 |
Description  |
|
|
TECHNICAL FIELD
The present invention relates to a wideband communication network using
radio either on a stand-alone basis or to supplement a hard-wired network
where complete portability of office design is desired.
DESCRIPTION OF THE PRIOR ART
Local Area Networks (LANs) have included many different architectures such
as the bus, loop, ring, star, tree, etc. One such LAN is disclosed in the
article "A New Local Area Network Architecture Using A Centralized Bus" by
A. Acampora et al. in IEEE Communications Magazine, Vol. 22, No. 8, August
1984, at pages 12-21. There, a centralized bus is used with all user
devices being hard-wired to a central node as shown in FIGS. 1-3 of the
article.
Indoor wireless communications networks have also been developed over the
years. In the article "Cordless Telephone System" by M. Komura et al.,
published in the Japanese Telecommunications Review, Vol. 15, No. 4, 1973,
at pages 257-261, a cordless radio telephone system is disclosed which
permits telephones to communicate via radio to a localized antenna which
is directly connected to a base station. Another technique for wireless
indoor communication is disclosed by F. Gfeller in the IBM Technical
Disclosure Bulletin, Vol. 24, No. 8, January 1982, at pages 4043-4046
wherein an infrared microbroadcasting network for in-house data
communication is disclosed. There, a host controller is directly connected
to a plurality of spaced-apart transponders for transmitting 2-way
communications via infrared signals with the various stations forming the
in-house system.
More recently, an office information network was disclosed in Globecom '85,
Vol. 1, Dec. 2-5, 1985, New Orleans, La. at pages 15.2.1-15.2.6 wherein a
slotted-ring access protocol and a dynamic bandwidth allocation scheme
offering preferential service to high-priority traffic is provided. There,
a dual optical fiber ring, transmitting in opposite directions, propagates
the communication signals to various nodes along the fibers. Connections
between the network nodes and local facilities or servers are copper pairs
or, where appropriate, wireless drops.
Indoor radio communication is not without problems, however. Buildings in
general, and office buildings in particular, present a harsh environment
for high-speed radio transmission because of numerous reflections from
stationary objects such as walls, furniture, and movable objects such as
people. The link between a given pair of transmitters and receivers is
thereby corrupted by severe multipath distortion arising from the random
superimposition of all reflected rays, and by shadow fading caused by the
absence of line-of-sight paths. At low data rates, the effects of
multipath can be characterized by Raleigh fading, while at higher rates
the channel additionally exhibits dispersion over the communication band.
Shadow fading is spectrally flat and characterized by a log-normal
distribution.
It is to be understood that all effects vary dynamically with time as the
environment slowly changes. Raleigh fading produces a wide variation in
the level of signals arriving at a particular receiver from different
transmitters, thereby precluding the use of standard techniques for
multiple access of the radio channel. Dispersion within the channel
produces serious intersymbol interference, thereby limiting the maximum
data rate of the channel and causing a fraction of users to experience an
unacceptably high bit error rate, and a link experiencing such condition
is said to have suffered an outage and is temporarily unavailable.
Therefore, the problem in the prior art is to provide a technique or
network which will permit as high a data rate as possible while
encountering changing conditions.
SUMMARY OF THE INVENTION
The foregoing problem in the prior art has been solved in accordance with
the present invention which relates to a wireless network for wideband
indoor communications using radio as the transmission medium either on a
stand-alone basis or to supplement a hard-wired network. The present
exemplary wideband indoor packet communications network comprises (1) a
plurality of transceivers; and (2) a central node. In addition, one or
more concentrators associated with certain separate groups of wireless,
and possibly hard-wired, transceivers, may be present. Each transceiver is
associated with a separate user of the network, and, some or all of the
plurality of transceivers communicate wirelessly with associated interface
units in the concentrators or central node. The central node comprises (a)
means for determining and communicating the necessary transmission
requireements to each of the active plurality of wireless transceivers
during a first subperiod of each frame period, and (b) means for receiving
packets of information from each of the plurality of transceivers,
transmitted as instructed by the communicating means, and retransmitting
the packets to the transceivers of the destined users during a second
subperiod of the frame period. More particularly, the communicating means
of the central node (i) determines the packet transmission requirements
and any transmission impairments of each network user when communicating
with the associated interface unit in a concentrator or central node, and
(ii) causes the transceivers associated with users determined to have
packet transmission requirements to transmit their packets of information
with a length sufficient to overcome the determined transmission
impairment.
It is also an aspect of the present invention to provide a wideband indoor
communications network as described above where (1) diverstiy antennas can
be used at the concentrators and central node, and one or more antennas
can be used at each transceiver, and (2) access to the radio channel used
by all wireless transceivers is performed by a modified polling scheme
which permits resource sharing to provide added protection against channel
impairments on an asneeded basis.
