Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
CA 02307218 2004-02-06
HYBRID RF PACKET NETWORK FOR BOTH MOBILE AND FIXED SUBSCRIBERS
FIELD OF THE INVENTION
The present invention relates to the field of wireless data communications -
using satellites and land-mobile
radio in the form of terrestrial RF packet networks. More specifically, the
invention relates to a system and
method of combining satellite and terrestrial RF packet networks into a hybrid
network and is embodied in
fixed and mobile equipment capable of using this hybrid network.
DESCRIPTION OF THE PRIOR ART
Since the mid-1980's, a variety of mobile data communications networks, or
RFpacket networks, have
been deployed in industrialized countries to provide services for packet data
transmission and reception via
wireless links to mobile vehicles. All of these networks have some form of
gateway through which a
computer in a fixed-location can transmit or receive data packets.
The connection between the gateway and a computer in a fixed-location is
typically established using some
data link technology such as X.25, Frame Relay, Ethernet and so forth,
provided by a public
telecommunications carrier, and with which the gateway equipment is
compatible. Because the end-users of
RF packet networks are typically charged on a per-packet basis by the operator
of the gateway and the
wireless data links, it is convenient to refer to the end-users as
subscribers, which connect to the RF packet
networks through access devices. Typically, a fixed subscriber connects to an
RF packet network through a
plug-in card adapter for the data link to the gateway network whereas a mobile
subscriber connects to an
RF packet network through a radio modem which implements the airlink protocol
of the specific RF
network technology. Both the adapters and the radio modems are called network
"access devices".
A frxed subscriber consists of a fixed-location computer with an access device
in the form of an adapter for
the data link to the gateway. A mobile subscriber consists of a mobile
computer and an access device in the
form of a radio-modem.
1
CA 02307218 2004-02-06
A common and persistent problem with RF packet networks is the limitation of
the territorial coverage
which they provide.
A number of mobile data communications networks based on satellites are
becoming available which do
not have these coverage limitations, but which are more costly to deploy. This
is reflected in the rates that
need to be charged to subscribers. Satellite-based networks are those which
use satellites to provide the RF
data link, called an airlink, to the mobile subscriber. This is in contrast to
terrestrial networks which use
fixed-location RF transceiver base stations to establish the airlink, deployed
similarly to those in cellular
telephony.
In order to provide the least expensive service possible simultaneously with
unrestricted coverage, there is
a recognition emerging in the mobile data communications industry, of the need
to combine terrestrial RF
packet networks with satellite networks. For a mobile subscriber, this would
take the form of a single
access device incorporating radio modems for each network in the combination.
The term currently being
used to refer to such a device is hybrid radio.
It is important to note the architectural distinction between this combination
of networks and the classical
interconnection of two networks using incompatible data link technologies. In
the classical interconnection
scenario, computing devices connected to two heterogeneous networks may
communicate with each other
through the mediation of a router which is connected to both. The best-known
and most ubiquitous
example of interconnect technology is the Internet. The specifications for the
Internet may be found in
Postel, J., "Internet Protocol", RFC 791, USC/Information Sciences Institute,
September 1981.
In designing a solution for the problem of combining complementary RF packet
networks, the opportunity
presents itself to enable the resulting combination to interconnect with the
lnternet. Consequently, the
concept of a router plays a role in the principal components of the present
invention. However, to the
extent that the Internet solves a problem which is different from the one
addressed by the present invention,
it does not represent significant prior art.
2
CA 02307218 2004-02-06
Nevertheless, it is possible to design a solution based on a conventional
Internet architecture. In such a
solution, both mobile and fixed subscribers would have Internet addresses,
called IP addresses, for each RF
packet network to which they are attached. Each subscriber would become what
is known, in the
terminology of the Internet, as a multi-homed host.
Any Internet node wishing to communicate with a destination which is a multi-
homed host, in such a
manner as always to use an available, yet least expensive, route.- needs to be
aware of all the alternative IP
addresses by which the destination can reached. It should also have knowledge
of transient conditions
along the alternative RF paths because these affect the choice of IP address
to use in transmitting data to the
destination. However, since the Internet mechanisms for propagating
information about transient conditions
are available only to Internet routers and not to Internet nodes in general,
any Internet node sending data to
a hybrid network subscriber cannot make an intelligent choice for the IP
address to use. For these reasons,
the multi-homed host solution would be inferior to the one provided by the
present invention.
