Note: Descriptions are shown in the official language in which they were submitted.
CA 02308819 2000-OS-15
Self-Organizing Network Architecture (SONA)
Ab3~act
Wireless con ivity holds the promise to revol ' a every aspect of modern
activity. The pac conversion fro fired" world to a "Wireless" world
continues to accelerate, ' h ually unlimited number of applications and
opportunities availabl a oited. The restrictions to this movement have
been more ec is than technica ' ature. The fundamental collapse of the
cost o chnology and associative to logical innovation has strongly
i nced the entry economics of wireless produc services in general.
Economic Gating Factors for Wireless Adoption
Wireless RF technology in its earliest stages was entirely Wide Area Network
(WAN) based. Microwave Relay, Satellite and Cellular services proved to be key
enabling technologies for the movement of high value voice, video and data
services. The key economic barrier that all three technologies overcame was
the
physical cost of providing service over both distance and broadly diverse
location, hence the use of the WAN nomenclature. Over the years, a number of
new WAN alternatives have proliferated, reducing the cost of entry
incrementally
as each new competitive WAN technology like PCS, Reflex 50, CDPD, etc.
entered the market. As valuable and useful as WANs have proven to be and
even with dramatic enhancements in the price/performance of WANs in general,
they have not proven to be the final answer for lower value wireless
transactions.
The Transition from Circuit Switched to Packet Switched Networks
Another economic breakthrough for wired and wireless services was the
migration of most new networks away from circuit switched to packet switched
data transmission. This transition opened up many new markets for transactions
that benefited from the increased efficiency of optimized bandwidth
utilization.
This new generation of hardware and packet driven protocols, such as TCPIIP
has more or less completed the conversion of the "Last mile" economy that used
to be dominated by wired service to the "Last Byte" economy that is
increasingly
dominated by low cost wireless services. The "Last Byte" refers to the search
for
new technologies that can profitably capture the vast volumes of low value and
comparatively smaller packet transactions that represent the next wave of
opportunity for wireless data services. These transactions needed to be
captured and managed locally to be of economic consequence for wireless data
services.
For localized replacement of close proximity wired services, new generations
of
wireless Local Area Network (LAN) technologies have developed. They now
represent an excellent way to deliver low value wireless transactions through
to
packet switched WANs. Wireless LAN technology costs in turn have collapsed
CA 02308819 2000-OS-15
to the point where very broad adoption of LAN technology has occurred in
midrange to lower value services applications. It is now common for WAN and
LAN technologies to be interfaced to enhance wireless price/performance
economics for entry into this vast array of lower value market applications.
This
new strategy works economically by concentrating a substantial number of lower
value transactions at the LAN layer into larger, higher value transactions for
transport through the WAN. The large volume of LAN devices being utilized
continues to collapse the costs of LAN devices.
Synchronous Versus Asynchronous Transmission
One of the main differences between a WAN and a LAN technology is the ability
to manage asynchronous transmission. At the device level, the ability to
negotiate a network connection in an asynchronous manner adds significantly to
the cost of the WAN device. For a WAN to be of practical use, it must manage
requests for transmission it receives from these remote transceivers. LAN's on
the other hand typically administer transmissions on the LAN in a synchronous
manner, whereby a LAN master polls other LAN devices for data waiting for
transmission. In this manner, any number of LAN devices within an area of RF
influence can be managed effectively with much lower component costs. There
is a limit however to the practical number of devices that can be effectively
managed in this manner.
There are many circumstances where the ability to deliver unscheduled
information such as alarms or status information, require an effective method
of
delivering this information quickly. In a polled environment, the status or
alarm
information only gets sent as often as the device is polled. In a WAN
environment, this same information can only be sent when the WAN device's
request to send is acknowledged. There is a practical latency to each of these
scenarios based on the number of transceivers involved. The fundamental
issues to be resolved for WAN or LAN selection purposes relate completely to
the price/performance requirements of the remote devices and the network layer
chosen.
