then be corrected by the above-mentioned additional error coding in the signalling patterns 31.
Fig. 16 shows a time domain representation of an example of frame 47 according to
15 the present invention. In the transmitting apparatus 54, after the frame pattern or
structure was generated in the frame forming means 59, the frequency domain frame pattern is transformed into the time domain by a transformation means 60. An example of a resulting time domain frame is now shown in Fig. 16. The frame 47 comprises a number of shortened training symbols 48, resulting from a mapping of pilot signals
20 only onto every m-th frequency carrier (m being a natural number larger or equal than 2) by a pilot mapping means 55, followed by a guard interval 49, a signalling symbol 50, a further guard interval 51 and a number of data symbols 52, which are respectively separated by guard intervals 53. While the situation that only a single signalling symbol is present in the time domain corresponds to the example shown in
25 Fig. 10, where only a single time slot with signalling patterns is present in the
frequency domain frame structure, the example of Fig. 13 with two time slots with signalling patterns 31a and 31b, respectively, would lead to the presence of two signalling patterns in the time domain, which are eventually separated by a guard interval. The guard intervals could e.g. be cyclic extensions of the useful parts of the
30 respective symbols. The synchronization reliability could be generally enhanced by
inverting the last training symbol, i.e. by inverting the phase of the last training symbol in respect to the preceding training symbols (which have all the same phase). In the example of an OFDM system, the signalling symbols and the data symbols, including their eventually provided guard bands, could respectively have the length of one

32
OFDM symbol. The time domain frames are then forwarded to a transmission means 61 which processes the time domain signal depending on the used multi-carrier system, for example by up-converting the signal to the wanted transmission frequency. The transmission signals are then transmitted via a transmission interface 62, which 5 can be a wired interface or a wireless interface, such as an antenna or the like.
The number of shortened training symbols 48 in frame 47 is depending on the wanted
implementation and the used transmission system. As a non-limiting example, the
number of shortened training (symbols 48 could be 8, which is a good compromise
10 between correlation complexity and synchronization reliability.
j
Fig. 16 further shows that a respective number of frames could be combined to super frames. The number of frames per super frame, i.e. the length of each super frame in
■i
the time direction, could be fixed or could vary. Hereby, there might be a maximum 15 length up to which the super frames could be set dynamically. Further, it might be
advantageous if the signalling data in the signalling patterns for each frame in a super frame are the same and if changes in the signalling data only occur from super frame to super frame. In other words, 'the modulation, coding, number of data patterns etc. would be the same in each frame of a super frame, but could then be different in the .20 succeeding super frame. Forexample, the length of the super frames in broadcast systems could be longer since the signalling data might not change as often, and in interactive system the super frame length could be shorter since an optimization ofthe transmission and reception parameters could be done on the basis of feedback from the receiver to the transmitter.
i
The elements and functionalities ofthe transmitting apparatus 54, a block diagram of
i| which is shown in Fig. 17, have been explained before. It has to be understood, that an
actual implementation of a transmitting apparatus 54 will contain additional elements
and functionalities necessary for the actual operation ofthe transmitting apparatus in
30 the respective system. In Fig. 17, only the elements and means necessary for the
explanation and understanding ofthe present invention are shown. The same is true for
the receiving apparatus 63, a block diagram of which is shown in Fig. 18. Fig. 18 only
shows elements and functionalities necessary for the understanding ofthe present
invention. Additional elements will be necessary for an actual operation ofthe

33
receiving apparatus 63. It has to be further understood that the elements and functionalities of the transmitting apparatus 54 as well as the receiving apparatus 63
i
can be implemented in any kind of device, apparatus, system and so forth adapted to perform the functionalities described and claimed by the present invention.
I
5 |
The present invention is furtner directed to a frame structure (and a correspondingly
adapted transmitting and receiving apparatus and method as described above), which,
as an alternative to the above described embodiments, does have a number (two or
more) data patterns in which 'at least one data pattern has a length which is different
10 from the length of the other data pattem(s). This structure of data patterns with a
'! variable length can be combined either with a sequence of training patterns with
II identical lengths and contents as described above, or with a sequence of training
patterns in which at least one training pattern has a length and/or a content different
from the other training patterns, i.e. a variable training pattern length. In both cases, the
15 receiving apparatus 63 will need some information about the varying data pattern length, which could be transmitted by means of a separate signalling data channel or by means of signalling data Comprised in signalling data patterns comprised in the frame structure as described above. In the later case, it might be a possible implementation if the first training pattern and the first signalling pattern in each frame
20 always have the same length so that the receiving apparatus can always obtain the
information about the varying data patterns by receiving the first training patterns and signalling patterns in every or the necessary frames. Of course, other implementations might be possible. Otherwise, the rest of the above description in relation to the training patterns, the data patterns and the signalling patterns as well as the possible
25 implementations in the transmitting apparatus 54 and the receiving apparatus 63 is still applicable.