The intelligent routing decision with which the present invention is concerned
relates to the choice of
alternative RF path between two neighbouring nodes in a network. The method
used to support this
decision is similar t the one described in Lee, William C., "Network
Connectivity Control by Artificial
Intelligence" March 1991 (U.S. Patent 4,999,833) [D1], but only in the sense
that it specifies an adaptive
mechanism driven by information about transient propagation conditions
received from an eternal source.
DI relates to a communications network which is subject to changing
transmission propagation conditions,
such as an RF network. The combination of a rule base (RB), a knowledge base
(KB) and an inference
engine (IE) is used to provide a mechanism for dynamically choosing alterative
paths to a designation.
This is achieved by enabling any node in the network to maintain awareness, in
its KB, of the changing
paths of connectivity to any destination. Any transmitting node can use this
awareness to optimally route
data packets to their destination, using the IE and RB to effect an
intelligent routing decision. For instance,
a packet from node A to node D may traverse either node B or C, depending on
the current state of
connectivity between the latter two nodes and D. Awareness of these
connectivity states is stored in A's
KB and updated as a result of broadcast "advertisements" from B and C.
3
CA 02307218 2004-02-06
This mechanism is similar in purpose to the routing protocol described in J.
Moy, "OSPF Version 2", RFC
2328, The Internet Society (1998). This document specifies the "Open Shortest
Path First" protocol,
whereby Internet routers are kept aware of transient conditions within the
Internet in order to optimize the
forwarding of data packets to their destinations in the face of changing
network topology.
By definition, therefore, the methods described above are concerned with
choosing between alternative
neighbouring nodes as the router through which to forward data packets towards
their ultimate destination.
This decision depends on the transient topology in that portion of the network
"beyond" the neighbouring
nodes, which a transmitter cannot directly "see" but about which it is
"informed" by the routers within its
neighourhood. By contrast, the present invention is only concerned with
choosing between alternative
progagation means to a specific neighbouring node. These alternative
propagation means constitute paths
within total independent RF networks. (There is no issue here of dynamic
changes in the topology of RF
networks. Although packets may traverse a number of routers in the "wired"
part of the lnternet, they never
transverse more than one node after having entered the wireless domain.)
SUMMARY OF THE INVENTION
The present invention provides a hybrid RF packet network which enables the
subscriber to obtain
coverage which is unrestricted in terms of geographic boundaries but which
makes use of the satellite data
links only when it is required. It also enables the subscriber to send or
receive Internet datagrams. The
system comprises two principal components.
I. A Hybrid Network Radio which constitutes a network access device for mobile
subscribers.
2. A Hybrid Network Gateway, which constitutes a network access device for
fixed location subscribers.
From the subscriber's perspective, this system treats the combined radio
network as a single abstract data
link. Both the Hybrid Network Radio and the Hybrid Network Gateway are
addressable as Internet nodes
which are neighbours attached to this link.
4
CA 02307218 2004-02-06
By definition, both the Hybrid Network Radio and the Hybrid Network Gateway
have attachments, called
interfaces, to two (2) or more RF packet network technologies. The actual
mechanism for transmission
across the abstract link between any Hybrid Network Radio/ Hybrid Network
Gateway pair is a function of
the relative costs of traversing the airlinks. For each airlink, this cost is
called impedance, the value of
which varies with transient conditions such as commercial terms for (a)
transmission rates at different times
of the day/week/month, and for (b) the length (in octets) of the data packet
being transmitted, as well as the
ability of a Hybrid Network Radio to transmit or receive over the airlink.
Although the formula for
impedance measurement may be different for any Hybrid Network or Hybrid
Network Gateway, its
application in terms of the variables used is the same for all datagrams
transiting these nodes, regardless of
their destination. Both the Hybrid Network Radio and the Hybrid Network
Gateway always route traffic
through the airlink with the lowest impedance and able to detect changes in
the impedance value of each
RF packet network.
BRIEF DESCRIPTION OF THE DRAWINGS
Figures 1(a) illustrates the classical interconnect scenario which is
characteristic of the Internet and Figure
1(b) shows a method, inferior to the present invention, for using the Internet-
based concept of a "multi-
homed host" to create a hybrid network.
