Note: Descriptions are shown in the official language in which they were submitted.
"A DISTRIBUTED CONTROL COMMUNICATIONS SYSTEM"
The present invention relates to a local area
network communications system with distributed control.
BACKGROUND ART
A Local Area Network (LAN) Communications System is
a communications system that is able to interconnect a
number of communication devices, such as infoemation
storage, retrieval and processing devices, over a limited
geographic area. Current LANs have a typical linear
10 topology from 500 metres to approximately 10 kilometres,
compared to telecommunication data networks which can be
world wide. However, whereas the maximum data rates
achievable with telecommunication data networks are limited
by the public telephone network to a few thousand
bits/seconds, LANs can achieve data rates of several
megabits/second.
LANs are able to provide these high data rates by
using a transmission medium such as low attenuation, wide
bandwidth coaxial cable which can support the data
modulation techniques used for high speed data
transmission. As the area serviced by a LAN is quite
small, the additional cost of the wide bandwidtb cable,
compared to normal telecommunications cable, is
insignificant in relation to the amount of information that
can be transferred over the net work-.
Baseband transmission LANs utilize the wide
bandwidth capability of coaxial cable to support the
transmission of high speed (several megabits/second)
digital signals along the cable. Because pulse
transmission techniques are used, only one data path can be
accommodated on the cable, as a wide frequency bandwidth is
required to maintain the symmetry and timing of the data
pu]ses.
Baseband L~Ns employ a shared channel topology and
Time Division ~ultiplexing is used to allow each device
attached to the network the opportunity to access the cable
to transmit information to another device. To increase the
network reliability, Distributed Network Control is used;
however, central network synchronization is commonly
~2~
re~uired.
Broadband transmission LANs use the wide frequency
bandwidth available with coaxial cable to support a number
of radio frequency carriers, which are separately modulated
by the data to be transmitted over the network. The
sroadband LAN subdivides the frequency spectrum of thQ
coaxial cable into a number of discrete radio frequency
carriers. Each carrier frequency in a Broadband LAN can
provide a separate data channel.
The radio frequency carriers in a Broadband LAN are
sinusoidal signals and only require a narrow portion of the
available frequency spectrum. The narrow bandwidth
required by each carrier frequency allows many separate
data channels to coexist on the one coaxial cable. Data is
transferred across the network by modulating the carrier at
the data rate using standard radio frequency modulation
techniques (Amplitude, Frequency, Phase etc.).
Unlike the Baseband LAN where a single data channel
is used and each device Time Division Multiplexes its data
onto the cable, the devices connected to a Broadband LAN
have many data channels cn which they can transmit data.
To maintain data integrity the network access control of a
Broadband LAN requires the non-conflicting assignment of an
appropriate data channel to each device requiring network
access.
A disadvantage of such local area network
communications system having central network control is
that failure of the central network control will prevent
the operation of the whole communications system.
U.S. Patent No. 3,573,379 discloses a
communications system for a plurality of subscribers having
random access capability without the requirement Eor the
usual central exchange. However, the system requires a
master timing clock to which the subscribers are
connected. Thus, centralised control is still present and
failure of the clock will prevent proper operation of the
system.
It is an object of the present invention to
overcome, or substantially ameliorate, the above described
~z~
-- 3
disadvantage by providing a local area network
communications system having distributed network control.
In the present invention, the failure of a control
unit will only cause the associated device to become
inoperative, and the rest of the system continues to
operate. Furthermore, due to distributed network control,
a cut or break in the transmission medium need not render
the whole system inoperative. Devices which are still
connected to each other can communicate with each other
regardless of a break in the line outside the link between
these devices.
DISCLOSURE OF THE INVENTION
According to the present invention, there is
provided a distributed control communications system for
providing communications between a plurality of subscriber
devices over a limited geographical area, said system
comprising a transmission medium having a plurality of
communication channels; a plurality of access means each
connected to said transmission medium and providing a
terminal for connection to a respective subscriber device,
whereby each said access means is adapted to provide a
logical connection between its respective subscriber device
and said transmission medium~ each said access means having
a control unit, said control units communicating with one
another via a common one of said communication channels,
characterised in that network control is provided by each
control unit progressively in turn on a cyclical basis.