34
1. Executive Summary
The following description is a suggestion for an advantageous implementation of the present invention in a future cable based digital video broadcast system, such as (but 5 not limited to) DVB-C2. The recent development of second generation physical layer standards for satellite (DVB-JS2) and terrestrial (DVB-T2) transmission has brought a need for cable operators to deliver an improved and competitive technical performance
and flexibility for digital broadcast and interactive services than can be achieved using
i) the current first generation DVB-C standard.
i) 10 The objective of this suggestion is to provide a complete system solution to the current
!J and anticipated future requirements of cable networks, but could be applied to
terrestrial networks as well.
This suggestion enables a significant improvement in throughput and system flexibility 15 through a number of new and improved features:
• Flexible and extremely efficient OFDM modulation scheme:
o Using not only the existing 8 MHz frequency raster but also larger
bandwidths with specified multiples of 8MHz allows an extremely
spectrally efficient transmission system to be realized.
20 o Reception based on frequency slices to allow for cost-effective receiver
implementation and increased system flexibility
!l o Notching of ©FDM subcarners to support efficient protection of
l( (security related) terrestrial services (Accumulated radiation of cable
li networks disturb terrestrial services)
25 • High order modulation of OFDM subcarriers provides a significant increase in
throughput over current DVB-C systems
o Up to <59.8Mbit/s using 1024QAM subcarrier modulation
(in 8 MHz reception bandwidth)
o Up to 83.7Mbit/s using 4096QAM subcarrier modulation
30 (in 8 MHz reception bandwidth)
• LDPC codec re-used from DVB-S2 and DVB-T2 with code rates optimized for
a cable system provides greater than 3dB gain over the current coding and
facilitating compatibility with second generation DVB systems

35

10

• Support for transcoding of satellite and terrestrial services to cable systems
• Support for several input stream formats (single and/or multiple Transport Streams (TS) and Generic Stream Encapsulation (GSE))
• Optimisation of throughput where return channels are available
o Low system latency to support interactive services
i
o Adaptation of OFDM subcarriers to optimise the throughput depending
on the location and frequency slice specific SNR conditions.
f| This suggestion is a complete system proposal and addresses all aspects of the
requirements. A detailed comparison to the C2 related requirements is given together
with the description of technology in paragraph 5.

The following abbreviations' are used:
ACM Adaptive Coding and Modulation
15 AWGN Additive White Gaussian Noise
BCH Bose-Chaudhuri-Hocquenghem multiple error correction binary block
code |
CAZAC Constant Amplitude Zero Autocorrelation Waveform
CCM Constant Coding and Modulation
20 CRC Cyclic Redundancy Check
FEC Forward Error Correction
GI Guard Interval
GS Generic Stream
GSE Generic Stream Encapsulation
25 GSM Global System for Mobile Communication
LDPC Low Density (Parity Check code
OFDM Orthogonal Frequency Division Multiplex
PAPR Peak to Average Power Reduction
PSI/SI Program Specific Information/Service Information
30 QAM Quadrature Amplitude Modulation
QoS Quality of Service
RF Radio Frequency
SMATV Satellite Master Antenna Television
SNR Signal to Noise Ratio
35 TS Transport Stream
VCM Variable Coding and Modulation
VoD Video on Demand
It has to be understood that all functionalities and requirements described below can be
40 implemented in the respectively suited means and elements of the transmitting apparatus 54 shown in and described in relation to Figure 17 and/or the receiving apparatus 63 shown in and described in relation to Figure 18. Further, it is to be

36 if understood that the below detailed description of an advantageous implementation is
not intended to restrict the scope of the present invention as defined in the claims.
i 2. System overview
2.1. Flexible «4k system ■
i
The proposed system includes a high level of flexibility regarding the mapping of the different input formats (single/multiple TS and GSE) onto the OFDM subcarriers.
The basic concept is to bundle and multiplex as many input streams as possible onto a related number of OFDM subcarriers that overall do not exceed the maximum tuner bandwidth on receiver side (e.g. 8 MHz, including the related guard bands). This is defined as a frequency data slice.
I
A subchannel denotes one 8MHz bandwidth block of the existing cable channel raster.
;i
The current DVB-C bandwidth (i.e. 8 MHz) can be used as a single channel. However
in order to increase the spectrum efficiency further n 8 MHz wide OFDM subchannels
can be combined or "bundled" together to create a larger channel. Several frequency
data slices may be combined within a channel. There is no fixed frequency bandwidth
) assignment for frequency slices, they are not necessarily aligned to the 8 MHz
subchannels. l>
Spectral efficiency is increased since the guard bands of the OFDM spectrum areused only once at each side of the overall channel bandwidth. The spectral shaping of the 5 guard band does not change with different channel bandwidths. Figure 19 illustrates
different channel bandwidth'examples with the related guard bands.
i


Figure 19: Example bandwidths of the proposed DVB-C2 OFDM channels
5 It is obvious that the higher the overall channel bandwidth the lower the spectral overhead of the guard bands. The upper limit of the overall channel bandwidth depends on the available technology (D/A converter) at the headend side. Table 22 illustrates the overhead percentage for different OFDM spectrum bandwidths if the same guard band shaping is applied: 10


5 The frequency data slice bandwidth is not related to any fixed frequency raster and can
be adjusted in a straight-forward way according to the bandwidth demands of the input
streams. The only requirement is that the number of allocated subcarriers does not
i exceed the tuner bandwidth on receiver side. Statistical multiplexing is applied over
i
(he data slice and benefits from bandwidths that are as large as possible.
i
10
The overall channel bandwidth should be a multiple n of the subchannel raster (8 MHz). This allows for simple network planning as well as sufficiently high tuning step sizes in the receiver tuner. The OFDM modulation is derived from the 4k operation mode used in DVB-H/T2,
!
15 being extended to a multiple jof the subchannel raster. The system is therefore called
«4k system (n indicating the number of bundled 4k modulation blocks)
I
2.2. Partial OFDM reception
To allow a cost efficient receiver implementation, an OFDM reception based on 20 frequency slices is proposed.
Segmented OFDM reception with fixed segment sizes has already been currently
I!
successfully deployed in ISDB-T. In these systems reception of individual segments or combined segments is possible. The main application in ISDB-T is to provide 25 mobile reception as well as fixed terrestrial reception within one RF channel.
The proposed C2 system contains an arbitrary adjustable assignment of subcarrier
\blocks, as shown in Figure 20. The proposed C2 headend is able to calculate for each
i) superframe the input stream specific distribution and frequency slice assembly of all
30 OFDM subcarriers. Ideally each input stream or each group of input streams is mapped
on the related subgroup of OFDM subcarriers. The number of allocated subcarriers can
be derived directly from thelinput data rate. This includes the combined overhead of