Where certainty is an advantage or a requirement, a polled strategy is
invariably
utilized that guarantees that the system operator will know immediately if
there
are any transceivers failing to respond, rather than waiting for the device to
send
information. In a polled LAN strategy, there is a limitation to the number of
devices that can cohabitate on a single LAN, is based on the length of time it
takes for the LAN master to poll the individual transceivers in the LAN.
Synchronous polling of devices also limits RF traffic to the communicating LAN
pair. This reduction in locally generated RF emissions, reduces contention.
The Last Fundamental Barriers to AMR
Today, for any high volume low value transaction opportunity, the main
economic
barrier is no longer the price/performance capabilities of the WAN portion of
the
WAN/LAN system or the cost of LAN devices. It is now, rather the costs of
2
CA 02308819 2000-OS-15
physical deployment, implementation and operation of the LAN devices
themselves. This is especially true in applications like Automated Meter
Reading
(AMR) where the sheer volume of units involved creates logistical difficulties
of a
considerable magnitude.
One of the most difficult requirements in overall WAN/LAN deployment and
implementation is to securely effect network connection, address assignment
and
subsequent transaction (data) routing from LAN devices to the WAN. GPS has
helped many to locate and manage these broadly dispersed assets more
effectively. What is now needed to eliminate the last fundamental barriers to
low
value wireless transactions is to have the LAN devices organize, connect and
address themselves efficiently. These capabilities would overcome the
"Organizational Barrier" and make AMR truly economic as a standalone
application.
AMR is certainly one of the most attractive ways to capture low value
transactions that occur in significant volumes. AMR also represents a
practical
way to access a vast array of other potential applications that could be
considered economic when the appropriate price/performance strategy is
employed.
Many companies have recognized the potential for these AMR driven wireless
networks to support the provision of a broad range of low cost transaction
services. Emerging AMR strategies typically contemplate the use of the LAN
module employed in the meter, as a sort of wireless gateway for the transport
of
numerous other transactions that could contribute services revenue. While this
strategy offers a way of generating theoretical support revenues that accrue
to
the AMR network owner, regulatory and revenue uncertainty continue to make it
difficult to use this additional economic support as the basis for adoption.
Today, this potential revenue offers the only offsetting economics to support
the
additional costs of organizing the LAN portion of large AMR networks.
Eliminating the Organizational Barrier to entry will drive adoption of
wireless AMR
network topologies as well as the adoption of a vast array of associative
services.
SONA (Self Organizing Network Architecture) will be a key enabling technology,
which will unlock the potential for high volumes of low value transactions.
SONA
transactions will also displace the commercial position of a vast number of
higher
value WAN transaction opportunities that exist within SONA coverage due to the
low entry cost of the devices and the reduction in WAN airtime costs.
3
CA 02308819 2000-OS-15
RF Technical Challenges to Broad Network Implementation
The difficulties inherent in the broad implementation of any RF technology are
that RF device emplacement and the resulting RF path expectations are at best
a
theoretical expectation based on sound engineering practice. The localized RF
environments within the larger RF domain will in practice be either slightly
better
or worse than the theoretical average RF path expectation. This uncertainty
has
created problems for RF engineers who need to design network architectures for
broadly dispersed transceivers. To add to this difficulty, the localized
density of
transceivers varies considerably, especially in a utility meter AMR
deployment.
Device-to-device RF based communications are optimally accomplished by using
the lowest amount of signal required to maintain acceptable signal to noise
ratios. Managing great variations in distance between transceivers is normally
handled by manipulating transmission power and antenna placement and
selection on a case-by-case basis. This is an impractibly complex and
expensive
process to manage individually in large deployments. To be economically
viable,
all of the transceivers must be identical in their characteristics. It should
also be
expected that personnel unskilled in RF would perform their implementation.