The present invention provides a self contained
communications system suitable for use with a large number
of information storage and retrieval devices which need to
be interconnected at random intervals for varying periods
of time for the purpose of transferring information.
In the preferred embodiment, the frequency spectrum
of a coaxial cable transmission medium is used to support
the slmultaneous transmission of a large number of radio
frequency carriers. ~ data channel consists of one or more
carrier frequen~ies~ Data is transmitted over the data
channel by modulating a channel carrier frequency at the
data rate by varying the Phase, Frequency and/or Amplitude
of the carrier radio signal energy in synchronism with the
data to be transmitted.
Typically, each access means is a Cable Access Unit
which provides the functional interface between the passive
coaxial cable transmission medium and a respective
communications device. The Cable Access Unit logically
resides between the device and the coaxial cable although
it may be physically incoporated in the device, or external
to the device, depending on the particular implementation.
The Cable Access Unit typically comprises a microprocessor
for its logic operation.
The System of the preferred embodiment has a
distributed control network with each Cable Access Unit
having a control unit managing all of the network functions
for its host device. The host device can only access the
network by requesting the Cable Access Unit to provide a
logical connection with another device in the network. The
Cable Access Unit uses an automatically sequenced network
access control function ~o communicate with the Cable
Access Unit attached to the requested device, and to select
the appropriate carrier frequencies for transmit and
receive. When the data channel has been established the
two host devices are able to commence data transfers with
each other.
Network access is performed by a Cable Access Unit
in response to a request from the host device. Once a data
channel has been established it is reserved for the
exclusive use of the two communicating devices. When the
data transfer is completed the Cable Access Units executes
a logical disconnection of the devices from the network.
The carrier frequencies associated with the data channel
are then available for use by othe~ devices attached to the
network.
Normal data communications functions such as device
to device protocol conversion, error checking and
correction, receive acknowledgments, retransmission
requests, code conversion and information type (voice or
data~ remain the responsibility of the attached subscriber
devices.
A device connected to the system only requires a
simple control sequence between itself and its Cable Access
Unit to establish a logical connection with another
device. If both devices have the same native protocol no
data transformation is required and error checking is
performed by the communication devices in the normal manner.
Eah Cable Access Unit can independently establish
a unique data channel between its host device and another
device attached to the network. In addition, the host
device can transfer control information, via the Cable
Access Units, to the re~uested device enabling data
transfer parameters such as transmission speed, protocol
and data formatting to be established between the two
devices prior to a data transEer.
The common channel is typically a separate fixed
carrier frequency data channel (System Control Channel) for
e~clusive use by the Cable Access ~nits atta~hed to the
network. Each Cable Access Unit preferably has two radio
transceivers. One transceiver is locked to the System
Control Channel and is set to the same carrier frequency in
all Cable Access Units. Any information transmitted on the
Control Channel is available to all Cable Access Units.
Host devices atta~hed to the network via the Cable Access
Units cannot access or transmit on the Control Channel.
The other transceiver is frequency agile, and can
be set by the Cable Access Unit to any of the predetermined
carrier frequencies provided by the System. The frequency
agile transceiver is the network data interface for the
host device and data transferred through this transceiver
is transparent to the Cable Access Unit.
~ ach Cable Access Unit is assigned a unique System
address, which is used as its identifier when communicating
with other Cable Access Units via the Control Channel.
As each Cable Access Unit is able to transmit on
the one single Control Channel, random transmissions would
cause data collisions and a collision detection and
recovery mechanism is preferably provided~ According to
this mechanism, the time at which each Cable Access Unit
can transmit information on the Control Channel is
-- 6
determined by a specific 'Transmission Authorisation' or TA
word (packet) of a predetermined number of bits. Each
Cable Access Unit monitors the TA packets on the Control
Channel until it recognises its own address. At this time
the addres~ed Cable Access Unit is the only device
'authorised' to transmit information on the Control
ChannelD A 'Network Access' or N~ word (packet) comprising
a number of bits is transmitted in response to the TA
packet, which is received by all other Cable Access Units
on the network.