39
mode adaptation, stream adaptation and FEC encoding and the gain due to the QAM
modulation. i
The partitioning of the overall OFDM channel into .the different frequency slices (also called frequency patterns or 'segments) is defined by the LI signalling (section 3.7.2). 5 The receiver tunes to the frequency that contains the wanted frequency data slice. The partial OFDM demodulation is applied to the selected 8MHz receive spectrum.

Figure 20: Partial reception of a broader OFDM transmit signal
Note: The width of the frequency data slice might be smaller than the receiver reception bandwidth. In this case the receiver selects after the OFDM demodulation only the information of the relevant subcarriers and forwards them to the following 15 decoding sections.
2.3. C2 system overview
■i
The following Figure 21 illustrates the top layer block diagram of the proposed C2
f
system:


! In a" first step in the proposed transmission system, the different input streams (single or multiple TS or GS) are merged and packetized to baseband packets similar to DVB-S2. This kind of mode adaptation allows stream specific (i.e. TS or GS) adjustment of the desired robustness level. It is possible to feed a single TS or GS onto a rather small number of OFDM subcarriers. However, in order to increase the diversity of the
) subchannel (i.e. by applying the frequency interleaver over a larger number of
subcarriers) it is advantageous to bundle as many input streams as possible to approach the maximum possible bandwidth (i.e. tuner bandwidth on receiver side).
The next stage is the stream adaptation stage which performs padding (if needed) and 5 applies baseband scrambling before the FEC encoding is applied.
The FEC encoding stage comprises a BCH encoder, a LDPC encoder as well as a bit
i) interleaver unit, similar to those used in DVB-T2. The normal output block size of the
I!. LDPC encoder is 64800 bits. However, in order to support low latency (e.g. as
f; 0 required by e.g. interactive services), shorter LDPC block sizes are also supported (i.e.
16200 bit as known from DVB-T2).
In order to remove error floors for high QAM constellations (1024-QAM and above),
an adjusted BCH with t-error correction of 12 bits is used .

41
Next the LDPC encoded FEC frames enter the BICM (Bit Interleaved Coded Modulation) stage. Here the output of the LDPC encoder is bit interleaved as in DVB-T2, with concatenation of parity interleaving followed by column twist interleaving and a demultiplexer. Bit interleaver extensions for the new, higher QAM constellations 5 are included in this document.
Afterwards the QAM encoder maps incoming bits into complex QAM symbols. QAM mapping is based on Gray coding, and an extension of the T2 mapping for 1024-QAM and 4096-QAM is proposed.
10 To provide flexible settings to cope with different requirements and environments, the modulation and FEC parameters can be modified. The proposed system provides two different operation modes:
* For broadcast streams, the settings of modulation and coding for each data slice
(i.e. the related number of OFDM subcarriers) are adjusted solely on transmitter
15 side. The settings are chosen to guarantee the desired Quality of Service level
within the whole network. Modulation and coding for each data slice can be changed from superframe to superframe. Each subcarrier within a data slice (also called data pattern or segment) has the same modulation and coding.
* If the cable network provides a return channel, the receiver can inform the
20 transmitter about its SNR condition in order to optimize the chosen modulation and
coding. This is especially of importance for optimizing the throughput for point to point interactive services (IP based, e.g. DOCSIS Internet traffic or Video on Demand VoD). Smaller multicast connections could also benefit from the SNR information if the transmitter selects modulation and coding according to the
25 receiver with the overall worst SNR on the related data slice.
The next stage is a time interleaver which can reduce the impact of impulsive noise and other noise bursts. The time interleaver is aligned to the overall frame length and can be switched off for time critical services, such as interactive services requiring low 30 latency.
Frequency interleaving is used to average the SNR ripple over the frequency slice width. Although the basic architecture is based on the frequency interleaver from DVB-T and DVB-T2, the width of the frequency interleaver is variable and is matched

42
to the number of subcarriers that are allocated by the specific data slice. The frequency interleaver specific memory inapping and demapping on the transmitter as well as on the receiver side can easily be done during operation.
The output signal of each symbol interleaver is then mapped onto one data slice (also 5 called data pattern):
The OFDM symbol builder combines all different incoming streams by mapping all of them on to the related necessary number of subcarriers, including the insertion of the appropriate pilot pattern. !
"r
10 The overall number of subcarriers for one OFDM symbol increases with an increasing
number of bundled 8 MHz channels (n4k system). The alignment of these data slices
does not have any segmentation restrictions, as shown in Figure 22. The only
l) requirement is that the width of one data slice (i.e. the number of allocated subcarriers)
must not exceed the receiver bandwidth (i.e. 8 MHz respectively the pass band
15 bandwidth of the receiver frontend.).