Other and further aspects of the present invention will become apparent
during the course of the following description and by reference to the
accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram of an exemplary arrangement of a wideband
communication network in accordance with the present invention including
various hard-wired and wireless user connections; and
FIG. 2 is a diagram of a media access technique using polling that can be
employed in the network of FIG. 1.
DETAILED DESCRIPTION
FIG. 1 illustrates an exemplary system topology which is functionally that
of a star Local Area Network (LAN) comprising a central node 30, remote
concentrators 20 and 21, and a plurality of user devices 10-19. Each user
device 10-19 is associated with a separate user of the network and can
communicate with central node 30 (1) via a hard-wired connection 26, as
shown for the indirect connections between user devices 10-13 and
concentrators 20 and 21; or (2) via a wireless link as shown for (a) the
channel comprising links 27 between a subgroup of user devices 18-19 and
central node 30, or (b) the indirect channel comprising links 28 between a
subgroup of users 14-15 and a subgroup of users 16-17 and concentrators 20
and 21, respectively. It is to be understood that user devices 10-19 can
each be coupled to a separate user terminal (not shown) such as, for
example, a data device, printer, personal computer, host computer,
telephone, etc.
Each of remote concentrators 20 and 21 is positioned between the associated
subgroup of user terminals 10-17 and central node 30 and is shown as
including (a) user interface modules (UIM) 22 and 23 which are each
coupled to a separate portion of the associated subgroup of user devices
10-17 (b) a clock module 24, and (c) a trunk module 25. It is to be
understood that each of exemplary concentrators 20 and 21 includes only
two UIMs, for purposes of simplicity and that additional UIMs could be
disposed in parallel with UIMs 22-23 shown, and connected to other
portions of the associated user devices (not shown) via either separate
hard-wired or wireless connections.
Each UIM 22 or 23 functions to translate the protocol of the signal
received from the associated user devices to a standard protocol of the
network as used by central node 30. The translated signal is then
transmitted, at the appropriate time, to trunk module 25 on a time
division multiplex (TDM) basis via a concentrator bus 29.sub.a for
transmission to central node 30, and vice versa for the other direction of
two-way communications using concentrator bus 29.sub.b. Where a user
device already transmits and receives signals using the standard network
protocol, an associated UIM need only transmit the received signal at the
appropriate time based on the received clock signals from clock module 24.
The trunk module 25 in each of remote concentrators 20 and 21 functions to
transmit each of the signals associated with that concentrator between
each of the UIMs 22 and 23 and central node 30 at the appropriate times.
The clock modules 24 provide the timing signals for each of the UIMs 22
and 23, and trunk module 25 to achieve coordinated operation within the
associated remote concentrator 20 or 21. Central node 30 is shown as
including a clock module 31 for providing the clock signals used in
central node 30; network interface units (NIU) 32-34 which are each
coupled either to a separate one of remote concentrators 20 or 21 or to a
separate subgroup of one or more user devices; a call processor 35; and
buses 36 and 37.
To describe the operation of the present network, the network components
associated only with hard-wired user devices, e.g., user devices 10-13,
UIMs 22 and NIUs 32 and 34 will first be considered. Each hard-wired user
device 10-13 is shown connected to the network via terminal interface
wires 26 and a UIM 22. Continuous (voice) or bursty (data) traffic
arriving at UIM 22 in concentrator 20 from user devices 10-11, or at UIM
22 in concentrator 21 from user devices 12-13, are formed into fixed
length packets for time-multiplexed high speed transmission to central
node 30 via trunk module 25. Each such packet is provided therein with a
logical channel number which allows central node 30 to re-route the packet
to the appropriate concentrator 20 or 21 where the indicated destination
user's device is connected. Central node 30 includes a contention bus 36,
37 operating at the speed of each high speed link, to accomplish this
re-routing. All traffic, including that traffic arising at a particular
concentrator 20 or 21 and destined for that same concentrator, is routed
through central node 30.
The receiving concentrator demultiplexes all arriving packets from central
node 30 for distribution via bus 29.sub.b to the appropriate UIM and
transmission to the destined user device. Logical channel numbers are
preferably assigned for the entire network at the beginning of a
predetermined time period of communications by call processor 35 in
central node 30. Additional device configurations and operational details
are described in the article "A New Local Area Network Architecture Using
A Centralized Bus" by Acampora et al. in IEEE Communications Magazine,
Vol. 22, No. 8, August 1984, at pages 12-21.