Figure 2 is a general schematic representation of the Hybrid Network Radio and
Hybrid Network Gateway
in relation to the combined RF packet networks.
Figure 3 is a detailed schematic representation of the Hybrid Network Radio.
Figure 4 is a detailed schematic representation of the Hybrid Network Gateway
in relation to the
proprietary gateways for each of the wireless networks and to the rest of the
lnternet.
Figure 5 is a schematic representation of the routing mechanism which
constitutes the essential innovative
characteristic of the present invention.
5
CA 02307218 2004-02-06
Figure 6 is a schematic representation of the behavior of either the Hybrid
Network Radio or the Hybrid
Network Gateway when a report is received from an RF packet network that a
previously transmitted
packet has failed to reach its destination, or that a prior attempt to
transmit a packet has failed.
Figure 7 Airlink Status Reporting is a schematic representation of the
behavior of the Hybrid Network
Radio when a report is received from the radio modem that RF contact with the
packet network
infrastructure has either been lost or re-established.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
Global System
Figure 1(a) is a schematic representation of the classical interconnect
concept as it is applied in the
architecture of the internet. This is shown in contradistinction to the nature
of, and the solution for, the
problem of hybrid networking. The internet solves the problem of data packets
traversing two or more
heterogeneous data links between source and destination computing devices,
whereas hybrid networking
deals with the choice of only one of several alternative RF paths to traverse
along a route between the
source and destination.
Figure I(b) illustrates the "multi-homed host" approach to solving the latter
problem as an example of a
solution which would be inferior to the present invention for the reasons
outlined in Prior Art.
Figure 2 shows the two (2) principal components of the present invention in
relation to the RF packet
networks that they combine. Hybrid Network Radio 20 is attached to a
terrestrial RF packet network 60 and
to a satellite network 65. Each of the RF packet networks have proprietary
gateways to which Hybrid
Network Gateway 80 is connected.
RF packet network 60 and its proprietary gateway 70 comprise a
telecommunications service as does RF
packet network 65 with proprietary gateway 75. There is not necessarily any
relationship, either technical
or commercial, between both services. In other words, both services may be
provided by two different
6
CA 02307218 2004-02-06
telecommunications carriers who can remain unaware that Hybrid Network
Gateways and Hybrid Network
Radios are effectively creating a combined network.
The combined network 105 is treated as a single abstract data link technology.
Any subscriber connected to
this data link can have a unique Internet address. More specifically, each of
the interfaces to this data link
from both Hybrid Network Radio 20 and Hybrid Network Gateway 80 has only one
Internet address.
Hybrid Network Radio
A Hybrid Network Radio in accordance with the present invention is shown in
Figure 3. The hardware
embodiment 20 of the Hybrid Network Radio contains reusable software module 30
at the core of which is
Internet Protocol module 35, called an IP module. IP modules are present in
all computing devices which
have an address on the Internet.
By definition, IP modules having only one interface to a specific data link
technology are called hosts. An
IP module with two (2) or more such interfaces, and with the ability to route
traffic, received from one
interface, to be transmitted on the other interface, is called a router.
The Hybrid Network Radio complies with the aforementioned definition of a
router. To support IP
communications with the mobile subscriber 10, it has PPP (Point-to-Point
Protocol) interface 15. The
specifications for the PPP may be found in Simpson, W., Editor, "The Point-to-
Point Protocol (PPP)", STD
50, RFC 16619 Daydreamer, July 1994.
For wide-area mobile communications, the Hybrid Network Radio has an array of
interfaces to RF packet
networks (data link technologies). Conceptually, these interfaces are for:
(1) A terrestrial RF packet network, labeled 45(a).
(2) A satellite RF packet network, labeled 45(b).
7
CA 02307218 2004-02-06
(3) Any additional RF packet networks which may offer a lower cost of
communications under specific
circumstances then either of the others. These are represented in Figure 1 as
the arbitrary n'th RF packet
network, labeled 45(c).
In a conventional Internet architecture, all these interfaces would have
Internet addresses, called IP
addresses. However, in terms of the conventional routing task of the IP
module, these are all combined into
a single abstract data link.