When a Cable Access Unit receives a TA packet which
has the destination address set to its own source address
it becomes the highest priority Cable Access Unit on the
network. The Cable Access Unit asserts its priority by
transmitting its NA packet which notifies all other Cable
Access Units that it has control of the network~
At ~he completion of the NA packet transmission the
Cable Access Unit then transmits a TA packet with the
destination address set to the next sequential Cable Access
Unit. The transmission of this TA packet advises the other
Cable Access Units that it is relinquishing control of the
network. The Cable Access Unit with the next sequential
source address then asserts its priority by transmitting
its NA packet. In this manner, a rotating transmission
priority structure is established wherein each Cable Access
Unit is allowed access to the control channel on a cyclical
basis.
The NA packet transmitted by each Cable Access
Unit, in response to a TA packet, contains the System
network control information. I'he TA packets only schedule
the orderl~ transmission of this in~ormation onto the
Control Channel by each Cable Access Unit in the correct
sequence required by the network control architecture. It
is the sequencing of Cable Access ~nit transmissions on the
Control Channel which prevents multiple transmissions
occuring and avoids the need Eor data collision detection
and recovery.
When a device requires a logical connection with
another device attached to the System it commands its Cable
Access Unit to make the connection, provides the address of
the destination device, the type of transmission required
and the device dependent parameters, if required.
The source device Cable Access Unit reserves an
appropriate data channel corresponding to the type of
transmission required and transmits a 'connect request' to
the destination device Cable Access Unit via a NA packet on
the Control Channel. The destination device Cable Access
Unit responds with its own NA packet and acknowledges the
request or advises that it is busy. If the source device
receives an acknowledge from the destination device it sets
its ~requency agile transceiver to the carrier frequencies
that it previously reserved, and advises the host device
that transmission can commence. The destination device
Cable Access Unit also sets its frequency agile transceiver
to the same carrier frequencies, and advises its hos-t
device that a logical connection from another device on the
System has been made. Any device dependent parameters that
were received from the source device are passed
transparently to the destination device.
The type of interface provided between the host
device and its Cable Access Vnit can vary depending on the
capability of the host device, the complexity of the
network, and the type and format of the data to be
transferred over the network.
With regard to devices that cannot provide direct
communication with the Cable Access Unit, such as simple
CRT terminals, the operator provides the Cable Access Unit
with the appropriate access request information via an
3~ attached keypad. When a conneckion is made the status is
displayed to the operator and the serial interface control
signals IClear to Send, Data Set Ready, Carrier Detect
etc.) are set to reflect the status of the logical
connection.
More comple~ devices, such as word processing
terminals, can have the Cable Access Unit tightly coupled
to their own internal control architecture. ~ this case,
the -terminal control processor would interact directly with
the Cable Access Unit, as i~ it were another peripheral
-- 8 --
device attached to its internal system structure. Although
the access request would be initiated by the operator, the
logical connection would be established by a control
sequence between the Cable Access Unit and the terminal
control processor. In this type of implementation the
ability to transfer device dependent parameters between the
source and destination devices, via the NA packets, allows
non-System functions, such as the data format, transmission
speed and protocol to be established before the data
transfer commences.
With the distributed control architecture of the
System there is no central control node. The failure of
any Cable Access Unit on the network has no effect on the
ability of other Cable Access Units to use the network.
Each Cable Access Unit in the network is never
idle. It is continuously receiving NA packets from the
other active Cable Access Units on the network and
transmitting its own NA packet in response to a TA packet.
Each NA packet has a special field, known as the Reserved
Channel field, which is used by a Cable Access Unit to
reserve one or more carrier frequencies for its own use, to
the exclusion of all other Cable Access Units.
When a host device requests a network access, in
order to establish a data channel to another device on the
network, the Cable Access Unit commences a Network Access
Sequence. At the time of the access request from its host
device, the Cable Access Unit is transmitting'idle' NA
packets on the Control Channel, and is monitoring the
incoming NA packets for a connect request from another
Cable Access Unit.
