Note: Descriptions are shown in the official language in which they were submitted.
CA 02387238 2002-04-11
WO 01/28162 PCT/LTS00/27959
-1-
Description
SPATIALLY SWITCHED ROUTER FOR
WIRELESS DATA PACKETS
TECHNICAL FIELD
The present invention relates to the field of
wireless communication in general, and in particular, to
a router for switched array antennas for high capacity
wireless broadband networks.
BACKGROUND ART
Wireless communication at high frequencies in
the range of l GHz to more than 100GHz are used exten-
sively for point-to-point (PP) and point-to-multi point
(PMP) communication. For these high frequencies, three
types of antennas are commonly used for spatial direc-
tional data transmission. Parabolic reflector antennas
are used for a fixed narrow spatial direction of trans-
mission. Sectorial horn antennas are used for fixed wide
area transmission. Patch antennas are used for fixed
direction transmission as well. Those antennas have
fixed lobe patterns aligned towards transceivers located
in a well defined spatial sector. Once the data link is
defined the antennas transmit and receive data from those
fixed directions, based on the MAC (Media Access Control)
layer either in a circuit connection form, in broadcast
form or in a polling form. In PP and PMP systems, the
transceivers' antennas at both sides of each link have to
be aligned to face each other and the antennas' alignment
is usually done manually during the initial link commis-
sioning. When setting up a PMP link, the antenna beams
at both sites have to be aligned simultaneously towards
each other to reach maximum received signal. In PMP
systems, the base station often includes fixed sectorial
antennas that are set initially to radiate in well de-
fined sectors, e.g. four low gain antennas of 90 degrees
that are positioned to cover 360 degrees. Thus, a sub-
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-2-
scriber's antenna has a narrower spatial divergence to
increase its gain, is aligned towards the base station
location in azimuth and elevation until maximum reception
is achieved. This alignment guarantees that the base
station is also receiving maximum transmission signal via
its large lower gain fixed sectorial antenna.
Data packets transmitted and received by the
antennas are coming from the same directions. In the
case of PMP system that uses FDM B frequency division
multiplexing, or TDM B time division multiplexing, or
other modulation technique, the base station can broad-
cast information dedicated to specific transceivers lo-
cated in a sector. All other transceivers in the same
sector will receive the data, decode it, but will ignore
it once it is found that the data is not aimed for them.
However, by sharing the sector among many transceivers,
only a limited amount of data packets can be forwarded
simultaneously among the transceivers when transmitting
at the same frequency.
The process of alignment in both PP links or in
the case of adding a new subscriber at a PMP system is
done off-line prior to service activation and involves
accurate mechanical adjustment while monitoring the re-
ceived signal level. In DBS (Direct Broadcast System - a
PMP using a satellite), antenna alignment is done in a
similar way to terrestrial PMP system. At the subscriber
location, the antenna is aligned towards a geostationary
satellite until a good signal is detected, and then it is
fixed mechanically towards that direction. In all of the
above-described cases, the antenna's aperture is aligned
mechanically towards the broadcasting source or towards
each other before establishing the communication link and
starting the service. Based on the received signal
level, the direction is mechanically adjusted, sometimes
by motor driven antenna, and fixed to the specified di-
rection of maximum reception and transmission.
Few techniques are used to route or direct data
towards different transmission directions. The most
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-3-
common is to locate a base station with multiple trans-
ceivers, each one with its own separate antenna, where
each antenna covers a different sector. The base station
MAC layer switches the data at baseband to the transmit-
s ter, which covers the sector that contains the subscriber
transceiver site where the data packets are aimed. At a
PMP base station, typical sectorial antennas such as horn
antennas are designed to cover fixed 90, 45, 30 or 15
degree lobes in the horizontal plane and about 7 degrees
in the vertical plane. The subscriber antenna, on the
other hand, is designed with much narrower beam
sensitivity, i.e. higher gain, with similar divergence in
horizontal and vertical planes, usually less than 7 de-
grees. Horn antennas, lens corrected horns and parabolic
antennas are commonly used for the subscriber
transceiver. Other PMP systems use a subscriber radio
with an antenna that receives the down-stream data from
the base station in one polarization, say horizontal, and
transmits upstream in a perpendicular polarization, say
vertical, towards the base station, thus increasing net-
work capacity. In all of the above cases, the spatial
capacity in a sector is fixed by the alignment of the
antennas.
