Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
CA 02302501 2000-02-28
1
PCT/EP 98/05451
DATA TRANSMISSION SYSTEM WITH RELAY STATIONS BETWEEN A
SOURCE STATION AND A TARGET STATION
The invention relates to a data transmission system for the
digital transmission of data, including voice data.
A decentralised data transmission network with numerous dis-
tributed stations is known from DE 33 37 648 C2, in which
direct data communication takes place only between neighbou-
ring stations. The transmission path from a source station to
a target station is defined by a special routing system and
the data are then transmitted from station to station in both
directions on different channels. In this context, each of the
stations transmits on a single channel, which is used only for
linking exactly two stations. This, however, requires a corre-
spondingly adapted data rate.
Packet-oriented data transmission between the stations of a
data transmission network is also known from the Internet. In
this case, data are grouped in packets and these packets are
transmitted separately via the most favourable transmission
path in each case. This kind of packet transmission causes
considerable delays, which are at least equivalent to the time
required to transmit one packet. Due to the associated delays,
packet transmission of this kind is unfavourable for a tele-
phone system. The delays would accumulate in accordance with
the number of stations involved in transmission, particularly
in a decentralised data transmission network which uses sta-
tion-to-station transmission.
The object of the invention is to design a decentralised digi-
tal data transmission system which enables the use of diffe-
rent transmission channels between two stations and greatly
minimises delays at the same time.
CA 02302501 2000-02-28
2
According to the invention, the object is solved by the featu-
res specified in Patent Claim 1.
The data transmission system according to the invention is
characterised in that the signals received at each station are
converted symbol-by-symbol from the reception channels to at
least one different transmission channel. This means that each
symbol flow coming in on a reception channel is converted to
the transmission channels. This process is similar to forming
information packets consisting of a single symbol. In the
simplest case, a symbol is one bit. However, it can also con-
sist of a number of related bits, such as the eight bits re-
presenting a letter symbol. During a transmission, the number
of bits per symbol position is constant within a sub-channel.
The number of bits per symbol is defined at the beginning of
transmission, depending on the required or desired degree of
transmission quality. Symbol-by-symbol conversion means that
only a delay equivalent to one symbol position of the symbol
flow is required at each station. This delay is related to the
fact that the symbol string on the incoming channels and the
outgoing channels is normally not synchronised, meaning that
a certain waiting time is required before the outgoing signal
can be transmitted in synchrony with the transmission chan-
nels. This delay, however, is minimal. In practice, it amounts
to roughly one to two symbol positions. The delays of the
individual stations accumulate. As a result of the minimal
delay at each individual station, the resulting total delay of
the transmission path is still acceptable.
According to a preferred configuration of the invention, the
transmission channels are divided into sub-channels, each of
which is suitable for transmitting a symbol flow and where the
symbols of all sub-channels of a transmission channel are
transmitted synchronously. This means that each station can
receive incoming signals on all channels. The outgoing sub-
channels can be transmitted in concentrated fashion on a sing-
le channel or a few selected ones. The symbol positions of all
sub-channels are transmitted synchronously on each channel,
CA 02302501 2000-02-28
3
which consists of a pre-defined number of sub-channels. Each
sub-channel supports unidirectional data flow. The data arrive
at the station on each sub-channel of the entire channel sy-
stem in a continuous data flow, without being divided into
"frames" or "packets". Consequently, the data flow does not
require headers or any other defining elements. Rather, each
symbol of the data flow is converted to the sub-channel selec-
ted for transmission within a very short time after reception
and transmitted synchronously with the transmission channel.
The sub-channels are allocated to a transmission channel bet
ween two stations such that the transmission frequency range
is dynamically adapted to the information content to be trans
mitted. This means that the number of sub-channels per channel
is variable.
Preferably, allocation for transmitting envisaged channels
occurs at a station in such a way that all transmitting sub-
channels of this station are located within just a few chan-
nels. This considerably reduces the number of channels to be
used. In this context, it must be borne in mind that, if a
station is transmitting on a channel, this channel cannot be
used by neighbouring stations, in order to avoid interference
or other disturbances . Even if a station uses only one sub-
channel of a channel, the entire channel is reserved for this
station. Therefore, all connections which run through a speci-
fic station are preferably distributed over sub-channels all
contained in the same channel.