Radio links may be introduced, as shown in FIG. 1, via either a wireless
link between a UIM 23 in either one of concentrators 20 or 21 as shown for
link 28, or a wireless link directly to a NIU 33 in central node 30 as
shown for link 27. For link 27, the high-speed links from trunk modules 25
in concentrators 20 and 21 to central node 30 have been augmented by the
inclusion of an NIU 33 in central node 30 which becomes a radio base
station providing a high-speed channel to collect traffic from a subgroup
of radio user devices 18-19 located throughout the building. It is to be
understood that the term channel hereinafter implies full duplex
operation, with separate bands used to transmit to and receive from NIU
33. This radio channel operates at a rate less than or equal to that of
the central node's contention buses 36 and 37 and each of the high-speed
links between trunk modules 25 and NIU's 32 and 34. With an appropriate
access protocol, the radio channel may be shared among all radio users
18-19 and appear, to central node 30, as a virtual concentrator. Fixed
length packets arriving over links 27 contend for the nodal bus 36 along
with packets arriving via high-speed buses at NIUs 32 and 34 from trunk
modules 25. The packets arriving from the wired links 26 may be rerouted
by central node 30 to a radio link 27, and vice-versa, according to a
destination address included in each packet.
A wireless link 28 establishes a communication path from each user of a
subgroup of users, 14-15 or 16-17, to an associated UIM 23 in one of
remote concentrators 20 or 21. Although multiple paths are established
within a subgroup of users associated with a UIM 23 or NIU 33, these links
time-share a single radio channel. More particularly, at any moment, only
one radio user of a subgroup of users may access the radio channel. It
should be noted that there is no need to provide an aggregate data rate
over all radio links 27 or 28 in excess of the transmission speed of
central node 30 since all packets must be routed through central node 30.
Therefore, it is pointless to reuse the radio channel among user
subgroups, as this increased capacity could not be used. Thus, by sharing
a single channel, operating at the speed of central node 30, among all
radio users, each user can potentially access the full system bandwidth,
and interference among clusters caused by simultaneous use of the channel
by users in different clusters is avoided. From the perspective of central
node 30, a radio link 28 established from each concentrator 20 or 21 to
each of its subgroups of radio users appears as another wired port (UIM
22) on the concentrator.
Regarding the radio or wireless links only, each of the UIMs 23 or NIU 33
are preferably equipped with multiple antennas for diversity to protect
against multipath fading, and each user device 14-19 is preferably
equipped with only a single antenna, although multiple antennas could be
used. The combination of limited diversity at the concentrators 20 and 21,
and central node 30, along with resource sharing can be used to provide
arbitrarily high availability. No direct communication is permitted among
users, since all users may communicate only with concentrators 20 or 21 or
central node 30. It should be understood that common media access
techniques, such as Carrier Sense Multiple Access (CSMA), are
inappropriate in the radio environment because free space path loss and
multipath fading result in too large a variation of signal strength to
insure that all channel usage can be detected. To keep the wireless user
devices 14-19 inexperience, sophisticated timing requirements should be
avoided. Finally, because of problems with delay spread, it is desired
that the throughput of the system not be significantly reduced by a media
access technique, and separate receive and transmit channels must be
provided to allow full duplex operation.
For the present network shown in FIG. 1, an exemplary modified polling
technique is used, with central node 30 controlling the transmit token.
Polling is performed by call processor 35 in central node 30; with the
radio UIMs 23, located at concentrators 20 and 21, and NIU 33, located at
central node 30, being slaved to processor 35 such that, at any point in
time, only one UIM 23 or NIU 33 is allowed to transmit the token to its
community of UDs. It should be understood that all of radio UDs 14-19 time
share a single radio channel without frequency reuse.
The present exemplary polling technique for use with the radio channel
associated with the wireless UDs 14-19 is shown in FIG. 2. There, time is
divided into a sequence of fixed length intervals called frames, as shown
at the top of FIG. 2. At the start of each frame a polling interval 40
appears, followed by multiple intervals for transmission of continuous
(voice) traffic packets 41, and bursty (data) traffic packets 42. The
length of the continuous traffic intervals 41 depends on the amount of
continuous traffic. This continuous traffic is transmitted periodically,
at least once per frame period, with the time interval between continuous
traffic intervals used for bursty traffic. Transmission of one fixed
length packet per continuous traffic interval constitutes some standard
grade service, e.g., 64 kbps. Continuous traffic UDs may request multiples
of this basic rate by accessing multiple time slots per continuous traffic
interval. The polling sequence is shown at the bottom two lines of FIG. 2
for transmissions from and to central node 30.
The following steps forming the exemplary overall transmission sequence for
the radio channel are:
1. Via the UIMs 23 located at concentration 20 and 21, and NIU 33, call
processor 35 at central node 30 sequentially polls each UD associated with
the radio channel using sub-packets P1-PN.
2. When polled, UDs 14-19 sequentially respond, after being polled, using
the associated one of packets R1-RN as to whether the UD has continuous or
bursty traffic, and, if bursty traffic, the number of blocks of data.