Phased array antennas allow beam steering by
controlling the phase of each antenna element relative to
phase of the other elements thus allows beam steering.
Those antennas are complicated to control in a very short
duration imposed by the burst nature of the packets of
data. Thus, phased array antennas are currently used
only in some advanced cellular base stations to establish
circuit connections for relatively long duration data
transmissions, such as in circuit oriented networks where
the duration of voice conversation is relatively much
longer than packets of data. Phased array antennas are
used primarily at low frequencies, typically less than
2.5 GHz, to get high directionality in a multireflections
environment. The complexity, high cost and high loss of
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-4-
components, namely phase shifters, at high frequencies
prevent use for mass commercial applications.
A simple solution for switching data packets
towards different transceivers at different directions is
by fast switching the final output energy between differ-
ent sectorial antennas located in different angles, in
say the horizontal plane, thus covering a large field of
view. This configuration, however, demands a multiplic-
ity of antennas, each one aimed in a different direction
with a multiplicity of transmitters and connection lines
to feed those antennas. RF energy needs to be switched
and then transported, via long waveguides or coax, to
each antenna. The distance from the switches to the
antennas creates large signal attenuation, which
increases at higher frequencies, and demands increased
antenna structural dimensions, cost, and can be environ-
mentally objectionable. Thus, an objective is to control
a very fast switch for millimeter waves using a high
frequency switched antenna array, with the switch located
in close proximity to the antenna array. This is needed
to allow high bit rate packets modulating high frequency
RF to be efficiently switched towards different trans-
ceivers in different spatial directions.
SUMMARY OF THE INVENTION
An example of spatial routing of data packets
in the space between arbitrarily distributed wireless
nodes is described by Berger at e1. in PCT document WO
00/25485, published May 4, 2000 and entitled "Broadband
Wireless Mesh Topology Network". The wireless network
nodes are designed to select a transmission direction and
a receive direction based on the routing address of the
data packets to be sent and/or received. The selection
of a transmission or receiving direction is done
instantaneously to accommodate short bursts of data
packets arriving from nodes located at different
directions or transmitted towards nodes located at
different directions, as defined by the scheduler of the
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-5-
MAC layer of the network nodes, as explained in the prior
applications. A communication protocol that is designed
to support the scheduling of spatially routed packets be-
tween network nodes in any generic topology such as mesh,
tree and branch and PMP. The description of the media
access control (MAC) layer is particularly pertinent.
The presently disclosed spatial switched router
(SSR) describes a way of designing a data packet switch-
ing and routing apparatus capable of switching data pack-
ets and transmitting them spatially between wireless
network nodes. The MAC layer defines, in real time, the
direction and time of the RF switching, thus directing
data packets based on the packets routes, destination and
the network's node spatial location. The prior applica-
tions explain that RF switching is established by sched-
ules, held at each node whereby packets are directed and
received from specified, spatially separated, nodes at
appointed times. This MAC protocol is assumed in the
present invention such that transmission and reception
timing and the corresponding desired direction of trans-
mission or reception are known in advance. However,
other packet protocols can be used with address decoding
and routing information obtained by decoding of the pack-
ets.
The SSR apparatus enables the switching of
transmitted and received data packets from one node to
other neighboring nodes and from multiple nodes located
at different direction and distances to other nodes in
their surroundings. Fast switching is accomplished by
applying fast, in the range of few nanoseconds to a few
microseconds, control signals to a series of microwave
switches, synchronously with the data packet transmission
and reception timing and synchronously with the direction
of transmission and reception. The fast RF switches are
designed in a configuration that delivers large isolation
between the receiver and transmitter input ports and
minimizing the RF losses. The design allows close prox-
imity of the switches to the output feeding ports to
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-6-
reduce coupling and transmission losses especially impor-
tant at very high frequencies (>20GHz). An nxm switch
assembly (n = number of input ports, m = number of output
ports) is designed based on a series of custom made 2x4
integrated RF switches made of GaAs integrated circuits
(MMIC), designed for the very high microwave frequencies
(>20GHz) and a switching array assembly structure closely
coupled to the focusing and collimating antenna struc-
ture.
