 |
Description  |
|
|
BACKGROUND OF THE INVENTION
The present invention relates to multi-station packet communication in
which a single transmit station sends the same message to a plurality of
receive stations and, more particularly, to a packet repeat request signal
transmitting method for a network in which all the receive stations
connected to a radio link or a bus type wire link are capable of observing
information which is being transmitted over the link.
Multi-station communication in a radio link, a bus type wire link and other
networks has the outstanding capability for delivering information to all
the receive stations in the network by a single communication because all
the receive stations are capable of observing the same information.
However, the problem with multi-station communication is that when a
plurality of receive stations respond with repeat requests, or
retransmission requests, at the same time due to transmission errors, the
repeat request signals conflict with each other in the link. As a result,
repeat requests cannot be correctly sent back to a transmit station.
The occurrence of a conflict particular to a situation wherein a plurality
of receive stations share the same communication link is well known in the
art as a problem with multi-access.
Multi-access systems may generally be classified into two types, i.e., a
transmission right control system which sequentially assigns a
transmission right to all the receive stations, and a random access type
which in the event of a conflict detects it and retransmits information
after a suitable delay time. Although the transmission right control type
system is free from conflicts, it needs to exchange some information for
controlling the repeat request signal returning timings among receive
stations, resulting in a complicated control. An example of retransmission
protocol of this kind is disclosed in S. B. Calo and M. C. Easton "A
Broadcast Protocol for File Transfers to Multiple Sites", IEEE
Transactions of Communications, Vol. COM-29, No. 11, November 1981, pages
1701-1707. The technique disclosed in this paper is such that a transmit
station sends N packets and each transmit station sends an ACK frame back
to the transmit station to report whether it successfully received the N
packets without error (CYCLE 1) and, thereafter, the transmit station
retransmits packets based on the returned ACK frames (CYCLE 2). Such a
procedure is repeated until all the packets have been transmitted. This
ACK frame scheme, however, brings about another problem that a
disproportionate period of time is necessary for all the packets to be
successfully transmitted. Also, the time necessary for transmitting a
repeat request signal becomes considerably long if the number of receive
stations is great. In addition, since receive stations for which a
multi-station communication is meant are usually changed at each
communication, the control over the returning timings among the receive
stations becomes further complicated. The random access type system, on
the other hand, involves a considerable probability of conflict which
limits the efficiency, because repeat request signals in most cases are
generated at the same time by a plurality of receive stations.
SUMMARY OF THE INVENTION
It is therefore an object of the present invention to eliminate the
drawbacks inherent in the prior art systems and provide a simple and
efficient repeat request signal transmitting method for multi-station
packet communication.
It is another object of the present invention to provide a generally
improved repeat request signal transmitting method for multi-station
packet communication.
In one aspect of the present invention, in a repeat request signal
transmitting method for multi-station packet communication in which a same
message is transmitted from a single transmit station to a plurality of
receive stations, each of the receive stations checks more than one
received packets for error and sends back to the transmit station a repeat
request signal which comprises a signal having a particular frequency
associated with information on a packet in which an error is detected,
while the transmit stations senses energy of respective frequency
components of the signals which are sent from the receive stations and
determines packets to retransmit in correspondence with the frequencies
the sensed energy of which is greater than a predetermined value.
In another aspect of the present invention, in a repeat request signal
transmitting method for multi-station packet communication in which a
single transmit station transmits a same message to a plurality of receive
stations, each of the receive stations checks more than one received
packets for error and sends back to the transmit station a repeat request
signal which comprises a burst signal appearing in a particular time slot
which is associated with information on a packet in which an error is
detected, while the transmit station senses energy of each of the bursts
of the signals sent from the receive stations and determines packets to
retransmit in correspondence with the time slots of the bursts in which
energy greater than a predetermined value sensed.