The biggest long-term problem faced in the RF environment is that it can and
will
change over time. Vegetation growth, new structures, and undetermined
external RF background noise will eventually impact portions of the RF
environment. In normal circumstances, path profile studies are performed to
characterize the RF environment prior to deployment to create some certainty
of
the theoretical performance that can be achieved. A path profile study of the
granularity necessary to ensure success in a high volume of widely distributed
RF devices is not economically viable, if possible at all, given the RF
environments propensity to be different at the time of deployment than during
testing.
In most circumstances, only the highest value applications can justify
individual
path profiles between transceivers. In an AMR deployment, economics do not
allow this type of approach. WAN concentrators are usually pole or tower
mounted, utilizing existing utility assets and leased locations to create an
economically viable line of sight path between LAN devices and the WAN/LAN
concentrators. WAN/LAN concentrator distribution requires a considerable
amount of detailed engineering and repeated follow up after deployment to get
a
system that is reliably operational. This would become a much more
manageable scenario if the meter based LAN modules could act as network
routers, acting as secure RF paths between themselves and the WAN/LAN
concentrators. In a small LAN grouping of known locations and positions, such
as a local office environment, reasonably successful network routing can be
achieved economically. However, in a typical million point AMR deployment, the
physical cost of mapping out and testing individual network routing
assignments
becomes a complex undertaking involving a great number of skilled technicians.
4
CA 02308819 2000-OS-15
The SONA Solution
During installation of the LAN meter nodes, the installer captures a GPS
reading.
These GPS coordinates, the meter ID, Customer ID and a unique password are
placed in a mobile data unit, carried by the installer for subsequent download
to a
database for verification and later reuse. Essentially, this information lets
the
utility know where every customer is, the meter type, ID, etc. The WAN/LAN
system can then use this information locally to create what we are now calling
a
"Self Organizing Network Architecture" (SONA).
SONA Device Descriptions
There are essentially two basic types of SONA devices:
~ The LAN node, or Local Area Network transceiver.
~ The WAN/LAN concentrator, which incorporates a WAN transceiver,
WAN/LAN interface board and a LAN node.
The LAN node incorporates a 902-928 MHz FHSS radio (or Other
Radio/Frequency), with on board microprocessor, flash memory, power supply,
and onboard omni directional antenna. For meter applications, either a serial
port connection or a pulse counting module is incorporated into the meter
prior to
installation. Any LAN node can become a WAN/LAN concentrator when a WAN
transceiver and WAN/LAN interface are installed. Any type of WAN connection
could be used, and as such is not a key factor in SONA deployment, other than
its capacity to manage anticipated WAN transaction requirements. Typical WAN
connections could include Two Way paging, CDPD, GSM, IDEN or RDLAP or the
newer Broadband network transceiver modules, as examples.
The single chip CMOS LAN transceiver section (or Other implementation) can be
safely operated at power levels between 200 to 300 milliwatts, without
incorporating additional amplifier circuitry, depending on overall RF circuit
attenuation. The transceiver section of the LAN node can have its power level
set
through a binary scale of 0 to 255, via the onboard processor, allowing a wide
variance in transceiver output.
Every LAN node incorporates sufficient memory to include the Media Access
Control layer functions required to effectively act as both a LAN node and a
WAN/LAN master. Additional memory is reserved for routing table information,
local data storage, parameter storage, as well as any executable code required
for local applications management, such as tracking pulses from the meter.
Each LAN module also supports the attachment of Sub-LAN devices/services
through a locally managed Sub-LAN addressing strategy. In Sub-LAN service,
the local LAN node acts as a master to other co-located LAN devices that
negotiate Sub-LAN assignment to the LAN node, which then effectively operates
as a Sub-LAN master, routing data through to the WAN.
CA 02308819 2000-OS-15
SONA Registration Management
Every device in the deployment, once it gets installed, has its GPS
coordinates
loaded into memory via communication with a GPS/LAN module held by the
installer or through an IRDA port or other means. WAN/LAN concentrators are
distributed in a systematic fashion, consistent with WAN to LAN ratio
requirements. i.e., every 200-240 LAN node devices a WAN/LAN concentrator
gets implemented. The exact ratio could be any number of devices, but is
strongly influenced by the amount of data that needs to be transported by the
WAN. This WAN emplacement can, without great difficulty, ensure that WAN
devices are placed evenly within the overall LAN node distribution.