A principal feature of the current spatially
switched router apparatus is its wireless spatial packet
routing and switching capability to form a "connection-
less" communication link between a multiplicity of dis-
persed nodes in a mesh topology network or any other
derivative of a mesh topology network such as tree and
branch and/or PMP. At the very high microwave frequen-
cies, the system may require a line-of-sight (LOS) be-
tween the communicating nodes. The spatial transmission
of data packets, such as Internet protocol (IP) packets,
towards specific directions of the destination nodes
allows multiple nodes to transmit at the same time, at
the same frequency band and in the same area with minimum
mutual interference. This synchronized mesh network
increases the available capacity of the network dramati-
cally relative to the common "connection oriented" net-
works, used in many PMP systems. In those PMP systems,
the bandwidth at certain sectors is defined up front by
the antenna's fixed illumination pattern. The spatially
switched router apparatus of the present invention can
perform fast route diversity and fast load balancing,
taking full advantage of the bursty nature of the IP data
packets traffic.
The present invention is optimized for the very
high radio frequencies, such as the FCC assigned LMDS
(Local Multipoint Distribution Systems) spectrum, 27GHz B
3lGHz, and other spectra that are assigned to operators
on a regional basis. Those frequencies bands allow large
amounts of data distributed at such frequencies, as
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-
10.5GHz (UK, Latin America) 24.5- 26.5 (Europe) 38GHz -
40GHz (US) etc. At those frequencies, the attenuation of
transmission lines is very high. Thus, the current de-
sign is made of a very compact switch array matrix that
is closely coupled to multiple feeding ports, which are
designed to feed multiple, focusing and collimating ports
of a beam forming optics apparatus operating at radio
frequencies. One of the beam forming apparatuses de-
scribed in this invention comprises of a known multi-
layer, graded-index, cylindrical lens that forms a one
dimensional, say horizontal focusing device, wherein the
other dimension, say vertical, divergence is defined by
the aperture size of the feeding port horn. This appara-
tus design allows the formation of beams with different
divergences in the horizontal plane, where the horizon-
tal plane is the switching plane, and the vertical plane.
In a different beam forming apparatus of the
present invention, the feeding ports feed a multi-layer
graded-index spherical lens, such as an RF Luneberg lens,
to form beams with similar divergence in the horizontal
(switching) plane, and the vertical plane. In both de-
vices, beam switching can cover angles in excess of 120
degrees with very high gain and collection efficiency
from different directions inside the sector. The packets
of data modulating the RF carrier are switched to focal
points, where beams from different focal points are
collimated to destination directions. All the beams
share the same lens, and use an overlapping aperture, of
the cylindrical or spherical lens, thus significantly
decreasing the size of the wireless node antenna. The
smaller size allows lower losses of RF energy coupled
through the antenna, lower weight and minimal intrusion
in the environment.
BRIEF DESCRIPTION OF THE DRAWINGS
Fig. 1 is a top plan view of a wireless node
with a spatially switched router apparatus for wireless
data packets comprising a spherical or cylindrical graded
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-g-
index RF lens to be used in a network in accord with the
present invention.
Fig. 2 is a side plan view of a cylindrical
type RF optics focusing and collimating antenna with a
feeding port used in the spatially switched router
apparatus of Fig. 1.
Fig. 3 is a top plan view of an RF switch
assembly for use in the spatially switched router
apparatus shown in Fig. 1.
Fig. 4 is a top plan view following Fig. 3, but
with low noise amplifiers in the RF switch assembly.
Fig. 5 is a top plan view following Fig. 4, but
with power amplifiers in the RF switch assembly.
Fig. 6 is an electrical schematic diagram of an
RF switch employed in the RF switch assemblies of Figs.
3-5.
BEST MODE FOR CARRYING OUT THE INVENTION
With reference to Fig. 1, a wireless router
transceiver node in accord with the present invention is
shown incorporating an RF switch assembly coupled to an
RF optics focusing and collimating apparatus. This appa-
ratus is a graded index RF lens which may have a spheri-
cal or a cylindrical lens shape. The term "RF optics",
as used herein, means a device operating at radio fre-
quencies, but having optical wave effects on radio waves,
such as lens effects. The transceiver comprises a local-
port 23 that communicates with local networks. The
local-port 23 and similar multiple ports, may incorporate
a wire connection such as twisted pair for 10/100 base T
Ethernet connection, fiber optic connection or wireless
transceivers, dedicated for separate local access via
frequencies such as ISM (industrial-scientific-
medical) bands (2.4GHz and/or 5.7GHz) or the MMDS
spectrum.