In another aspect of the present invention, a repeat request signal
transmitting method for multi-station packet communication in which a
single transmit station transmits a same message to a plurality of receive
stations, each of the receive stations determines a number of packets
which are received with errors in a binary code and sends back to the
transmit station the binary code, which is obtained after reception of a
predetermined number of packets, each bit in a time slot which is
different from the other bits, while the transmit station senses energy of
bursts which are sent from the receive stations and appear in the
respective time slots to recognize a number of check packets which the
receive stations need for error correction and retransmits check packets
based on the recognized number of check packets.
The above and other objects, features and advantages of the present
invention will become more apparent from the following detailed
description taken with the accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram representative of a first embodiment of the
present invention;
FIG. 2 shows frequencies which are assigned to repeat request signals in
accordance with the embodiment of FIG. 1;
FIG. 3 is a block diagram showing a second embodiment of the present
invention;
FIG. 4 shows frequencies assigned to repeat request signals in accordance
with the embodiment of FIG. 3;
FIG. 5 is a diagram demonstrating the operation of the embodiment of FIG.
3;
FIG. 6 is a block diagram showing a third embodiment of the present
invention;
FIG. 7 shows burst positions of repeat request signals in accordance with
the embodiment of FIG. 6;
FIG. 8 is a block diagram showing a fourth embodiment of the present
invention;
FIG. 9 shows burst positions of repeat request signals in accordance with
the embodiment of FIG. 8;
FIG. 10 is a block diagram of a fifth embodiment of the present invention;
FIG. 11 shows signals sent back from receive stations and their positional
relationship in a communication link; and
FIG. 12 shows a relationship between packets and check packets.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
While the repeat request signal transmitting method for multi-station
packet communication of the present invention is susceptible of numerous
physical embodiments, depending upon the environment and requirements of
use, substantial numbers of the herein shown and described embodiments
have been made, tested and used, and all have performed in an eminently
sastisfactory manner.
FIRST AND SECOND EMBODIMENTS
The principle underlying a first and a second embodiments of the present
invention will be described first. Briefly, utilizing the fact that the
signals to be transmitted are repeat request signals, the method in
accordance with any of the first and second embodiments allows repeat
request information to be correctly delivered from a plurality of receive
stations to a transmit station even if the repeat request signals conflict
with each other due to simultaneous generation. Specifically, where
receive stations simply code packet numbers which are assigned to faulty
packets and return them to a transmit station, the transit station fails
to correctly recognize the faulty packet numbers particular to the
respective receive stations in the event of a conflict. In accordance with
the first and second embodiments, the same number of discrete frequencies
as the packets are prepared so that a sinusoidal wave having a particular
one of the frequencies which is associated with any faulty packet may be
sent back, where information on packets are implemented with packet
numbers. Some signal if not without interference will necessarily be
observed in a frequency associated with a packet number of a packet which
was received with an error by more that one receive stations, while no
signal will be observed in frequencies associated with packet numbers of
packets which were successfully received by all the receive stations.
Therefore, repeat request from all the receive stations will be fulfilled
if the transmit station is constructed to observe received energy in the
respective frequency bands and retransmit only those packets with packet
numbers associated with frequency bands in which energy greater than a
predetermined level has been observed.
The packet-by-packet repeat request signal transmission effected by the
receive stations as described above is not limitative. Alternatively, an
arrangement may be made such that the number of faulty packets received by
each receive station is sent back as information on faulty packets, while
a transmit station sends to the receive stations packets which correspond
to the number of faulty packets in the form of check bits, which are
produced by error-correction-coding data signals. In this case, each of
the receive stations discards faulty packets and decodes information using
retransmitted check packets, the transmit station therefore needing no
information showing which packets are faulty. This kind of transmitting
method, too, may match different numbers of faulty packets to different
frequencies in order to enhance efficient transmission of repeat request
signals as in the previously stated method.
The first embodiment will be described in detail referring to FIGS. 1 and
2. In this particular embodiment, assume that a relatively long message
such as a facsimile signal is transmitted in a plurality of discrete
packets.