Figure 22: Data slicing in the overall channel bandwidth (32 MHz channel example)
20 The proposed frequency slicing provides a very efficient mapping of the accumulation
i
of the bandwidth demand of all different input streams onto an overall large bandwidth without any significant stuffing overhead.
!
f Afterwards a guard interval is pre-pended to each OFDM symbol. This document
25 proposes three different guard interval lengths to provide the possibility to optimize the
guard interval to the network specific environment (i.e. maximum echo length).

In the final framing section, each 320 data OFDM symbols are separated by a preamble, consisting of a training sequence phase (allowing all important synchronization as well as initial channel estimation functions) and two 16.QAM 5 modulated LI signalling symbols (containing all important physical layer information for the upcoming frame).
2.4. DVB-S / DVB-S2 service transcoding
11
10 For transcoding satellite streams into CI cable networks the block diagram in Figure
i 21 is usually valid: TS level is used as interface between satellite decoding and C2
specific encoding. The TS based output streams of the DVB-S system are therefore
encoded according to the upper signal chain.
i
15 In order to perform correct adaptation of all PSI/SI information entries within all
transport streams, an additional PSI/SI reprocessing block is included at the beginning of the proposed C2 encoding.

Figure 23: DVB-S(2) transcoding: Interface is TS level, including PSI/SI processing
20 ■ I
Note: The same TS based processing might be applicable for transcoding DVB-T or
| DVB-T2 transport streams into the cable network.

44
For SMATV headends that transcode DVB-S2 services into smaller cable networks, PSI/SI processing might not applied (similar to DVB-C SMATV systems). In this case it is not needed to reverse all encoding steps to insert the signals into the cable network. Moreover, the DVB-S2 signals are decoded only until baseband packet levels. These baseband packets are then inserted directly into the proposed C2 system. Figure 24 shows the related block diagram:

i Figure 24: Block diagram of a SMATV headend: Baseband packets of DVB-S2 services are used
as input data i;

45
3, System Description
3.1. Mode adaptation
Mode adaptation is reused as much as possible from DVB-S2. The system works with
Transport Stream Input or Generic Stream Input (DVB GSE protocol to adapt an IP
stream into a generic stream)'. Both formats support single and multiple input stream
'I modes, as shown in Figure 25.
This kind of mode adaptation allows stream specific (i.e. TS or GS) adjustment of the desired robustness level. The higher the SNR, the higher the 'ModCod' mode is used
i
(i.e. combination of modulation scheme and selected FEC mode).
In a cable channel the level of SNR ripple is limited compared to terrestrial systems. -
Therefore the emphasis in this proposal is on simplicity and reduction of the signalling
overhead. |
Similar to DVB-S2 various stream configurations are supported to provide the required system flexibility:
* Single transport stream input (CCM): All services of the input stream are protected
i with the same FEC level1 by the system. VCM is not directly available at single
transport stream level.
• Multiple transport stream input (CCM and VCM):
o Each transport stream can be protected separately with a single FEC level o Protection may be differentiated in different transport streams (VCM).

3.2.3.1. Bit interleaver
A bit interleaver shall be employed to optimize assignments between LDPC code bits 10 and Gray mapped QAM symbol bits. As in DVB-T2 it shall consist of a block interleaver and a de-multiplexer.
In the block interleaver part, as shown in Figure 26, the output of the LDPC encoder shall be parity interleaved first and then be stored into memory of Nc columns and Nr 15 rows. The data are written column by column with column twisting offset tc and are read row by row.

3.2.3.2. Time Interleaver
10 To mitigate the influence from impulsive or burst noise, a time interleaver is proposed
i| ■ for broadcast services. The interleaving length of the time interleaver is kept short in
comparison to DVB-T2.
Figure 27 shows the time interleaver operation. The time interleaver takes the output 15 from the QAM encoder and writes the data into columns. The outputs are passed to the frequency interleaver by reading out the interleaver cells in rows.
• The number of rows.R is a fixed value of 40. This value assumes a 2.5%
erasure rate, i.e. approximately one in every 40 symbols is lost due to
20 interference. I
• The time interleaver length is aligned to the frame length (section 7.5) for
i! simplicity.

50
• ine numoer 01 couimns NL in the time interleave matches the number of
subcarriers in the required service.
• The usage of time interleaving for each block of the segmented OFDM system
is signalled in the LI packets.
5 • Transmitter memory requirement: 4096*12*40 = 1966080 = 1.97 Mbit


51
calculated dynamically by the transmitter as well as the receiver (i.e. the interleaver size varies depending on the number of subcarriers allocated).

53
3.3. QAM subcarrier modulation
The modulation of the OFDM subcarriers shall be regular Quadrature Amplitude Modulation 5 (QAM): The use of the following constellations is proposed, based on the definitions ofDVB-T2.
• 16-QAM
• 64-QAM
• 256-QAM
10 In order to increase the throughput rates of the proposed C2 system, the following higher constellations are proposed for broadcasting services
• 1024-QAM (using Gray mapping)
• 4096-QAM (using Gray mapping)
i
15 In addition, even higher QAM constellations may be applicable for interactive services
that can exploit the advantages of ACM (adaptive coding and modulation), i.e.
transmitter and receiver exchange OFDM tonemaps that signal the chosen QAM
constellation for each data slice. SNR-dependant adjustment of the chosen
I) constellation and coding is possible.
ii
20 il i
3.4. OFDM parameters I
This chapter proposes the OFDM structure to use for each transmission mode. The
transmitted signal is organized in frames as described in section 3.5.. Each frame has
duration of TF, and consists of LF OFDM symbols. Each symbol is constituted by a set
\\ 25 of K carriers transmitted with a duration Ts. It is composed of two parts: a useful part
l| with duration Tu and a guard interval with duration A. The guard interval consists of a
cyclic continuation of the useful part, Tu, and is inserted before it.
l| The symbols in an OFDM frame are numbered from 1 to LF- All symbols contain data
l| and/or reference information.
30