A central processing unit 22 processes data
packets received and transmitted via the local ports 23
and over the air from the remote network nodes. The CPU
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-g-
22 supplies control signals to activate a switch assembly
in a synchronous timing sequence to switch data packets
arriving from different directions to the proper receiver
and transmitted data packets towards the proper remote
node directions. Assume that the remote nodes are lo-
cated at different directions, say directions correspond-
ing to the beams propagation towards directions 12, 13
and 14. The MAC layer used in this invention as an
example is based on a pre-established schedule of routing
information, particularly the data packets' routing path
and destination, priority, and links availability,
determines the data packet routing direction and
transmission or reception timing at each node, as more
fully described in application S.N. 09/328,105, mentioned
above. For example, packets received from a node
associated with direction 14, shown in Figure 1 by an (RF
Pulse) propagating towards the RF lens, are stored,
processed, sorted and routed to either the local port 23
or towards nodes associated with other remote nodes such
as directions 12 or 13 (see "RF pulses" shown in Figure 1
propagating in the direction 12 and 13). If routing
information is not known beforehand, packets must be
decoded to obtain the information.
The receiver 21 comprises an RF low noise am-
plifier, down converter, IF receiver, analog-to-digital
converter and demodulator. The transmitter 20 comprises
a modulator, digital-to-analog converter, IF transmitter,
up-converter, and RF power amplifier. The system 20 and
21 also includes local oscillators, high frequency oscil-
lators and phase lock loops in accordance with standard
high frequency radio transceiver design.
The RF microwave switch assembly 19 comprises
an array of RF switches to be described below with refer-
ence to Fig. 3. The microwave switch assembly is acti-
vated by the control signals applied by the central pro-
cessing unit 22. The control signals switch data pack-
ets, now modulating the radio frequency signal generated
by the transmitter 20, to one of the multiplicity of RF
CA 02387238 2002-04-11
WO 01/28162 PCT/LTS00/27959
-10-
feeders, such as feeders 16, 17, or 18. The feeders may
be standard microwave horns. Those feeders direct the RF
signal to the focal point of the RF lens 11. Lens 11 is
preferably a Luneberg lens of the type shown in U.S. Pat.
No. 4,309,710. Based on the synchronous control inputs
from the central processing area 22, the fast microwave
switches routes packets of data to and from the different
ports 16, 17, 18, and others. The feeders transfer the
RF frequency to the focal points around the lens 11.
Lens 11 collimates beams, for example from feeder 17 to
the remote node or nodes, located in direction 12, or
from feeder 16 to the direction 13.
The same feeders may also serve as receiving
ports for RF microwave energy arriving from different
antennas. For example, the RF beam from remote node in
direction 14 is focused towards feeding-port 18 and
synchronously, switched by the switch assembly 19 towards
the receiver 21. Thus, a single lens and a switch array
assembly structure is used for both transmitting and
receiving packets of data, minimizing the size of the
transceiver antenna together with an increase of the
field of view. In a time division duplex (TDD) mode, the
same switch assembly 19 is used for the transmit and
receive functions, thus reducing the cost of the
transceiver and allowing better matching of the burst
flow associated with data packets relative to the
transceiver output and input bandwidth.
The RF lens 11 has a radom 15 designed to
protect the RF lens material from accumulation of dirt,
ice, and rain directly on the lens surface. A heat sink
24 dissipates heat generated by the transceiver and the
router electronics.
Microwave lens 11 focuses microwave beams from
different directions, say from directions 12, 13 and 14
into the waveguide feeders, say feeders 17, 16 and 18,
respectfully, located around the circumference of the
lens to cover a large field of view that can extend over
120 degrees. The focusing of the RF energy is done by
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-11-
means of graded index dielectric layers centered one
around the other, where the high index of refraction
layers are in the center. The index of refraction
decreases as the diameter of the lens increases. The
index of refraction is based on Luneberg expression,
where the index of refraction of a spherical lens has
ideal imaging properties.
The RF frequency index of refraction distribu-
tion n(r) for a Luneberg lens extends over a finite ra-
dius r, and is given by:
n(r) - (2-rz/az)liz, for r <1. Equation (1)
n(r) - 1 for r >1 or r =1. Equation (2)
where r is the distance from the sphere (or cylinder)
center and a is the sphere (or cylinder) radius.
Luneberg lenses use a spherical symmetry and have maximum
concentration at focal points near the sphere or cylin-
drical surface for beams of rays arriving from different
directions.
Due to the longer wavelength of microwave RF
frequencies, usually many millimeters, the lenses are
usually implemented by the formation of multiple
dielectric spherical shells, usually made of half shells,
inserted inside each other and made of variable steps of
index of refraction.