In a transmit station 10, information coming in through an input terminal
102 is once stored in a buffer 104. A coder 106 adds error detection codes
to the information which is sequentially read out of the buffer 104. A
modem 108 modulates the output of the coder 106 and sends it through an
antenna 110.
Receivers 20A-20C receive the information at the same time. All the
receivers 20A-20C are identical in construction and, therefore, the
following description will concentrate to the receiver 20A by way of
example. As packets are received through an antenna 202 of the receiver
20A, they are demodulated by a modem 204 and, then, checked for error
packet by packet by an error detector 206. The received packets are stored
in a buffer 208. At the same time, results of decision showing whether the
discrete packets have been received without failure are sequentially
applied from the error detector 206 to a shift register 210. When all the
packets have been received, information on the respective packets
concerning errors have been loaded in the shift register 210. Assuming
that N packets are received, the shift register 210 has N shift stages. An
oscillator bank 212 has N oscillators which oscillate different
frequencies. The outputs of the oscillators in the bank 212 are
selectively gated by a gate circuit 214 based on the information stored in
the shift register 210. Specifically, only those oscillator output
frequencies which are associated with faulty received packets appear at an
output terminal of the gate circuit 214. The outputs of the gate 214 are
combined by a combining circuit 216 which comprises an analog adder. The
output of the combining circuit 216 is modulated by the modem 204 and,
then, transmitted through the antenna 202 as a repeat request signal.
Assume that four packets are received, N=4, and that an error has been
found in the first and third packets in the receive station 20A, in the
third packet in the receive station 20B, and in the fourth packet in the
receive station 20C. Shown in FIG. 2 are signals which are sent back as
repeat request signals from the respective receive stations 20A-20C and
their relationship in a link. In FIG. 2, frequencies f.sub.1, f.sub.2,
f.sub.3 and f.sub.4 are assigned to the repeat request signals which
respectively are associated with the first, second, third and fourth
packets. In this particular example, the receivers 20A-20C respectively
generate repeat request signals having frequencies which are shown in
R20a, R20b and R20c of FIG. 2.
The transmit station 10 receives a signal T10 shown in FIG. 2 through the
antenna 110. The modem 108 demodulates the received signal and applies its
output to a filter bank 112, which comprises N filters. In the example
shown in FIG. 2, the filter bank 112 comprises four filters which
respectively are designed to pass the four frequencies f.sub.1, f.sub.2,
f.sub.3 and f.sub.4. The outputs of the filter bank 112 are routed to a
detector bank 114. By envelope detection, the detector bank 114 detects
power of the respective filter outputs by envelope detectors and, thereby,
senses energy of each of the filter outputs. The outputs of the detector
bank 114 are applied to a decision bank 116 which then decides that, when
the output of any of the detectors is above a predetermined level, a
retransmission of a packet associated with it has been requested. In the
example of FIG. 2, a retransmission of the first, third and fourth packets
are requested. The results of such decision are stored in the shift
register 118. A gate circuit 122 gates outputs of a packet counter 120,
which are representative of addresses of the buffer 104, thereby causing
only the requested packets to be retransmitted. Then, each of the receive
stations 20A-20C checks only a necessary one or ones of the retransmitted
packets for error and, if found any faulty packet, then returns another
repeat request signal to the transmit station 10. In this manner, all the
information can be delivered without failure.
Referring to FIGS. 3 and 4, the second embodiment of the present invention
is shown. In a transmit station 30, information inputted through an input
terminal 302 is once stored in a buffer 304. A coder 306 adds error
detection codes to the information which is sequentially read out of the
buffer 304. A modem 308 modulates the output of the coder 306 and then
transmits its output through an antenna 310.
Receive stations 40A-40C receive the information from the transmit station
30 at the same time. Again, the receive stations 40A-40C are identical in
construction and, therefore, the operation will be described concentrating
to the receive station 40A only. Packets received through an antenna 402
are demodulated by a modem 404 and, then, checked for error by an error
detector 406. The received packets are stored in a buffer 408. The error
detector 406 produces a result of decision which shows whether each of the
packets has been received without failure, while a counter 408 counts
faulty packets. In this particular embodiment, after all the packets have
been received, the content of the counter 408, i.e., the number of faulty
packets is returned as information. Specifically, the output terminal of
the counter 408 is connected to a frequency control terminal of a
synthesizer 410 so that frequencies associated with the content of the
counter 408 are applied from the synthesizer 410 to the modem 404. The
modulated output of the modem 404 is sent to the transmit station 30.