Hybrid Network Gateway
A Hybrid Network Gateway in accordance with the present invention is shown in
Figure 4, relative to the
proprietary gateways for each of the wireless networks and to the rest of the
Internet. The Re-usable
Software Module 30 of the Hybrid Network Radio is, as the name implies, re-
used in Hybrid Network
Gateway 80 and therefore all of its components, including the RF path switch,
are identical in functionality.
Similarly to the Hybrid Network Radio's relationship with the mobile
subscriber, the Hybrid Network
Gateway interfaces to a fixed subscriber 100 through a generic IP interface
90. In other words, 90 is any
interface to a network which can be assigned an Internet address and therefore
enables the Hybrid Network
Gateway to route between the Hybrid Network and the Internet. In this context,
the fixed subscriber 100 is
an Internet node.
Routing Mechanism
From the perspective of mobile subscriber 10 in Figure 3, the IP module in the
Hybrid Network Radio has
the same behavior as an IP module in any conventional router. It routes
traffic between the PPP interface
and the abstract data link which combines the RF packet networks However once
traffic has been routed to
the abstract data link, a choice must be made between an array of alternative
RF paths. Module 40, which is
not part of a conventional EP module, determines this choice, and is called an
RF path switch.
The functionality of the RF path switch should be explained in contrast to the
conventional Internet routing
mechanism, which is as follows. The basic data unit in the Internet is called
a datagram. As datagrams
8
CA 02307218 2004-02-06
transit Internet routers, the routing decision taken by the IP module is a
function of the destination address
encapsulated in the header of the datagram. A router will mask out the host
portion of the destination
address in order to extract only the network portion. If the router
incorporates an interface to that network,
then the datagram can be delivered directly to its destination. Otherwise, it
examines its routing table to
find a route associated with the target address and, if there is a route, it
delivers the datagram to the network
interface associated with that route. In other words, it passes the datagram
along to the next hop indicated
by the route.
In contrast to this, the datagram's destination address is not sufficient for
the Hybrid Network Radio to
determine the RF packet network interface to which the datagram should be
delivered. Figure 5 shows that
this is purely a function of the relative costs of sending the datagram,
(encapsulated in a data link layer
frame) along any one of the wireless data link paths. RF path switch 40
chooses the path of least
impedance. Therefore if the impedance I1 of RF packet network 60 is less than
impedance 12 along RF
packet network 65, RF path switch 40 chooses 60 as the medium through which to
transfer the datagram.
In a conventional IP router, the entries in its routing table can be
dynamically changed as topological
conditions in the Internet change. As routers are added or removed, or traffic
congestion problems are
reported, the routers which detect these phenomena can propagate the
information throughout the Internet
via a set of protocols in which only routers participate. As new information
is received, the router may
change some of its route entries.
In contrast, changes to the entries in the routing table of the EP module 35
are brought about by changes to
the impedance values for the RF paths by which the entries in the routing
table can be reached. These
changes are the result of error reporting and the airlink status reporting
mechanisms of the RF packet
networks, which are described in the following two (2) sections. Furthermore,
the IP module 35 and its RF
path switch 40 do not propagate changes to routing table entries because the
combined RF packet networks
constitute, from an external perspective, a single abstract data link. Any
transient conditions within this
data link, such as the relative cost of traversing any of the RF media, are
not significant to the "outside
world".
9
CA 02307218 2004-02-06
Error Reporting
Most RF packet technologies include a mechanism for reporting of transmission
errors to the users which
initially requested the transmission. The cause of errors can vary from
failures of the airlink to a temporary
condition of insufficient holding buffers in a modem driving the transmitter.
The network interfaces 45(a),
(b) and (c) from Figure 3 propagate these errors to the IP module 35 by
indicating a unique identifier for
the datagram as well as a unique code which represents the nature of the
error.
Figure 6 shows the behavior of a network interface and the IP router on
reception of an error report. Input
I from a network access device to an RF packet network interface 45 represents
an error report which is
assumed to encapsulate both the cause of the error and an identifier for the
transmitted packet which failed
to cross the airlink. (The network access device is shown to be a radio modem
50 (a), (b) or (c), as in the
case of a hybrid Network Radio, but it could also be an adapter 85, as used in
a Hybrid Network Gateway).
The RF packet network interface propagates this report to the IP module 40,
which, in turn, produces two
(2) outputs.