In the current invention, a cylindrical lens is
also described in order to achieve an asymmetric beam
forming apparatus. This cylindrical lens is also shown
in Fig. 2. The graded index multi-layer microwave cylin-
drical lens is made of multiple cylindrical tubes with
variable diameter one inside each other, where the graded
index of each layer is reduced as the layer diameter
increases. The core is a cylindrical rod with the high-
est index of refraction approaching n(r ~ 0) ~ (2)l~z.
The one dimensional cylindrical microwave lens
31 implementation of the current invention is easy and
inexpensive to manufacture. The cylindrical design of
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-12-
Fig. 2 allows collimation of the diverging feed energy
from feeding port 33 in a divergence based on the lens
graded index variation relative to the cylinder radial
dimension a in Equation (1). The use of mufti-layer cy-
lindrical elements to form lens 31 allows the design of
high volume manufacturable lenses by forming long tubes
with different indices of refraction (say approximately
different type of tubes), inserting them inside each
other and then cutting the cylinder to the appropriate
10 length dimension to generate multiple lenses. The length
of the cylindrical lens is defined by the size of the
feeding aperture and is made to be compatible with the
beam vertical diffraction inside the cylindrical lens.
In the cylindrical configuration design of Fig.
2, the vertical aperture is defined by the horn vertical
aperture which defines the vertical beam 32 diffraction
from the feed from source 33 at the elongated focal plane
along the cylindrical surface and parallel to the lens
axis. A degree of freedom exists to design a different
beam divergence for the horizontal and vertical planes
based on the applications and the node covering angle.
In a spherical lens design case, a similar beam
divergence is formed for the horizontal and vertical
dimension. The output port can be implemented by a
waveguide feed or a patch feed design. Other techniques
such as flared waveguides can be used to form beam gain
variation in specific planes.
In Fig. 2, the horn 33 is feeding graded index
cylindrical lens 31. Multiple horns are located in a
sector around the cylinder, as shown in Fig. 1. The
divergence of the beam in the vertical plane is reduced
by controlling beam expansion from the feeding-port 33
using horn 32, which expands the RF beam in one dimension
only, along the cylindrical axis of the antenna. The
horn leads from the switch array assembly, to the focal
surface of the lens 34. The horn 32 has a horizontal
dimension kept smaller approximately the size of ~1/2 the
RF frequency wavelength, thus allowing the horizontal
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-13-
emitted beam to be expanded and collimated by the
cylindrical lens graded index structure. The narrow
horizontal horn dimension allows the positioning of
multiple horns side by side to achieve high resolution in
the switching plane. The vertical divergence 35 is
usually less than 7.5 degrees and can be designed to be a
different size based on the needed gain for each of the
beams in the horizontal plane 36, i.e. the switching /
steering plane. The lens 31 is seen to have a variation
in the graded index layers. The lens 31 has an output
aperture 37 that is partially shared among the
multiplicity of the feeding ports. Feeding port 33 has
an input polarized beam 38, where E is the field
polarization, coming from the switch assembly and
extending to the feeding-ports horn 32.
An advantage of a cylindrical design for the
lens is having feeding ports with a narrow aperture in
the horizontal plane, i.e. the switching plane, thus
allowing the co-location of multiple feeding ports around
the lens periphery, thus increasing the resolution and
gain of the multiplicity of beams. The multiplicity of
narrow beams each one of them with narrow divergence and
higher gain (high spatial resolution) in the switching
plane, allows the multiple use of the same frequencies in
the same lens coverage area and also allows an increased
bit rate in each direction due to higher gain in the
aimed narrower sector.
The RF beam polarization is defined by the
orientation and dimensions of the design of the horn
waveguide feeds, or alternatively, the patch antenna feed
design. The graded index lens configuration can be used
for both polarizations in cases where the system design
uses transmitting in one polarization and receiving in
the perpendicular (other) polarization. In FDD
(frequency division duplex) PMP (point to multipoint)
systems for example, the transmitting feeds could be
located above or below the receiving feeds using the same
cylindrical lens. In this way, both transmission and
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-14-
receiving can take place simultaneously at a different
frequencies. The same can be done at different
frequencies and different polarizations to increase
transmit-receive isolation even farther while transmit
and receive take place simultaneously. In those PMP FDD
systems, the single lens configuration allows high
resolution in sectorial selections around the base
station location.