Assume that the same errors as those in the first embodiment have been
detected in the respective receive stations 40A-40C. In FIG. 4,
frequencies f.sub.1, f.sub.2, f.sub.3 and f.sub.4 are assigned to repeat
request signals which are associated with one faulty packet, two faulty
packets, three faulty packets, and four faulty packets, respectively. In
the transmit station, the received signals are demodulated by the modem
308 and then applied to a filter bank 312, which comprises N filters. In
the example shown in FIG. 4, the filter bank 312 comprises four filters.
Each of the filters is designed to pass a particular one of the
frequencies f.sub.1, f.sub.2, f.sub.3 and f.sub.4 therethrough. The
outputs of the filter bank 312 are routed to a detector bank 314. The
detector bank 314 subjects the respective filter outputs to envelope
detection to extract their low frequency components and, thereby, detect
energy of each filter output. The outputs of the detector bank 314 are
applied to a decision bank 316. When the output of any of detectors in the
bank 314 has exceeded a predetermined level, the decision bank 316 decides
that packets corresponding in number to those particular detector outputs
are faulty. A maximum detector 318 to which an output of the decision bank
316 is applied functions to discriminate one of the detected frequencies
which is indicative of the largest number of packet errors. In the example
shown in FIG. 4, the maximum detector 318 detects the frequency f.sub.2 as
such a particular frequency and determines that the largest number of
faulty packets is two.
In the second embodiment, the transmit station 30 responsive to an output
of the maximum detector 318 applies error correction coding to the
information stored in the buffer 304 and transmits the redundant bits as
packets. The number of redundant packets corresponds to that of the faulty
packets. An example of such technique is disclosed in John J. Metzner "An
Improved Broadcast Retransmission Protocol", IEEE Transaction of
Communications, Vol. COM-32, No. 6, June 1984, pages 679-683. The
disclosed technique will be outlined with reference to FIG. 5.
In the example shown in FIG. 5, designated by symbols P.sub.1 -P.sub.4 are
packets which have already been transmitted. Each of the packets P.sub.1
-P.sub.4 includes check bits for error detection. An error correction
coder 320 (see FIG. 3) reads already transmitted data out of the buffer
304 and, then, codes them in a direction indicated by arrows in FIG. 5 to
provide error correction codes. In this condition, data stored in packets
P.sub.5 and P.sub.6 shown in FIG. 5 serve as check bits for the
information packets P.sub.1 -P.sub.4 which are associated in position
therewith. Each of the receive stations 40A-40C, therefore, is allowed to
correct faulty packets using the packets P.sub.5 and P.sub.6 which it
receives. The information which the transmit station need be supplied is
not which packet or packets have been faulty but how many packets have
been received with errors by the receive stations.
The coder 306 adds error correction codes to the output of the error
correction coder 320. The output of the coder 306 is modulated by the
modem 308 and then transmitted through the antenna 310. In the receive
station 40A, the modem 404 demodulates the retransmitted check bit packets
as ordinary packets, while the error detector 406 determines whether the
demodulated packets are faulty. If the retransmitted check bit packets are
faulty, each of the receive stations 40A-40C repeats the above procedure
for packet retransmission. Upon receipt of error-free check bit packets,
an error correction decoder 412 retrieves the previously received packets
from the buffer 408 and, after removing errors, applies them to a terminal
414.
As described above, the second embodiment, like the first embodiment,
detects repeat request signals using a method which relies on a filter
bank and detection of energy. Hence, it allows repeat request signals to
be correctly detected even if they are in a conflict. Another advantage
attainable with the second embodiment is that what is requied is not
sending a repeat request signal for each packet but simply sending
information indicative of the total number of failed packets, cutting down
the number of frequencies necessary for retransmission request.