54
Since the OFDM signal comprises many separately-modulated carriers, each symbol
i i
can m turn be considered to be divided into cells, each corresponding to the modulation carried on one carrier during one symbol.
5 The OFDM symbols contain pilots that can be used for frame synchronization,
frequency synchronization, time synchronization, channel estimation, and can also be used to track the phase noise.
The carriers are indexed by k e [K,^,,; K^J and determined by Kmin and Kmax . The
i
spacing between adjacent carriers is l/Ty while the spacing between carriers K^,, and
i 10 K^ are determined by (K.-\\)/Tu.
i
The OFDM parameters are summarised in Table 10. The values for the various time-related parameters are given in multiples of the elementary period T and in microseconds.
15
A «4k mode of operation is proposed as a good trade-off between symbol length,
i
phase noise sensitivity as well as spectrum side lobe steepness. This is based on the
DVB-H/T2 4k mode, withinfan 8 MHz channel. The system bandwidth can be
l| extended to n multiples of 8 MHz.
■i
20 The following table illustrates the settings for several channel bandwidths where n varies from 1 to 4.


Table 10: 4nk OFDM modulation for different channel bandwidths
5 With the added advantage of channel bundling, the proposed OFDM values are very, similar to the main parameters of the DVB-H/T2 4k mode, including carrier spacing as welt as the symbol duration.'!
Note:
10 Other channel bandwidths can be obtained by adj usting the elementary period T. For
f example, a 6 MHz channel bandwidth can be derived from an 8MHz channel by
i| changing the elementary period from 7/64 \is to 7/48 |is.
li
3.5. Framing 15 The framing structure is shown in Figure 29, which is similar as to the above described Figure 16. The super frame is divided into C2 frames which are further divided into OFDM symbols. A C2 frame always starts with one preamble symbol then two layer 1 signalling symbols and finally Lp-3 data symbols.

57
LI signalling dafa can be changed only at super frame boundaries. For broadcast-only
services the super frame period can be set to its maximum value of (216 - 1)* TF ,
which is approximately 2 hours 37 minutes, as LI signalling parameters are not
envisaged to change frequently. For either interactive-only or mixed
5 broadcast/interactive services then the super frame length can be shortened as required.
^ The super frame period is provided as an LI signalling parameter.
Zapping time without knowledge of the frequency data slice location is expected to require up to two complete G2 frame periods (288ms) depending on the relative timing 10 of the start of the channel change to the start of a C2 frame.
3.6. Pilot carriers in data symbols
The scattered pilot density is derived from
15 • The maximum delay length of the multipath channel to determine the repetition
rate in frequency direction • The maximum Doppler frequency of the cable channel to determine the repetition rate in time direction



3.7. Preamble
20 A preamble defines the start of a new C2 frame. The preamble must allow the following functionalities:
• Frame and initial OFDM symbol synchronization
• Initial offset correction (frequency and sampling rate offsets)
• Initial channel estimation
i
25 • Information about the basic physical layer parameter for the next frame: o Guard Interval ■

o OFDM subcarrier allocation
l|i
■ Basic structure of different subcarrier segments
• Start / stop carrier, block width, ...
■ Segment specific subcarrier modulation scheme
5 ■ Segment specific subcarrier FEC settings
o Frequency notchfindication
The preamble is divided into a training sequence phase and a LI signalling phase. The
training phase consists of 8 shortened training symbols; the overall length is one
t 10 OFDM symbol (4096 samples). The succeeding two OFDM symbols contain the LI
signalling (including the related Guard Interval).



5 The proposed preamble provides all typical important functionalities independent from the tuning position.
• Time / frame synchronization
• Coarse / fine frequency offset estimation
.1
• Initial channel estimation
10 • LI signalling |
The ability to perform all preamble functionalities independent from the tuning position allows the usage of an arbitrary data slicing in the frequency domain. In particular the width (bandwidth) of the data slices does not have to be aligned to any 15 fixed segment size. The functionality of the different blocks is described below:

63
3.7.2. LI signalling
,L1 signalling provides information on all relevant physical layer specific parameters.
As illustrated in Figure 32, VI signalling follows the training sequence phase in each 5 frame. The duration of the LI signalling is two OFDM symbols. The bandwidth of the LI signal is 4MHz, eachtwo^l blocks (also called signalling patterns) are aligned to the initial 8 MHz raster.
i
The frequency behaviour of the LI signalling has to reflect typical filter characteristics of a receiver as well as the overall spectrum mask:
10 In order to allow proper LI decoding on each arbitrary tuning position, the LI block does not use all subcarriers in its 4 MHz block. Additionally the guard band characteristics from the overall channel bandwidth are reused. In any n4k mode, 343 subcarriers on each border are not used for data transmission (guard band). The same number of unused carriers is used for the LI signal, therefore the available number of
15 carriers per LI block is:
3584/2 - 2*343 = 1106 carriers
li Figure 34, which is similar to Figure 15 explained above, illustrates the carrier
l| allocation of the LI symbols (signalling patterns):
ll
20 j|

66
Calculating the resulting maximum number of LI signalling bits results in an overall number that fits into two consecutive (in time direction) QAM modulated LI symbols with 4 MHz bandwidth, including the overhead of an appropriate FEC scheme.