Instead of waveguide beam feeders, a multiple
patch antenna's feeds direct RF energy to a curved
surface located at focal points of the lens, can be used
as well. In the case of patch antenna feed, a symmetri-
cal dimension feed is used for spherical lens feeds or an
asymmetrical design is used for cylindrical lens feeds.
Certain lens feed designs may also incorporate structures
that employ apodization for side lobe suppression.
Another design that incorporates a cylindrical
lens can be used simultaneously for transmit and receive
signals, usually at different frequencies. This design
locates the receiver's feeds in one plane around the lens
and the transmitter's feeds in a plane above or below the
receiver feeds. In this design, the receiver feeds are
isolated from the transmitter feeds, thus simultaneous
operation can take place, either at the same frequency
with different timing or on separate frequencies at the
same time. This allows the elimination of a transmit and
receive diplexer in a frequency division duplex (FDD)
design or the T/R (transmit/receive) switch in the TDD
design, Receiving and transmitting from different
direction at the same time can also increase the
isolation. In both cases, lower RF losses are achieved
together with an increase of bandwidth.
In Fig. 3, a data packet switch employs 2x16
ports in a switch array. Different switch configurations
such as 2x4, 2x8, 2x32 share a similar design concept and
could be used. In the 2xm designation, 2 defines the
input and output ports, from the transmitter and towards
the receiver, the m defines the number of feeding ports
CA 02387238 2002-04-11
WO 01/28162 PCT/CTS00/27959
-15-
towards and from the antenna. Those ports correspond to
the different spatial switching directions. A key
advantage of the current data packet switched router
invention design is the high isolation achieved between
the transmitting signal at transmitter input port 41 and
the receiver input port 52. That isolation exceeds 60dB,
typically better than 75dB. This is necessary in order
to be able to switch from transmitting of short bursts of
data packets to the receiving of short bursts of data
packets, without saturating the receiver or changing the
operating point of the power amplifier. This large
isolation is achieved by having switches designed with
greater than 30 dB isolation between each one of the
ports and having 2xm (m>1) type switches 46, 45, 54, 55
between the transmit switch 40 and the receive switch 50.
In general, switches 40 and 50 could be either 1xn or 2xn
type switches. In the 2xn switch, the extra port is used
for testing purposes to allow monitoring with external
test equipment of the different parameters of the data
packet switch array. Different monitoring and testing
can be performed, including routing signals from
different ports to the transmitter test ports 42 and/or
the receiver test port 51.
When any feeding port is switched to the
receiver, the associated transmitter port is switched
off. Thus, isolation is achieved by the double cascaded
isolation between the transmitter port and the received
signal path. Say that a signal from the feeding port 49
is fed to port 48 of switch 46. The received signal is
switched by switch 46 to line 47, and fed into receive
switch 50, which in turn is switched to connect line 47
to a port with output line 52 that leads to the receiver.
The transmitter port 41 is switched off at the time the
receive packets arrive. Thus, port 41 is isolated from
port 44 by more than 30 dB, and port 44 is switched off
as well at switch 46, thus adding additional isolation of
greater than 30 dB between port 44 and any ports at the
switch 46. Thus, line 47 of switch 46 is isolated from
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-16-
port 41 by more than 60dB. This leads to isolation
greater than 60dB between the receiver port 52 and the
transmitter port 41.
The RF signal from the transmitter is fed into
switch 40 via input port 41. Synchronously with the
arrival of the data packets to the input port 41, the
switch 40 is momentary enabled by control signals on line
40a, changing the voltage on its internal input diodes
connected to port 41, and one of the diodes at one of the
output ports. The RF energy that carries the data pack-
ets is transferred via the switched diodes, which form
the ports, towards one of the switches in the second
layer that feeds the antenna. For example, the control
voltage switches the RF energy entered at input port 41
towards transmission line 74 that transfers the modulated
RF to switch 46. The control voltage signal is also
applied synchronously to switch 46 using control signals
46a, which switch RF energy to port 48 that couples the
energy to waveguide 49. In turn, waveguide 49 feeds the
lens 61 which collimates the beam in the direction 65.
At the end of the duration of the data packets, the
control voltage is changed to turn off the RF path con-
necting ports 48 and 44 of switch 46 as well as the RF
path between ports 42 and 41 of switch 40. During the
reception of data packets via port 49, from direction 65,
a control bias voltage is applied to the switching diodes
at port 48 and 44 of switch 46, and the diodes connected
to transmission line 47 and port 52 at switch 50, as
explained below with reference to Fig. 6. This enables
the connection of feeding port 49 to the feeding port 52
that leads to the receiver.