Once the access re~uest is received from the host
device the Cable Access Unit goes 'busy' and waits for its
next TA packet. The next NA packet transmitted is an
'idle' and 'busy' packet which informs all other Cable
Access Units that is unavailable for a connection. After
transmitting this NA packet the Cable Access Unit monitors
the Reserved Channel field in each NA packet received from
the other Cable Access Units on the network. After one
complete Control Channel TA packet sequence rotation all
~2~t~
g
active Cable Access Units have transmitted their NA packet,
and all of the carrier frequencies that have been reserved
by other Cable Access Units are known to the Cable Access
Unit requiring network access.
The Cable Access Unit requiring network access is
the only one that is authorised to transmit at this time.
A Channel Number corresponding to an unused channel is
arbitarily selected by the Units logic and a 'connect
request' NA packet is transmitted on the Control Channel.
This NA packet is received by all other Cable Access Units
on the network and advises them that the carrier
frequencies corresponding to the Channel Number in the
Reserved Channel Field have been reserved and are
unavailable for allocation. As no other Cable Access Unit
can transmit at this time the NA packet from the access
requesting Cable Access Unit reserves its carrier
frequencies exclusively and without ambiguity.
Except for the Cable Access Unit whose Source
Address is contained in the NA packet Destination Address
Field, all other Cable Access Units only note the change in
network status, and that the carrier frequencies
corresponding to the Reserved Channel Field are reserved.
The addressed Cable Access Unit recognizes a valid access
to its host device by another device on the network and,
with a NA packet, responds to the requesting Cable Access
Unit after it receives its TA. The eequesting device Cable
Access Unit is advised of the current acti~ity status of
the requested device. If the requested device is 'busy'
the requesting device Cable Access Unit can either cancel
the requestt wait until the requested device is 'ready' or,
if it has a higher network access priority, force a connect
by exercising its priority status.
If the requested device is not 'busy' an
acknowledge is transmitted to the requesting device Cable
Access Unit in its next NA packet transmission. Both Cable
Access Units then set their frequency agile transceiver to
the carrier frequencies previously reserved and a logical
connection is established between the two host devices.
Each Cable Access Unit advises its own host device that the
- 10 -
data channel is established and the data transfer can
commence.
During the period of the logical connection the NA
packets transmitted by the Source and Destination Cable
Access Units maintain exclusive use of the reserved carrier
frequencies by transmitting the Channel Number in the
Reserved Channel Field. When the data transfer is
completed a disconnect NA packet sequence is exchanged
between the requesting and requested device Cable Access
Units validating the disconnect. The next NA packet
transmitted by both Cable Access Units has the Reserved
Channel Field set to zero which relinguishes the reserved
status of the carrier frequencies previously used Eor the
data channel. As these carrier frequencies are no longer
being used they are available for use by any other Cable
Access Unit on the network.
BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 is a schematic block diagram of a limited
area network communctions system according to a preferred
embodiment of the present invention.
Figure 2 is a schematic block diagram of a host
device connection to the cable of the system of Figure 1.
Figure 3 is a schematic block diagram of a Cable
Access Unit of the system of Figure 1.
Figure 4 is a schematic block diagram of a control
unit of the system of Figure 1.
Figure 5 shows a transmission authorization packet
format.
Figure 6 shows a network access packet format.
Figure 7 illustrates the time sequence Eor TA and
NA packet transmissions.
Figure 8 illustrates the time sequence for TA and
NA packet transmissions for a non-responding CAU.
Figure 9 illustrates the time sequence for TA and
NA packet transmissions during recovery from a
non responding CAU.
Figure 10 illustrates the time sequence for TA and
NA packet transmissions for two successive non-responding
CAUs.
-- 11 --
DESCRIPTION OF THE PREFERRED EMBODIMENT
-
A communication system in accordance with a
preferred embodiment is illustrated schematically in Figure
1. A number of host devices 130, 230, 330, 430..Øare
connected to a transmission medium such as a coaxial cable
100 via respective Cable Access Units 120, 220, 320,
420Ø..and line couplers 110, 210, 310, 410.~... Although
the transmission medium is typically a coaxial cable, it
can also be an~ other suitable medium such as a light fibre
or even a wireless link.