Output 3 is an instruction to generate, and queue for transmission, an ICMP
(Internet Control and Message
Protocol) error message destined for the source of the datagram which failed
to be transmitted. ICMP is the
method used within the Internet for nodes to communicate "out-of-band"; i.e.
to report problems and to
implement diagnostic request-response protocols such as the well-known "ping".
The specifications for
ICMP may be found in Postel, J., "Internet Control Message Protocol", STD 5,
RFC 792, USC/Information
Sciences Institute, September 1981.
The ICMP message type used in Output 3 is commonly called DESTINATION
UNREACHABLE, which
means that the router was unsuccessful in forwarding the datagram. The
specific cause of the failure is
derived from the error report received from the interface 45, and is recorded
in the CODE field of the
ICMP message.
Output 4 is an instruction to modify, in accordance with the nature of the
error, the impedance value of the
RF path for the entry in the routing table which corresponds to the original
destination of the datagram.
CA 02307218 2004-02-06
When the source of the datagram, either mobile subscriber 10 or a fixed
subscriber (see Figure 7), receives
the ICMP DESTINATION UNREACHABLE message, it may choose to resend the
datagram. Therefore,
the choice of path for the re-transmitted datagram will take into account the
new impedance value of the
first RF path.
The formula for calculating the impedance of an RF packet network may vary
with the technology and the
commercial terms offered for the service. For instance, if commercial rates
vary with the number of octets
in a packet and with the time of day, impedance measures should take these
factors into account. When an
error report is translated into DESTINATION UNREACHABLE, the impedance on this
RF path is set to a
value which cannot be exceeded on any other path, so that the RF path switch
will avoid this path until
conditions change.
Airlink Status Reporting
Some RF packet network technologies provide a mechanism for a mobile network
access device, i.e. a
radio modem, to generate reports regarding the status of the airlink, or the
ability of the device to transmit
on the airlink. With respect to the functionality of the present invention,
the most important of these reports
are the establishment and loss of RF contact with the packet network.
Figure 7 illustrates this mechanism in terms of how these reports are used by
the Hybrid Network Radio to
modify the routing table entries.
Input 1 is a status report from the modem 50(a) that RF contact with the
network infrastructure has been
lost. The RF packet network interface 45(a) propagates this report to the IP
module (Output 2), which
activates timer 36 in order to wait a suitable delay t before raising the
impedance of RF path 60 to its
maximum possible value (Outputs 4,5). Subsequently, the RF path switch will
avoid sending datagrams
along RF path 60 because its impedance will be as great as or greater than all
other RF paths for packet
networks to which the IP module is attached.
The impedance for the RF path 60 is reset to its original value when a status
report is received from the
radio modem 50(a) indicating that RF contact with the network infrastructure
has been re-established. This
11
CA 02307218 2004-02-06
is illustrated by Input 6 in Figure 7, which produces two (2) outputs. Output
7 is an instruction to reset the
impedance measure for RF path 60 in the routing table. Output 8 is an
instruction to generate and transmit
across the airlink a control packet indicating that the mobile subscriber is
within RF contact.
For any specific RF packet network, a control packet is defined as a packet
which does not carry a payload
for a higher-level protocol such as an [P datagram. In most RF packet
networks, the packet header includes
a field which is used to specify a type, wherein control packet types can be
distinguished from other packets
carrying payloads for higher-level protocols. If an RF packet network does not
define such a field, then it
must be defined, at the beginning of the user data area of each packet, in
such a manner as to enable the
transmitting and receiving network interfaces to recognize it. The packet
header also includes an
identification of the sender in the form of an address which is "native" to
the RF packet network. This is
commonly called a "hardware address" and is mapped within the receiver's
routing tables to an IP address.
The control packet resulting from Output 8 in Figure 7 is called an RF
PATH_UPDATE packet. It informs
the receiver that the sender has entered the coverage area of the RF packet
network. The receiver can
therefore change, if required, the impedance value for the RF path to the
sender.
The timer 36 in Figure 7, activated when loss of RF contact is reported, is
used to avoid unnecessary
transmission in the case where there are spurious oscillations between loss
and re-establishment of RF
contact. In other words, if contact is lost, the timer may still be canceled
if, before the interval has elapsed,
a status report is received indicating re-establishment of contact. In such a
case, the RF_PATH_UPDATE
control would not be generated and transmitted. Interval t can be calibrated
to each RF packet network
technology.
12