Switching the transmit packets to direction 62
is done in a similar way by applying the control signals
40a to switch 40 and 55a to switch 55. The switch
assembly thus connecting the signal from input port 41,
to port 57 of switch 55, transmitting port of switch 55,
and then to port 58 leading to the feeding-port 59.
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-17-
To receive data packets from direction 62, the
control signals 50a and 55a switch both the receive
switch 50 and the feeding switch 55 to allow the
modulated RF signal from direction 62 to pass through
from the feeding port 59, port 58 to the receiver port 56
of switch 55 and from there to the output receiver port
52 that leads to the RF receiver.
Different combinations of switching routes are
possible with the above-described array, among others.
The ability to transmit simultaneously into multiple
directions by splitting the energy to multiple feeding
ports is present.
The present invention uses a new 2xm port
switch assembly comprising very high-speed switches
connecting to antenna feeding ports. This type of switch
design, with separate transmit and receive ports, allows
in addition to high isolation, the creation of a loop
back of RF signals. The loop back of RF signal is
enabled by, for example, switching port 41 to 44 in
switch 40, and port 44 to line 47 in switch 46 and line
47 to port 52 in switch 50. This allows measurements of
the losses and functionality of the different sections
and elements of the data packet switch array.
In the design of Fig. 3, the loss from the
feeding port through the two cascaded switches to the
receive port 52 is typically from 5 to 6 dB. The high
noise figure as a result of this high loss could be
partially overcome by adding multiple low noise
amplifiers (LNA) in the receiving ports of the 2xm
switches. Fig. 4 described the LNA incorporation. LNA
66 amplifies the signal from port 56 of switch 55 feeding
it to the receive switch 50. Similar LNAs exist for
other ports of the receive switch 50. An output port of
switch 46 has LNA 76. An output port of switch 45 has
LNA 75. An output port of switch 55 has LNA 78. Since
LNAs are relatively low cost, the advantage of adding
them is significant.
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-18-
Fig. 5 describes the case where higher power is
necessary. Power amplifiers can be added to the switch
array between the switch 40 and the switches 45,46,54 and
55. For example, power amplifier 67 is located between
switches 40 and 55. Power amplifiers 81, 83 and 85 may
overcome the losses added by the switch 40 and the
transmission lines before and after it. The cost of
power amplifiers is higher than LNAs, thus a trade off of
power versus cost should be considered. Due to close
proximity of the power amplifiers and the LNA, careful
layout should be applied to avoid coupling.
With reference to Fig. 6, a custom made GaAs
MMIC of a star shaped array of 6 diodes 91, 92, 93, 94,95
and 96, are each connected to a transmission line
corresponding to RF ports 101, 102, 103, 104, 105 and
106. All diodes are biased to short the connected RF
transmission line, by conducting to ground with bias
applied at the ports 81-86, in an identical manner. When
a control voltage is applied to the corresponding ports
81, 82, 83, 84, 85, and 86 the RF conductivity of the
diodes is reduced, enabling the transmission of the RF
signal from the RF port to the center 100. V~hen the
control signal is applied to two diodes, say 81 and 83
simultaneously, an RF signal from port 101 can flow to
port 103 and vice versa. This six diode switch,
fabricated as a GaAs integrated circuit, is used as a
millimeter wave switch for the 20-40 Ghz. The switch is
used as a 2x4 switch, and each of the 4 output ports can
be switched. The switch is part of the larger 2x16
switch of the present invention. A similar design is
also applicable for switches operating at frequencies
above ~40 GHz and below ~20 GHz.
By applying the control signal to switch "OFF"
any two diodes, say 91 and 94, a low loss RF path is
enabled from one RF port to another port, say ports 101
and 104 correspondingly. The switch allows any other
combination of diodes to act as switches when the control
signal are "OFF". At ~20 to ~40 Ghz the following
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-19-
performance characteristics exist. When all of the
diodes are "ON", the port to port isolation is greater
than 60 dB. The isolation between a port that is in an
"OFF" state and a port that is in an "ON" state is
greater than 30dB. The switch loss in the ~20 to ~40 Ghz
range is about 1.5 dB.