In a preferred configuration of the invention, appropriate
error correction bits are added to the contents of the syn-
chronously transmitted symbol positions of the sub-channels of
a channel, and error correction is carried out at the recei-
ving station, in order to reduce the probability of error in
the data link. Known methods can be used for error correction,
such as the FEC method (Forward Error Correction) or the ARQ
method (Automatic Re-transmission Request). The special featu-
re in the present case is that the contents of the synchro-
CA 02302501 2000-02-28
4
nously transmitted symbol positions of the sub-channels of a
channel are used for error correction, where completely inde-
pendent information contents flow in the sub-channels. This
means that error correction is carried out on the basis of
bits which are associated with different information and which
merely happen to be located at the synchronous positions of
the channel.
The use of an error correction method is only sensible if a
bit error ratio of less than roughly 10-3 is to be achieved.
Simple error detection is adequate for higher bit error ra-
tios, in order to at least obtain information on the quality
of the connection between the two participating stations.
Error detection of this kind can be provided by a redundant
error control element (e. g. parity bit), this error control
element being added to the synchronously transmitted symbol
positions of all sub-channels of a channel. It is alternative-
ly or additionally possible, after each transmission of a pre-
defined number of symbol positions of a sub-channel, to gene-
rate an error detection bit for each sub-channel, which corre-
sponds to the consecutive information contents of this sub-
channel, error detection being carried out at the receiving
station.
A practical example of the invention is described in more
detail below based on the drawings.
The drawings show the following:
Fig. 1 A diagram of part of the data transmission system
showing the distributed stations,
Fig. 2 An example of a connection from a source station to
a target station,
Fig. 3 A coupling matrix for frequency conversion at each
station,
CA 02302501 2000-02-28
Fig. 4 An example of data flows running through the coup-
ling matrix shown in Fig. 3, and
Fig. 5 A diagram of the consecutive symbol positions in a
5 channel with error correction bits and error detec
tion bits.
The data transmission system consists of numerous distributed
stations S, where each station represents a subscriber posi-
tion. Each station contains transmitting and receiving equip-
ment. Two frequency bands of 12.8 MHz each are available for
radio transmission of the data. The two frequency bands are
separated from one another by duplex spacing. One frequency
band is designated as the uplink and the other as the down-
link. In order to establish a connection, a channel in the
uplink is used for the connection in the one direction and a
channel in the downlink for the connection in the other direc-
tion, so that the two directions are completely decoupled from
one another in terms of frequency.
In this practical example, the two frequency bands of 12.8 MHz
bandwidth each are divided into a total of 1,280 channels with
a width of 20 kHz. Some of these channels are used as informa-
tion channels for establishing a connection and for other
purposes. Each station can receive on any of the available
channels and transmit on any of the available channels.
In Fig. 1, it is assumed that a connection is to be establis-
hed between a source station S61 and a target station 565.
This connection runs via stations S60 and 563, which act as
relay stations. In addition, a connection from station S62 to
Station S64 is relayed by station 560.
In the example of an established connection shown in Fig. 2,
data transmission from S61 takes place on channel C1, data
transmission from S60 to S63 on channel C25 and data trans-
mission from S63 to target station S65 on channel C12. Station
560, which is given special consideration in this example,
CA 02302501 2000-02-28
6
also transmits on channel C25 for stations S61 and 564.
Routing, i.e. selection of the stations through which the
connection is to be established, and selection of the channels
are handled by way of a dialogue conducted between the parti-
cipating stations. Routing (path-finding) and establishment of
the connection are not the object of the present invention.
Figure 3 shows a coupling matrix KM, which is contained in
each station. For reasons of simplicity, each symbol position
represented by a box is assumed to consist of one bit in this
practical example.
Each station contains a channel register CR-1...CR-n for every
channel C1.:.Cn. Channel register CR-1 contains eight informa-
tion symbol positions 1...8, where each of the symbol posi-
tions corresponds to one sub-channel SC. Thus, channel C1 is
divided into eight sub-channels 1...8. Each sub-channel has a
bandwidth of 20 kHz, where the frequencies of all sub-channels
1-8 are consecutive. A unidirectional data link can be estab-
lished over one sub-channel SC.
Figure 3 shows the time-slot patterns for sub-channels 4 and
5 of channel C1, in which symbols are transmitted to channel
register 1. Transmission takes place at a frequency of 20 kHz
in a continuous symbol flow.