The host unit can be any form o information
retreaval, processing or storage device, such as video
terminals, data base processors, personal computers,
telemetry devices, or analog communication devices such as
telephones. The host devices are connected to their
respective Cable Access Units by a link appropriate to the
type of signalling taking place, but typically by twisted
pair wires.
Communication takes place over the coaxial cable
100 by means of radio frequency signalling. The bandwidth
of the coaxial cable is several 100 MHz (e.g. 300 M~Z is
accepted by the cable TV industry), hence a high number of
discrete radio channels can be supported by the cable, the
number of channels being a function of the information
bandwidth.
Communication is established between a pair of host
devices when one host, say 130, instructs its respective
Cable Access Unit 120 that a connection i5 required to
another host, say 330. Cable Access Units 120 and 320
communicate over a common channel, being one of the
plurality of channels provided on the cable, and known as a
'control channel' which, for a given network, is set to a
radio frequency known to all Cable Access Units. When
Cable Access Units 120 and 320 have established that
communication is possible (e~g. that host 330 is not
already communicating with some other host or i5
disconnected or inoperative), Cable Access Units 120 and
320 select a pair of radio frequency data channels for the
exclusive use of their hosts which then exchange
- 12
information as required.
After exchange of information is finished/ either
Cable Access Unit 120 or 320 is so advised and the duplex
pair of radio fre~uencies channels allocated to hosts 120
and 320 are released for use by other host devices.
Taking host 130 as an example, figure 2 illustrates
the connection of the host device to the cable via a Cable
Access Unit 120. The Cable Access Unit 120 may be
physically part of the host device 130 (e.g. if the host is
a telephone or personal computer) or may be a separate key
pad 125 for devices such as CRT terminals.
The Cable Access Unit 120 is connected to a line
coupler 110 by means of a 'drop' coaxial cable which is
relatively short (less than 500 metres) and has the same
electrical impedence as the main cable 100. Line coupler
110 provides a junction in the cable to allow an external
connection while providing the correct electrical
impendence to each coaxial cable terminal. Such line
couplers are in common use throughout the cable television
industry, and in various forms are known as 'line
couplers', 'splitters', 'directional couplers' and
'splitter-combiners'.
If other transmission media (e.g. optic fibers) are
used, appropriate couplers are provided to tap into the
medium.
The Cable Access Unit 120 is shown in more detail
in Figure 3~ the other Cable Access Units being similar in
architecture. The unit 120 has a control unit 121 which
comprises a microcomputer and associated logic and memory.
30 A transmitter 123 and receiver 124 are connected between
the cable 100 and the control unit 121, and form a fi~ed
frequency transceiver for communicating over the common
control channel. The transmitter 123 includes a modulator
for digital signalling and can modulate the carrier using
amplitude or angle techniques or a combination of these.
The receiver 12~ has a demodulator capable of demodulating
the transmitted control carrier and output digital data.
Transmitter 126 and receiver 127 form a frequency
selectable transceiver which can be tuned over a set of
discrete fre~uencies, the number of discrete ~requencies
being a function of the in~ormation bandwidth, and the
dynamic range o~ a frequency synthesizer 122 which is used
to tune the transceiver 126, 127.
The transmitter 126 also includes a modulator which
may accept analog or digital information depending on the
Cable Access Unit's function. The carrier signal can be
either amplitude or angle modulated, or a combination of
these. The receiver 127 demodulates the carrier signal and
provides the requisite analog or digital demodulated signal
as output to a user interface 129.
Any suitable known frequency synthesizer 122 can be
used. The control unit 121 provides a digital number to
the frequency synthesizer corresponding to a selected
frequency. The output of the frequency synthesizer (a low
level RF signal) is then used to tune transmitter 126 and
receiver 127. The input to transmitter 126 and output from
receiver 127 are the signals passing between the frequency
agile transceiver and the user interface 129. These
signals may be analog or digital. User interface 129
provides the correct electrical level etc., between the
user host device and the frequency agile transceiver~
User interface 128 provides the electrical
interface between the host device 130 and the control unit
121. This interface exchanges digital data relating to the
addresses of sending host device and receiving host device,
and other information required to establish a connection~
This interface may be driven by a simple key pad in the
case of a basic CRT terminal or both interface 128 and
Cable Access Unit itself can be embedded in the host device
120, as would be the case with a personal computer or data
base processor.