While the first and second embodiments have been shown and described as
adding error detection codes to each packet, the error detection codes may
be replaced with error correction codes.
THIRD AND FOURTH EMBODIMENTS
The principle which underlies a third and a fourth embodiments will be
described first. Utilizing the fact that signals to be transmitted are
repeat request signals, each of the third and fourth embodiments makes it
possible for a plurality of receive stations to accurately deliver repeat
request information to a transmit station even if the repeat request
signals conflict each other. Where a number assigned to a faulty packet is
simply coded and transmitted, a transmit station cannot correctly
recognize packet numbers when the packet numbers are in a conflict. In
accordance with the third and fourth embodiments, time slots equal in
number to packets are set up so that burst signals may be sent from
receive stations to a transmit station in the time slots which correspond
to informations associated with faulty packets.
In a situation where packet numbers are adopted as informations associated
with packets, some burst signal will be observed in a time slot associated
with a particular number of a packet which has been received with an error
by more than one receive stations even if undergone interference, while no
signal will be observed in a time slot associated with a number of a
packet which has been correctly received by all the receive stations.
Hence, repeat requests from all the receive stations will be fulfilled if
a transmit station observes received energy in each time slot and
retransmits only those packets having numbers in which energy greater than
a predetermined level has been observed.
The packet-by-packet retransmission of repeat request signals effected by
each receive station as described above is not limitative. Alternatively,
an arrangement may be made such that while each receive station sends back
a repeat request signal associated with the total number of faulty packets
as information on packets which have been received with errors, a transmit
station retransmits packets which correspond in number to the faulty
packets in the form of check bits provided by error-correction-coding data
signals. In this case, each receive station discards faulty packets and
decodes information using retransmitted check packets, eliminating the
need for the transmit station to know which packets are faulty. Such will
realize efficient transmission of repeat request signals as in the
previously stated procedure, if numbers of faulty packets are matched to
time slots.
The third embodiment of the present invention will be described in detail.
Again, assume that a relatively long message such as a facsimile signal is
transmitted in a plurality of discrete packets.
In a transmitter 50, information applied to an input terminal 502 is once
stored in a buffer 504. A coder 506 adds error detection codes to the
information which is sequentially read out of the buffer 504. A modem 508
modulates an output of the coder 506 to transmit it through an antenna
510.
Receive stations 60A-60C receive the above information at the same time.
Since the receive stations 60A-60C are identical in construction, only the
operation of the receive station 50A will be described in detail. As
packets are received through an antenna 602, they are demodulated by a
modem 604 and then checked for error packet by packet by an error detector
606. The received packets are stored in a buffer 608. At the same time,
results of decision indicative of whether the respective packets have been
correctly received are applied sequentially from the error detector 606 to
a shift register 610. When all the packets have been received, information
on the packets concerning errors have been stored in the shift register
610. Assuming that N packets are received, the shift register 610 has N
shift stages. Based on the content of the shift register 610, the receive
station 60A generates a repeat request signal. Specifically, a gate
circuit 612 gates an output of an oscillator 614 responsive to the content
of the shift register 610. An example of outputs of the gate circuit 612
is shown in R60a of FIG. 7.
Assume that four packets have been received, i.e., N=4, and that an error
has been detected in the first and third packets in the receive station
60A, in the third packet in the receive station 60B, and in the fourth
packet in the receive station 60C. Shown in FIG. 7 are signals sent from
the receive stations 60A-60C as repeat request signals and their
positional relationship in a communication link. In FIG. 7, positions
b.sub.1, b.sub.2, b.sub.3 and b.sub.4 on the time axis are the burst
positions assigned to repeat request signals which are associated with the
first, second, third and fourth packets, respectively. In the above
embodiment, the receive stations 60A-60C send repeat request signals in
the time slots as shown in R60a-R60c of FIG. 7, respectively. The
intervals between burst positions are directed to absorbing delay
differences between the receive stations 60A-60C.