■I either aligned or not aligned to the 8 MHz raster in cable networks. In this position, the
tuning window covers a complete preamble sequence and two complete LI signalling
blocks. Therefore the receiver is able to synchronize, to perform an initial channel
estimation and to extract Ll'signalling. From the LI signalling, e.g. the current n of
5 n4k information, the receiver has knowledge of the location of the received and
\ decoded signalling pattern(s) in relation to the present frame and can then tune to the
frequency of the wanted data slice (the data slice is typically not aligned to the 8MHz
raster) and is able to receive and decode all wanted data slices in all succeeding frames
of this superframe in this tuning position.
10

3.8. Data slicing
As explained in the previous chapters, the preamble is designed in a way that allows all
important frame related functionalities (i.e. receiver synchronization, channel
j
estimation and LI decoding) in any tuning position.
i
L 5 Therefore the data slices, i.e.? the data patterns as explained in relation to Figure 10, need not follow any fixed segment allocation. An appropriate number of OFDM subcarriers can be allocatedTThe only condition regarding the width of one data slice is that it does not exceed the receiving bandwidth (i.e. 8 MHz minus 2* guard band (e.g. 7.6. MHz)). 10 Each data slice has aconstant number of data bits (i.e. data carriers) per frequency slice per super frame. This number of data bits per data slice may change from superframe to superframe.
ii
Figure 35 illustrates the building of the overall OFDM signal as the combination of
i 15 several OFDM subblocks (data slices). Each signal encoding chain is mapped onto the
ij matching number of subcarriers.

I .1
The smaller the bandwidth of the data segment the lower the interleaving gain from the
i'l 5 frequency interleaves Bundling of several streams with the same QoS requirements in
the mode adaptation is one method to deploy frequency diversity in the best possible
way.

72 3.9. Notching
Terrestrial services and the DVB cable system often share the same frequency range. Interferences between both services reduce the SNR of the affected service. Radiation . from cable networks disturbs the operation of the terrestrial services. Similarly, the transmission quality of cable services suffers from the ingress of terrestrial services by causing additional noise on the cable medium. An example is shown in Figure 36. Notching of OFDM carriers is used to protect the different communication systems
from each other: The OFDM carriers allocated to the same frequency range(s) are
i omitted from the data communication.
I An example of the system described above on terrestrial side is flight security services
It and many more.


73
3.10. OFDM adaptivity for Interactive Services
If the cable network is return channel capable, the proposed C2 system shall be usable
as downstream medium for interactive data services, as shown in Figure 37.
Similar to the existing DVB-C system the C2 system shall be able to integrate the
5 DOCSIS downstream data traffic. The upstream channel is provided in a DOCSIS
*: compliant way and is out of the scope of this document

10
Figure 37: C2 as downstream channel for DOCSIS data
Examples of these kinds of interactive services are all DOCSIS based data communication, including all IP based services or Video on Demand (VoD). 15


Figure 38: DOCSIS communication in the proposed C2 system
5 In this scenario the proposed system is able to deploy the advantages of adaptive
OFDM (ACM - adaptive coding and modulation): For interactive point to point
i communication services the modem and the transmitter can exchange their SNR
i. conditions in their assigned frequency slice in order to optimize their data throughput.
This technique provides a precise protection of the data slice as well as dynamic link
10 adaptation to propagation conditions by targeting each individual terminal (C2 modem
/ receiver).
In Figure 38 an example cable network with the C2 headend and a number of
connected C2 receivers / modems is depicted. Depending on channel influences like
ij 15 attenuation or multipath ripple the available SNR in each location changes. For
example, the C2 modem/receiver 1 is quite close to the headend and therefore any
attenuation in the downlink spectrum is low. The modem will inform the headend
about its good channel conditions, the headend selects an appropriate combination of

75
modulation and coding with a very high throughput rate. In contrast, the distance between the C2 headend and'the C2 modem/receiver 2 is assumed to be very long, resulting in a higher attenuation in the receive spectrum. The available SNR range is therefore significantly lower,'the C2 modem/receiver 2 informs the C2 headend to use 5 a more robust combination of modulation and coding.
Theoretically it would be possible to signal the SNR condition of each individual
OFDM subcarrier back to the C2 headend. Another alternative that is widely used in
I
other communication systems like PLC (Powerline Communication System) is to send
10 one SNR value per coherence bandwidth slot.
However, this document proposes to use just one overall combination of modulation and coding for each data slice that is used for interactive services. The main reasons are:
• LI signalling / OFDM tonemap complexity: If each subcarrier or coherence
15 bandwidth slot would be'treated separately, the overall amount of LI signalling
data as well as the OFDM tonemap data (i.e. feedback data containing information containing the carrier specific SNR conditions) would increase significantly.
• Limited SNR ripple: Due to the rather low amplitude levels of the echo signals the
resulting SNR variation in the related frequency slice of the receive spectrum is not
20 too big (e.g. overall frequency slot ripple below 3 dB). Carrier specific SNR
treatment is typically targeted by using different modulation schemes for different OFDM subcarriers while the complete data slice is encoded with same FEC settings (i.e. LDPC coding). The overall small level of amplitude ripple can't be
covered in an efficient way by the rather high SNR steps between the different
!l
25 constellations (e.g. roughly 6 dB between neighboured square constellations).
• If interactive service data slices select just one overall modulation and coding
setting, they fit very well to the overall proposedC2 architecture in which each
different broadcast stream is allowed to use a specific 'modcod' setting, too. .
Despite the additional!exchange of SNR conditions between transmitter and
30 receiver the system uses exactly the same data slicing and LI signalling
mechanisms.