The control unit 121 can be implemented in several
ways including the use of random logic and gate arrays, but
the preferred embodiment uses a microcomputer and
associated memory and logic. The control unit 121 is shown
in more detail in the block diagram of Figure ~. The
control unit comprises a microprocessor 500 which can be
"single chip" device which includes on-chip Random Access
- 14 -
Memory, Read Only Memory, etc., or it may be the central
processing part only of the microcomputer. The
microprocessor 500, Random Access Memory 510 and Read Only
Memory 520 communicate via control bus 530. The frequency
synthesizer 122, together with its RF outputs 550 and 560
which drive the frequency agile transmitter 126 and
receiver 127, are also connected to the control bus 530.
Digital latches and associated electronic circuits 570 and
5~0 provide the interfaces between the host and the Cable
Access Unit, and staticizer-serializer 590 provides the
serial interface to the fixed frequency control channel
transmitter 123 and receiver 124.
Each Cable Access Unit is continuously monitoring
the control channel for control information known as
transmission authorisation (TA) packets and network access
(NA) packets.
Each TA packet has three fields which command a
particular Cable Access Unit to transmit its status and
network control in~ormation on the control channel. The ~A
packet format is shown in Figure 5 and comprises a
destination address 50 (address of the channel controller
to which the TA packet is being sent), a transmission
control 51 (defines the particular transmission function
required) and an error check 52 (calculated fro~ fields 50
and 51 and used to determine the TA packet validity).
Each NA packet has six specific fields which convey
the intention and status of the transmitting Cable Access
Unit to all other Cable Access Units of the network. The
NA packet format is shown in Figure 6 and comprises a
source address 61 (address of the CAU transmitting the NA
packet), an access rontrol 62 (defining the access mode and
the current status of the CAU transmitting the NA packet),
a destination address 63 (address of the CAU to which the
NA packet is being sent; an "idle" ~acket has this field
set to zero.), a reserve channel 64 (data channel number
which has been reserved by the source CAU for the exclusive
use of its host device and the destination address device;
if a data channel is not reserved this field is set to
zero), a device control ~5 (a free format field which
- 15 -
contains device to device control information; it is
received from the source address device and passed
transparently to the destination address device) and a
cyclic redundancy check 66 (calculated from the previous
five fields and used by each CAU to check the received ~A
packet validity).
A transmission priority structure is maintained by
the CAUs on the network whereby only one CAU at any one
time has the "authority to transmit" on the control
channel. When a CAU receives a TA packet which has the
destination address 50 set to its own source address, it
becomes the highest priority CAU on the network. The CAU
asserts its priority by transmitting its NA packet which
notifies all other CAUs that it has taken control of the
network. At the completion of the NA packet transmission,
the CAU then transmits a TA packet with the destination
address 50 set to the next sequential CAU source address.
The transmission of this TA packet advises the other CAUs
that it is relinquishing control of the network. The CAtU
with the next sequential source address then asserts ~r
priority by transmitting its NA packet. A rotating
transmission priority structure is established without the
requirement of a central control. Packet collision is
prevented since each CA~ can only transmit af~er it
receives a TA packet assigning the control channel to it to
the exclusion of all other CAUs for the duration of the NA
packet transmission.
In the time domain, the control channel can be
considered to be a ring structure around which circulate
the TA and NA packets generated by each CAU. Figure 7
illustrates, in the time domain, sequential TA and ~A
packet transmissions for a network having N CAUs. Each
packet is a fixed length. The time to transmit a TA packet
is I'ta and the time to tran~mit a NA packet is Tna.