The transmit station 50 receives a signal T50 shown in FIG. 7 through an
antenna 510. The received signal is demodulated by the modem 508 and then
applied to a window circuit 512. The outputs of the window circuit 512 are
sequentially detected by a detector 514 and then checked by a decision
circuit 516 as to whether their energy is as high as a predetermined
level. The results of decision are sequentially loaded in a shift register
518. The received signal is shown in T50 of FIG. 7; it will be seen that
the first, third and fourth packets are faulty. In the transmitter 50, a
gate circuit 522 gates outputs of a packet counter 520 responsive to the
content of the shift register 518, thereby retransmitting only the
requested packets. Meanwhile, each receive station checks only associated
ones of the retransmitted packets and, if any one of them is faulty, sends
another repeat request signal to the transmit station 50. In this manner,
all the informations are transmitted from the transmit station 50 to the
receive stations 60A-60C.
Referring to FIGS. 8 and 9, the fourth embodiment of the present invention
is shown. In a transmitter 70, information entered through an input
terminal 702 is once stored in a buffer 704. The information is
sequentially read out of the buffer 704, while a modem 708 adds error
detection codes to the information. The output of the modem 708 is
transmitted through an antenna 710.
Receivers 80A-80C receive the above information at the same time. Since the
receivers 80A-80C are constructed in exactly the same manner, the
operation will be described taking the receiver 80A for example. Packets
coming in through an antenna 802 are demodulated by a modem 804 and then
checked for errors packet by packet by an error detector 806. The received
packets are stored in a buffer 808. At the same time, results of decision
showing whether the respective packets have been correctly received are
applied from the error detector 806 to a counter 810, which then couns
faulty packets. In this particular embodiment, after all the packets have
been received, the content of the counter 810, i.e., the number of faulty
packets is returned to the transmit station 70 as information. A gate
circuit 814 gates an output of an oscillator 812 only in those time slots
which are each associated with the content of the counter 810. The outputs
of the gate circuit 814 are modulated and then sent back to the
transmitter 70. Assuming that the same errors as in the third embodiment
have been developed in the respective receivers 80A-80C, repeat request
signals will be sent from the receivers as shown in FIG. 9. In FIG. 9, the
time slots b.sub.1, b.sub.2, b.sub.3 and b.sub.4 represent respectively
the burst positions which are assigned to repeat request signals
associated with one faulty packet, two faulty packets, three faulty
packets and four faulty packets.
The signals received by the transmitter 70 are demodulated by the modem 708
and then applied to a window circuit 712 which is adapted to produce
signals appearing at the burst positions b.sub.1, b.sub.2, b.sub.3 and
b.sub.4 only. The outputs of the window circuit 712 are sequentially
detected by a detector 714 whose outputs are applied to a decision circuit
716. This circuit 716 determines if the energy of each detector output is
as high as a predetermined level. The results of decision are stored in a
maximum detector 718 which is adapted to store a particular burst position
where a burst signal above a predetermined level has appeared last. In
this manner, the largest number of packets which were erroneously received
by the receivers is reported to the transmitter 70. The transmitter 70
retransmits packets which correspond in number to the largest packet
number after applying error correction coding to the content of the buffer
704. Specifically, an error correction coder 720 responsive to the output
of the maximum detector 718 manipulates the information stored in the
buffer 704 for error correction coding and sends the redundant bits as
packets. The number of the redundant packets corresponds to the faulty
packets which were received by the receivers 80A-80C. As previously stated
in relation to the second embodiment, an example of such a technique is
described in John J. Metzner "An improved Broadcast Retransmission
Protocol", IEEE Transaction of Communications, Vol. COM-32, No. 6, June
1984, pages 679-683. The output of the error correction coder 720 is
applied to the coder 706 which then adds error detection codes to the
input. The output of the coder 706 in turn is modulated by the modem 708
to be transmitted. In the receiver 80A, the check bit packets from the
transmitter 70, like ordinary packets, are demodulated by the modem 804
and then checked for error by the error detector 806. If any of the
retransmitted check bit packets is faulty, each receiver 80A-80C repeats
the above procedure to send another repeat requst to the transmitter 70.