76 Note: The message formats of exchanging SNR conditions or signalling appropriate
i
combinations of modulation and coding is the subject of higher layers and is out of scope of this proposal.

To improve the out of band OFDM spectrum characteristics and to achieve the 20 required isolation between channels at the border frequency between two channels, filtering is required. Basically two methods are applicable.
• Windowing: Makes the amplitude go smoothly to zero at the symbol
boundaries (time domain). The windowing in the time domain means the
resulting spectrum is a convolution of the spectrum of the windowing function
25 with a set of impulses at the subcarrier frequencies.
• Conventional filtering techniques (digital and/or analogue)

I Windowing and filtering are dual techniques to reduce out of band spectrum.
The cut off behaviour of conventional filtering has a potential impact on the
performance of high subcarrier modulation modes. In contrast, windowing in the time
5 domain causes no system degradation. The drawback of windowing is a partial overlap
A between consecutive symbols and the related degradation of the usable guard interval
fraction. Figure 40 shows the basic principle of windowing.

10 Figure 40: Windowing of OFDM symbols in time domain
The overlap between consecutive OFDM symbols has the duration TTR. The higher the TTR value, the more the level of the out of band spectrum is reduced.
15 The final channel isolation as well as the related side lobe attenuation has to be investigated in adjacent channel system simulations.
!
3.12. PAPR
20 Low complexity solutions for PAPR reduction on the transmitter side should be
investigated. Larger overall ;FFT sizes by channel bundling are expected to increase the probabilistic crest factor of the OFDM system slightly. For example, the usage of a 32KIFFT on transmitter side is expected to increase the probabilistic crest factor of the OFDM system by less then 0.5dB as compared to an 8K FFT based transmitter.
Additionally it is known that an increase of the order of the QAM modulation has no negative effect for the probabilistic crest factor for OFDM systems with 1K or higher

FFT sizes. Therefore the optimization problem for crest factor reduction for the proposed OFDM system is similar to the one for DVB-T2.
It should be noted that the method of active constellation extension will be less
5 efficient than in DVB-T2due,to the very high order QAM constellations typically used
ii'
for cable transmissions.

79
4. System performance / throughput
(
4.1. Throughput rates
The following tables list different throughput rates of the proposed «4k C2 system for 5 8MHz and 32MHz channel bandwidths. In addition a comparison to the current maximum DVB-C throughput is given (DVB-C 256-QAM). The calculation considers the following system overhead:
• Guard interval (1/64.U28,1/256)
• LDPC codec ! 10 • BCH codec
• Pilot pattern overhead
1
• Framing overhead (3 preamble/signalling symbols out of 323 symbols)
Note: Potential windowing overhead to increase the OFDM spectrum shaping is not 15 (yet) considered.

Claims
I. Transmitting apparatus for transmitting signals in a multi carrier system on the basis of a frame structure, each frame comprising at least two signalling patterns adjacent to each other in the frequency direction and at least two data patterns, said transmitting apparatus comprising
signalling mapping means adapted to map signalling data on frequency carriers of each of said at least two signalling patterns in a frame, each signalling pattern having the same length,
. data mapping means adapted to map data on frequency carriers of said at least two data patterns in a! frame, transforming means adapted to transform said signalling patterns and said data
patterns from the frequency domain into the time domain in order to generate a
>l time domain transmission signal, and
I) transmitting means adapted to transmit said time domain transmission signal.
2. Transmitting apparatus according to claim 1,
wherein each frame comprises at least two additional signalling patterns
succeeding said at least two signalling patterns in the time dimension, each of
i| said additional signalling patterns having the respective same length as the
'1
corresponding one of said at least two preceding signalling patterns.
3. Transmitting apparatus according to claim 1 or 2,
wherein each frame comprises at least two training patterns, said transmitting apparatus comprising a pilot mapping means adapted to map the pilot signals on frequency carriers of each training pattern in a frame, and wherein the signalling patterns are aligned to the training patterns in the frequency direction.

4. Transmitting apparatus according to claim 3,
wherein every training pattern has the same length.
il
5. Transmitting apparatus according to one of the claims I to 4,
'I wherein each signalling pattern of each frame comprises the location of the
respective signalling pattern in the frame.
6. Transmitting apparatus according to one of the claims 1 to 5, wherein the signalling patterns of each frame comprise signalling data indicating the number of data patterns comprised in the frame.
7. Transmitting apparatus according to one of the claims 1 to 6,
wherein the structure'lof the signalling data in the signalling patterns supports a limited maximum number of data patterns in the frequency direction of each frame.
i
8.' Transmitting apparatus according to one of the claims 1 to 7,
wherein the signalling patterns of each frame comprise individual signalling data for each data pattern comprised in the frame.
9. Transmitting methodTor transmitting signals in a multi carrier system on the
basis of a frame structure, each frame comprising at least two signalling
patterns adjacent to each other in the frequency direction and at least two data
patterns, comprising the steps of
mapping signalling data on frequency carriers of each of said at least two signalling patterns in a frame, each signalling pattern having the same length, mapping data on frequency carriers of said at least two data patterns in a frame, transforming said signalling patterns and said data patterns from the frequency
j
domain into the time domain in order to generate a time domain transmission
signal, and
i transmitting said time domain transmission signal.
10. Frame pattern for a multi carrier system, comprising at least two signalling
patterns adjacent to each other in the frequency direction and at least two data
I
il