Typically, for a control channel data rate of 128
kBits/sec, Tta is 375 microsecond for a 32 bit packet,
and Tna is 437.5 microsecond for a 56 bit packet. Each
packet transmit time is equal to the number of data bits
multiplied by the control channel data transmission bit
rate. ~n the described embodiment, all of these paremeters
are constants and each CAU can receive and decode every
packet on the control channel.
A second predetermined time Tw, the inter-packet
wait time, is the minimum separation time between
sequential packets. Typically Tw is of the order of 50
microsecond. During time T the control channel is
idle. When a CAU determines, from the last received TA
packet destination address that it is the current highest
priority CAU, it delays a time of Tw from the end of the
TA packet before commencing transmission of its own NA
packet. At the completion of the NA packet transmission,
the CAU waits time Tw and then transmits its TA packet.
CAUs which are not powered up, not physically
present or faulty, are not able to transmit their NA
packets when they are selected by TA packet. Figure 8
illustrates the TA and NA packet transmission in the time
domain for a non-responding CAU. CAU (M-2) transmits its
TA packet, with the destination address set to M-l, to
CAU (M-l). It then monitors the control channel for the
transmission of the corresponding NA packet from the CAU
(M-l). CAU (M-l) waits time Tw from the end of the TA
packet and then asserts its control of the network by
transmitting its NA packet. It waits other time Tw and
transmits its TA packet to CAU (M).
As CAU (M) is powered off it cannot respond to the
TA packet Erom CAU (M-l) and the control channel remains
idle. CAU (M-l) waits time Tw, the minimum time from the
completion of a TA packet that CAU (M) will commence
transmission of its NA packet. CAU (M-l) then waits an
additional time Tpd, which corresponds to the maximum
signal progation delay time for a cable segment.
Typically, Tpd is 4.5 microsecond/kilometre of coaxial
cable~ At the end of time Tw + Tpd from the
transmission of its TA packet, CAU (M-l) assumes that
CAU (M) is unable to transmit its NA packet. CAU (M-])
increments the destination address to M~l and transmits the
next TA packet to CAU (M-~l). CAU (M~l) is active and
asserts its control of the network by transmitting its NA
~2~
packet~ The recovery from the non-responsing CAU is shown
in the time domain in Figure 9O
If more than one sequential CAU is unable to
respond to its NA packet, the CAU which has control of the
network waits time TW -~ Tpd from the end of each
successive TA packet transmission and then transmits a TA
packet to the next sequential CAU. This sequence continues
until an active CAU responds with its NA packet and takes
control oE tbe network. The TA and NA packet sequence for
two sequential inactive CAUs is known in Figure 10.
Therefore, the CAUs can be powered on or off
without the requirement for synchronisation with the other
CAUs. The presence or absence of a NA packet in response
to the TA packet is sufficient to advise all other CAUs
attached to the network of the status and availability of
the respective CAU.
For the time delay figures quoted above, the total
period of a control transfer is typically 913 microsecond,
and the poll rate is therefore approximately 1095 dual
packets/second.
There are other packet formats which can be
implemented to provide a similar network access mechanism
to that described above. By increasing the complexity of
the packet control structure, an increase in network access
performance is achievable. The length of the NA packet can
be varied, depending on the information bein~ transferred
on the control channel. An idle CAU need only notify other
CAUs that it is active. This can be achieved by
transmitting a truncated NA packet which contain the source
address field, the access control field and the CRC field.
The same truncation of the NA packet can be used if a
device control field transfer to the re~uested device CAU
is not re~uired. In each case, the access control field
would specify the type of N~ packet being transmitted,
which would allow receiving CAUs to determine the start of
the CRC field for packet validation.
The establishment of a data channel between two
devices can be significantly shortened if the CAU
requesting a device access can reserve a number of
- 18
sequential time slots on the control channel. The
reserving of a number of time slots would enable the
reguesting and the requested device control units to
exchange sequential NA packets and synchronise the logical
connection without waiting for a complete control channel
time rotation for each NA packet transmission. This would
require the control channel transmission priority to be
passed, via the exchange of TA packets~ between the
requesting and requested CAUs enabling NA packets to be
exchanged beEore the control channel transmission priority
is passed to the next sequential CAU.