The synchronously received symbols (in this case: bits) of the
sub-channels of a channel enter a receiving register ER1...ERn
and are transmitted from there to the respective channel regi-
ster CR-l...CR-n with a delay of two symbol durations. Channel
registers CR-1 ...CR-n are each associated with the columns of
the coupling matrix. The coupling matrix has n rows and m
columns, where each row and each column is assigned to a dif-
ferent sub-channel and a different frequency. The rows of
coupling matrix KM each correspond to one sub-channel or
transmitting frequency. Each channel has one channel register
CR-1...CR-n, which contains one symbol position for each sub-
CA 02302501 2000-02-28
7
channel 1...8. The symbol positions of all transmission-side
channel registers are associated with the rows of coupling
matrix KM. Each transmission-side channel register CR-l...CR-n
is assigned a transmission register SR1...SRn.
Coupling matrix KM is of integrated circuit design, where
corresponding control signals can connect the nodes to the
intersections of a row and a column. The associated node re-
mains connected during connection.
In the practical example shown, it is assumed that the infor-
mation received on sub-channel No. 1 of channel C1 is to be
relayed on sub-channel No. 2 of channel 25. There is a connec-
ted node KP at the associated intersection of the coupling
matrix, so that the bit located in position No. 1 of receiving
channel register CR-1 is transmitted to position No. 2 of
transmitting channel register CR-25 for channel C25.
In the same way, the signals received on sub-channel No. 4 of
channel C2 are transmitted to sub-channel No. 6 of channel C25
and sent out on this channel.
Figure 4 shows an example of the timing of signals received on
the sub-channels of channels C1, C2 and C3. At the associated
station, such as station S60 in Figs. 1 and 2, the signals
received there and intended for relaying are converted to
channel C25. For station S60, a dialogue with the neighbouring
stations previously determined that channel C25 is available
for data transmission.
As Fig. 2 shows for the selected practical example, station
S60 receives the data from station S61 on channel Cl which it
is supposed to relay to station 563. Consequently, these data
are converted to channel C25 at station S60. At station 563,
the same data are converted to another channel, such as C12,
and transmitted to target station 565.
In the selected example, station S60 considered here receives
CA 02302501 2000-02-28
8
signals from station S62 on channel C2. These signals are to
be relayed to station 564. Channel C25 is again selected for
this purpose. Finally, signals are also to be transmitted from
station S60 to station.S6l, for which purpose another sub-
s channel of channel C25 is selected. Everything station S60
transmits is on channel C25, but on different sub-channels.
Figure 4 shows the conversion of the data at station S60,
which were received by stations S61 and S62 on channels C1 and
C2. These data are converted to channel C25, but to different
sub-channels. In this context, the time axis is designated as
"t" in each case. The top line of Fig. 4 shows that the symbol
positions transmitted on channels C1, C2 and C3 are delayed
relative to one another by a maximum of the duration of one
symbol position. For this reason, the data are retained in
channel register CR-1...CR-n (Fig. 3) until the associated
symbol position has been received for all channels. Conversion
to the outgoing channels is then carried out simultaneously in
coupling matrix KM.
In addition to the symbol positions of sub-channels 1...8,
which transmit the information, three other bit positions have
been added to each channel for error correction bits A, B, C.
The contents of these additional bit positions are analysed in
receiving register ER1...ERn and used to correct errors in the
information bits received simultaneously on one channel. Only
the corrected information bits are entered in the correspon-
ding channel register CR-l...CR-n.
In transmission registers SRl...SRn, error detection bits A,
B, C are added to the eight information symbols of a channel,
before the entire bit volume is transmitted. These error de-
tection bits are generated by an error detection algorithm in
accordance with the contents of the information symbol posi-
tions. After receiving the entire signal, error correction is
performed in the same way using the algorithm.
- CA 02302501 2000-02-28
9
Figure 5 shows a channel's individual symbol positions in
relation to their timing, where the numbers 1...8 refer to the
information symbol positions and represent sub-channels. These
sub-channels have different frequencies. In Fig. 5, frequency
f increases from left to right with increasing ordinal number.
The three error correction bit positions A, B, C have been
added to the last sub-channel (channel "8").
In the practical example shown in Fig. 5, a total of eight
consecutive symbols in the channel is followed by an addi-
tional symbol position P, which contains a parity bit for each
sub-channel that also serves the purpose of error detection.
The addition of the error detection bits and the error correc-
tion bits, and the analysis of these bits on the basis of the
information content, are carried out separately for each
transmission link. These additional bits are not involved in
frequency conversion.
As an alternative to the practical example described above, in
which the assignment of the sub-channels to the frequencies is
fixed, the assignment of the sub-channels to the frequencies
can be changed after each symbol step. This makes it possible
to ensure that a disturbance cannot permanently interfere with
a sub-channel.