In summary, the present invention describes a
"data packet spatially switched router" comprising an "RF
switch assembly" with multiple feeding ports in close
proximity to "RF optics focusing and collimating
apparatus", that allows selective reception of data
packets from multiple narrow sectors, covering a large
field of view and selective transmitting of data packets
to multiple narrow sectors covering a large field of
view, where the same RF optics apparatus and switch
assembly, switch data packets to multiple directions and
receive data packets from multiple directions, based on
their routing information.
The data packet spatially switched router is
used either in the case of wireless time division duplex
(TDD) mode systems or frequency division duplex (FDD)
mode and can be implemented in mesh topology wireless
networks as well as PMP systems (a subset of mesh
topology) and Tree and Brunch systems to increase data
flow in the same coverage spatial region.
From the above, it will be seen that a network
topology may be implemented based on the spatially
switched router of the invention. Furthermore, when the
spatially switched router for data packets described in
the present invention is used in the wireless nodes that
comprise wireless data packet networks, a new type of
data packet network can be implemented where data can be
routed to different directions with timing and directions
based on the packet routing information. For example,
some of the new network topologies that can be supported
are:
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-20-
(a) Mesh Network - The network nodes incorporate the
"spatially switched router". A network node can
communicate with multiple neighboring nodes. The nodes
use TDD (Time Division Duplex) to communicate with each
other. Nodes transmit and receive at different times.
The transmit and receive intervals could be done at the
same frequency or at different frequencies. In some
network designs, the frequencies for transmit and receive
are fixed (e.g. by government regulatory agencies. The
agencies may assign specific frequencies for use by
specific networks). The nodes may also use FDD
(frequency division duplex) to communicate with each
other. In the FDD case, a node can transmit and receive
at the same time, but uses a different frequencies, with
large spectral separation to avoid the signal from the
transmitter interfering with the received signal.
(b) Mesh of Base Stations, wherein each base station acts
also as a point to multipoint distribution node to edge
nodes. This type of mixed network includes two types of
nodes. A first group of nodes are the main network nodes
that incorporate the "spatially switch router" and
operates as "Base Stations" to communicate data packets
to and from a second group of nodes. A second group of
nodes are simpler and standard in the prior art. These
are edge nodes or "leaf nodes" which do not incorporate
the spatially switch router of the present invention.
Those nodes have usually a single sector defined by a
fixed antenna lobe and are fixedly aligned toward the
main network node. The nodes of the second group, each
receives data from the main node and drop the data to its
local port or ports. The nodes of second group also
getting data from their local ports and transmitting
towards a fixed direction of the main nodes. Once getting
packet traffic from an edge node, the main nodes of group
one can drop the data packets of traffic directly to the
network backbone via its local port, route the data
packets to other base stations with whom it communicates
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-21-
via wireless mesh trunking network topology as described
in (a) above or route the data packets to other edge
nodes of the second group, located in one of its sectors.
As described, the simpler second group nodes
communicate by transmitting and receiving data from a
specific spatial direction of the main node, which
operates as a routing base station. Each of the group
two nodes has a fixed beam usually aligned mechanically
towards one or more base stations. In cases where the
main node do not have line of sight with subscriber edge
node another main node which includes the spatially
switched router could be located in a position which has
line of sight with the main node and the leaf node and
acts as a wireless repeater. The point to multipoint
section of the network may operate in either TDD mode or
FDD mode based on the system design and the regulation
for the operating frequency in the area of operation. In
FDD the subscriber will communicate with a main node,
i.e. a base station, by transmitting at a first frequency
and receive information at a second frequency. The base
station of the first group will receive the first
frequency and will transmit at a second frequency.
Combinations of frequencies may be used for different
sectors by each one of the first group of nodes avoiding
mutual interference and promoting maximum data packet
flow in the network.
(c) Basic point to multipoint system - In this network
topology case, member of the first group of main nodes
incorporate the spatially switched router of the present
invention used as base stations. Members of the simpler
group two nodes are used as subscriber nodes. The main
nodes communicate traffic to the backbone and the simpler
subscriber nodes communicate to and from the base
stations to the local subscribers. The spatially
switched router in this point to multipoint topology
allows the adaptive routing of data packets from and
towards different sectors covered by the spatially
CA 02387238 2002-04-11
WO 01/28162 PCT/US00/27959
-22-
switched router antenna based on data packet routing
information. The multiple sectors allow for a high
degree of frequency reuse and a higher antenna gain that
allows longer distances inside of each sector. Different
sectors can be assigned different frequencies based on
system design and data packet routing information.