99
patterns, wherein signalling data are mapped on frequency carriers of each of said at least two signalling patterns in the frame, each signalling pattern having the same length, and wherein data are mapped on frequency carriers of said at least two data patterns in the frame.
11. Receiving apparatus for receiving signals in a multi carrier system on the basis
of a frame structure in a transmission bandwidth, each frame comprising at
least two signalling patterns adjacent to each other in the frequency direction
i
each with signalling data mapped on frequency carriers and at least two data
patterns with data mapped on frequency carriers, each of said at least two
signalling patterns having the same length,
said receiving apparatus comprising
receiving means adapted to be tuned to and to receive a selected part of said
transmission bandwidth, said selected part of said transmission bandwidth
having at least the length of one of said signalling patterns and covering at least
one data pattern to be received, and
evaluation means adapted to evaluate the signalling data comprised in a
received signalling pattern in order to enable the receipt of said at least two
data patterns. I
12. Receiving apparatus according to claim 11,
'I comprising a reconstructing means adapted to reconstruct the original
signalling pattern from said received selected part of said transmission
bandwidth.
i
13. Receiving apparatus 'according to claim 12,
wherein said reconstructing means is adapted to rearrange received signalling signals into the original signalling partem in case that the selected part of said transmission bandwidth to which the receiving means is tuned does not match with the signalling pattern structure.
14. Receiving apparatus according to claim 12,
wherein each frame comprises at least two additional signalling patterns
ij succeeding said at least two signalling patterns in the time dimension, each of
l!

100
said additional signalling patterns having the respective same length as the corresponding one of said at least two preceding signalling patterns, wherein said reconstructing means is adapted to rearrange received two or more signalling patterns succeeding each other in the time dimension into the original signalling pattern.
15. Receiving apparatus according to one of the claims 11 to 14,
wherein the signalling data of the signalling patterns comprise an error
correction coding, and wherein said reconstructing means is adapted to
perform an error correction decoding on said received signalling signals in
i
order to reconstruct the original signalling pattern.
16. Receiving apparatus according to one of the claims 11 to 15,
wherein the signalling patterns of each frame comprise signalling data with the location of each signalling pattern in the frame, wherein said evaluation means
is adapted to extract said location information.
f
r
17. Receiving apparatus;according to one of the claims 11 to 16,
wherein the signalling patterns of each frame comprise signalling data with the number of data patterns comprised in the frame, wherein said evaluation means is adapted to extract said signalling data with the number of data patterns from a received signalling pattern.
18. Receiving apparatus according to one of the claims 11 to 17,
wherein the signalling patterns of each frame comprise individual signalling data for each data pattern comprised in the frame, wherein said evaluation means is adapted totextract said individual signalling data for each data pattern , from a received signalling pattern.
19. Receiving apparatus according to one of the claims 11 to 18,
i
wherein said receiving means is adapted to be tuned to and to receive a selected part of said transmission bandwidth so that an optimized receipt of a signalling pattern in the selected part of said transmission bandwidth to be received is

101 enabled.
20. Receiving apparatus according to one of the claims 11 to 19,
wherein said receiving means is adapted to be tuned to and to receive a selected part of said transmission bandwidth so that said at least one data pattern to be received is centred in relation to the selected part of said transmission bandwidth to be received.
21. Receiving apparatus according to one of the claims 11 to 20,
I) wherein said receiving means is adapted to be tuned to and to receive a selected
If part of said transmission bandwidth on the basis of signalling data received in a
f" signalling pattern of a previous frame.
22. Receiving method for receiving signals transmitted in a multi carrier system on the basis of a frame structure in a transmission bandwidth, each frame comprising at least two signalling patterns adjacent to each other in the frequency direction each with signalling data mapped on frequency carriers and at least two data patterns with data mapped on frequency carriers, each of said at least two signalling patterns having the same length, comprising the steps of receiving a selected part of said transmission bandwidth, said selected part of said transmission bandwidth having at least the length of one of said signalling patterns and covering at least one data pattern to be received, and evaluating the signalling data comprised in a received signalling pattern in order to enable the receipt of said at least two data patterns.
23. System for transmitting and receiving signals, comprising a transmitting apparatus according to one of the claims 1 to 8, and a receiving apparatus according to one of the claims 11 to 21 adapted to receive said time domain transmission signal from said transmitting apparatus.
I:
I .
24. Method for transmitting and receiving signals, comprising a transmitting
method for transmitting signals in a multi carrier system on the basis of a frame
structure, each frame comprising at least two signalling patterns adjacent to
each other in the frequency direction and at least two data patterns, said

1|. 102
- rj
transmitting method comprising the steps of - .
mapping signalling data on frequency carriers of each of said at least two °
signalling patterns in a frame, each signalling pattern having the same length,
mapping data on frequency carriers of said at least two data patterns in a frame,
5 transforming said signalling patterns and said data patterns from the frequency
} domain into the time domain in order to generate a time domain transmission
signal, and. !J , '