3. Processor 35 then sequentially sends a signal, i.e., transmit token,
TS.sub.I -TS.sub.J, to each continuous traffic user to send one fixed
length packet, designated V.sub.I to V.sub.J, including a preset number of
data symbols in each packet.
4. Processor 35 then sequentially sends a signal, designated TS.sub.K
-TS.sub.L, i.e., a transmit token, to each bursty traffic user to send
their first data block designated packets D.sub.K -D.sub.L, then the
second data block, etc.
5. During steps 3 and 4, while the UDs are transmitting to concentrators 20
and 21 in blocks V.sub.I -V.sub.J and D.sub.K -D.sub.L, processor 35,
through the UIMs 23 at the concentrators, is transmitting voice and data
to the UDs 14-17 in associated blocks 44.
6. When it is time again for continuous traffic to be transmitted, then
step 3 is reiterated.
7. When it is time again for polling, i.e., the beginning of another frame,
then step 1 is reiterated.
The above described polling technique meets necessary requirements since
(a) the system handles continuous traffic, i.e., periodic data or voice,
with priority, (b) the system has the same maximum data rate for each use,
i.e., a fair distribution of resources, which depends on the system
loading, (c) there is no timing requirements at the remote UDs 14-19, (d)
the throughput on the channel is not significantly reduced by this
technique because the polling has a low duty cycle, mainly due to the
short propagation delays between the concentrators 20, 21 and the remote
UDs 14-19, and (e) the system has duplex operation.
What must also be considered is that in a multipath environment, paths of
different lengths cause delay spread at a receiver. The delay spread,
i.e., the dispersion or frequency selective fading in the channel,
produces intersymbol interference which limits the maximum data rate in a
given building and depends primarily on the rms delay spread and not the
delay spread function. Thus, within the coverage area, there is some
probability that the received signal bit error rate (BER) for each UD is
more than the required value, hereinafter called the outage probability.
If one UD 14-19 does not work in one location, the user can move the UD or
its antenna. However, the delay spread may vary slowly with time as people
and objects move within the building. Therefore, it is desirable to keep
the outage probability due to delay spread as low as possible so that the
wireless system is almost as reliable as any wired portion of the system.
In addition to the technique described above, resource sharing can be used
to increase the maximum data rate and/or decrease the outage probability.
With resource sharing, users normally transmit at some high rate R.sub.1.
When channel conditions between concentrators 20 or 21, or central node
30, and a particular UD no longer permits operation at this high rate.,
the rate is lowered to some value such as R.sub.2 such that the BER
objective is maintained. Such techniques are well known in the satellite
system art as disclosed, for example, in the articles by A. S. Acampora in
BSTJ, Vol. 58, No. 9, November 1979, at pages 2097-2111; and IEEE Journal
On Selected Areas In Communications, Vol. SAC-1, Jan. 1983, at pages
133-142 where a pool of spare time slots are used, and each packet is
transmitted with or without coding, to reduce the outage probability.
Although it takes longer to complete transmission at this lower rate, the
number of users simultaneously slowed down is usually a small fraction of
the total population, and the overall throughput remains high. More
particularly, during non-fade conditions, convolutional codes with a large
channel signaling alphabet are employed to permit a high rate of
information transfer as described hereinbefore for the 7-step transmission
sequence, and when the fade depth exceeds the built in fade margin, the
signaling alphabet is reduced and enough time slots are borrowed from a
resource sharing reserved time slot pool to maintain the data rate at the
fade site. From the prior art, it is known that a small pool of spare time
slots can protect a large community of users. In the present technique,
the use of coding during fade events is not considered because the channel
is dispersive.
Implementation of resouce sharing with two transmission rates requires
modification of the 7-step media access technique described hereinbefore.
With resource sharing, transmission would normally be at the higher rate
R.sub.1 during non-transmission impairment periods. If errors are detected
at the higher rate via standard error detection techniques, a receiver in
UDs 14-19, UIMs 23, or NIU 33 can request call processor 35 to schedule a
retransmission of the last block of data at the lower rate R.sub.2. Call
processor 35 would then cause the transmitter to retransmit the last block
of information during a subsequent corresponding continuous 41 or bursty
42 traffic period at the lower data using, for example, a longer block
V.sub.i, D.sub.i, or 44, or two or more equal length blocks. A transmitter
for accomplishing such technique of resource sharing is described, for
example, in U.S. Pat. No. 4,309,764 issued to A. Acampora on Jan. 5, 1982,
and the previously cited article to A. Acampora in BSTJ, Vol. 58, No. 9,
November 1979, at pages 2097-2111. Periodically the transmitter can retry
transmission at the higher rate. The frequency of retries depends on the
dynamics of the delay spread in the channel. Requests for lower rate
transmission and retries at the higher rate need only occur infrequently
since the channel normally varies very slowly with time.
* * * * *
|
|
 |
|
 |
Description |
|