If the check bit packets are error-free, an error correction decoder 816
retrieves the previously received packets from the buffer 808 and, after
readout error correction, applies them to a terminal 818. The procedure
described so far is repeated to retransmit packets.
FIFTH EMBODIMENT
Referring to FIGS. 10-12, a fifth embodiment of the present invention is
shown. In a transmit station 90, information applied to an input terminal
902 is stored in a buffer 904. A coder 906 adds error detection codes
packet by packet to the information which is sequentially read out of the
buffer 904. A modem 908 modulates an output of the coder 906 to transmit
it through an antenna 910 to each of receive stations 100A-100C.
Each of the receive stations 100A-100C receives the signal from the
transmit station 90 through an antenna 1002. The signal is demodulated by
a modem 1004, then checked by an error detector 1006 for error packet by
packet, and then stored in a buffer 1008 on a packet basis. At the same
time, resuls of decision made by the error detector 100 as to whether the
respective packets have been correctly received are applied to a counter
1010. Specifically, the counter 1010 counts faulty packets using binary
codes. The content of the counter 1010 is fed to a gate circuit 1012. Also
fed to the gate circuit 1012 is an output of an oscillator 1014. In this
condition, the gate circuit 1012 is abled and disabled depending upon
information of each bit associated with the content of the counter 1010;
when abled, the gate circuit 1012 delivers the oscillator output to the
modem 1004. That is, the gate circuit 1012 functions to control the output
of the oscillator 1014 responsive to an output of the counter 1010 so that
information associated with the content of the counter 1010 is sent to the
transmit station 90. In this particular embodiment, the transmission of
such information from the receive station 100A is effected after the
reception of all the predetermined number of packets. This allows the
number of faulty packets which are held in a binary code in the counter
1010 to be sent as information to the transmit station 90. This signal
returned to the station 90 is the repeat request signal. The reference
numeral 1016 designates an error correction decoder installed in the
receiver station 100A.
The transmit station 90 includes a window circuit 912, a detector 914, a
decision circuit 916, a shift register 918, and an error correction coder
920. As will be described later in detail, these circuits 912, 914, 916,
918 and 920 serve to process repeat request signals from the stations
100A, 100B and 100C to retransmit necessary check packets.
Assume that eight packets have been transmitted from the transmit station
90 to the receive stations 100A-100C, and that two packets have been found
faulty at the receive station 100A, one packet has been found faulty at
the receive station 100B, and three packets have been found faulty at the
receive station 100C. Shown in FIG. 11 are repeat request signals
generated by the respective receive stations under the above condition and
their positional relationship in a communication link. In FIG. 11, a
symbol 110A indicates bursts, or repeat request signals, while a symbol
110B indicates conflicts of the bursts each occurring in the same time
slot. Also, a symbol b.sub.1 designates a least significant bit (LSB), a
symbol b.sub.2 a second bit from the last, and b.sub.3 a most significant
bit (MSB). These bits each represent a particular time slot in which a
burst constituting a repeat request signal in a binary code is generated.
The time intervals between the nearby bursts 110A are directed to
absorbing delay differences between the receive stations 100A, 100B and
100C. In this particular example, a conflict 110B of the bursts from the
receive stations 100B and 100C occurs at the position b.sub.1, while a
conflict 110B of the bursts from the receive stations 100A and 100C occurs
at the position b.sub.2. At the position b.sub.3, no signal and,
therefore, no conflict occurs. The repeat request signals from the receive
stations 100A-100C are each generated by the previously stated counter
1010, gate circuit 1012, oscillator 1014 and modem 1004.
The bursts 110A or their conflicts 110B are received by the transmit
station 90 through the antenna 910. The received signals are demodulated
by the modem 908 and then applied to the window circuit 912 which
functions | | |