The LAN node address structure uses a single address byte that supports 255
individual addresses, which is incorporated into the protocol packet for
transmission. A second address byte incorporated into the protocol packet
allows for the attachment of 255 Sub-LAN addresses. Distribution ratios during
initial deployment of the network should necessarily contemplate new addresses
being assigned over time. The initial ratio of WAN/LAN to LAN devices is best
kept between 200 and 240 units depending on average device densities and the
potential for higher densities of devices in the future. A number of addresses
are
reserved for maintenance and supervisory services. Maintaining a consistent
set
of reserve addresses throughout the domains provides a method for mobile LAN
devices to be used by Utility personnel for local interrogation of meters and
other
transaction services. This methodology also allows recovery of meter data when
a WAN device fails or if the WAN network is unavailable via a LAN transceiver
using a reserve address to make a LAN connection anywhere in the network.
Every WAN/LAN concentrator and LAN node starts listening actively once it is
installed and GPS located. The GPS coordinates, Device ID, Meter ID and
Customer ID as well as a password for access are downloaded into the Network
Management System (NMS) from the installers portable units. Every device
deployed will only respond to a packet that incorporates the unique LAN ID,
password and GPS coordinate of each LAN node or WAN/LAN concentrator.
The NMS incorporates this information into a GPS/GIS oriented database. This
database is structured to support an algorithm that organizes LAN nodes that
surround each WAN/LAN concentrator into individual domains. The WAN device
ID is used to give each domain a unique ID. If a WAN device fails or is
changed
over to a new network, the domain name changes to the new WAN device ID.
The average domain size is based on the ratio of WAN/LAN concentrators to
LAN nodes deployed.
Since the WAN device ID is typically either a unique 32 bit or a 64 bit
number,
depending on the type of WAN employed, there are virtually an unlimited number
of domains that can be created.
6
CA 02308819 2000-OS-15
At this initial stage, the domain file created is still a theoretical
extrapolation,
based on the GPS information provided. This file determines the LAN nodes that
the WAN/LAN concentrator is expected to capture during its SONA registration
sequence to enable effective territory wide SONA network coverage. The NMS
creates a domain file based on this extrapolation and downloads this file to
the
WAN/LAN concentrator that will operate as the domain master for the LAN nodes
nominated by NMS.
The domain file provides the LAN node ID, the GPS coordinate of the LAN node
and the password assigned during installation. As part of the domain
organization strategy, all of the LAN nodes to be captured are assigned a
position in a defined acquisition sequence, based on the relative distance
between the WAN/LAN concentrator and the individual nodes. The sequence
also contemplates the relative vector angle of the LAN nodes in reference to
the
WAN/LAN concentrator. This effectively forces a spiral structure to the
sequence
starting from the LAN node closest to the WAN/LAN concentrator and ending at
the most distant.
The WAN/LAN concentrator transmits its first request at a preset power level.
This power level is a theoretical calculation of the transmission power
required
for secure transmission between the two devices. The power level versus
distance algorithm is a preset value, based on the known capabilities of the
transceiver. The device meant to receive the packet sends its response at the
same power level. Assuming that the bit error rate is acceptable, the WAN/LAN
concentrator then registers the device. Registration of the node, assigns the
domain, sets the device status, sets the power level for transmissions, and
time
synchronizes the device to the network. If the polled device does not respond,
then the message is sent again at a higher power level. This methodology
guarantees that the lowest practical transmission power is used to arbitrate
and
register each new LAN node in the sequence. (A more detailed description of
Power management is provided later in this document.)
Upon receipt of a valid transmission from a polled meter in the sequence, the
WAN/LAN concentrator registers the node. During registration, the LAN node is
given a unique LAN address that corresponds to its number in the domain
polling
sequence. The power level needed to establish a secure RF connection and the
bit error rate achieved are also written into the table upon registration. A
Received Signal Strength Indication (RSSI) can also be incorporated if
available.
Once a LAN node is registered into a WAN/LAN domain, it ignores all
transmissions except those registered to that particular domain. The WAN/LAN
concentrator repeats this routine for every LAN node on its domain listing. In
most residential applications, low meter densities or local RF path
obstructions
will not allow the capture of all of the LAN nodes in the domain list during
the
initial capture sequence. When the full domain list of LAN nodes have been
either captured and registered or polled without achieving capture, the
WAN/LAN
concentrator sends a file to the NMS comprised of the successfully captured
LAN
nodes, the power levels used and the achieved bit error rates.
7
CA 02308819 2000-OS-15
The NMS places this information into the main database. The power level and
achieved bit error rate information is the equivalent of an automated path
profile
study between the WAN/LAN concentrator and the captured LAN nodes. This
data can be used to interpret the local RF environment by the NMS. It can also
be used for comparative analysis in the future to determine whether
transceiver
degradation or RF path degradation is impacting LAN node performance over
time.
The NMS then overlays the resulting "path profiles" between the WAN/LAN
concentrator and the individually captured LAN nodes onto the GPS/GIS map
area that contains the remaining LAN nodes that were not captured during the
initial registration sequence. This information provides a detailed
understanding
of the local RF environment surrounding the captured devices. This data also
provides an excellent statistical probability of the quality of the RF
environment in
those areas in the domain where LAN nodes failed to respond in the initial
sequence. An algorithm that interprets this information can then select a
subset
of the captured LAN nodes that have the highest probability to capture the
balance of the LAN nodes that were not captured during the initial SONA
sequence.
A new download file is then constructed by the NMS, based on the results of
the
data interpretation. This file contains the sequence numbers of the LAN nodes
chosen to act as LAN node concentrators. A LAN node concentrator acts
effectively as a repeater between LAN nodes that failed to register during the
primary SONA sequence and the WAN/LAN concentrator. The file also
describes the most probable polling path or route to each LAN node that failed
to
respond during the initial registration sequence. This file is incorporated
into a
routing table. The routing table describes the proposed path for each LAN node
that was not captured during the original polling sequence.
The WAN/LAN concentrator contacts each LAN node selected for use as a LAN
node concentrator and reregisters the LAN node as a LAN concentrator. During
the reregistration a file containing a list of LAN node addresses is sent that
represent LAN nodes that the concentrator is expected to manage routing for.
Whenever the WAN/LAN concentrator for the domain sends a message for one
of these addresses, the LAN concentrator that has received one of these
addresses will repeat the message at the power level indicated by the
registration packet sent by the WAN/LAN concentrator. The LAN concentrator
will repeat any response from the LAN node being contacted, since its address
is
incorporated in the return packet. The capture strategy employed by the
WAN/LAN concentrator during the initial sequence is employed as before, with
the only difference being that it factors in the additional delay of the
repeater
function into its capture strategy. Any LAN nodes captured during the new
registration sequence are registered and given their LAN address to the domain
as before. Upon completion of the new registration sequence, the power level
and bit error rates of the newly acquired LAN nodes are sent to NMS for
8
CA 02308819 2000-OS-15
incorporation into the database and the GPS/GIS path profile map for the
domain.
In most circumstances, this secondary capture will complete registration for
the
domain. In circumstances where this is not the case, the NMS will reinterpret
the
newly gathered path profile data, and either select newly captured nodes or
alternate nodes to be used as LAN concentrators to capture the balance of the
missing LAN nodes from the original domain sequence.
The same registration strategy is employed as for the secondary capture, with
a
key difference being that more than one LAN concentrator will be part of the
route. The new LAN node addresses are added to the repeater listing of those
LAN nodes that become part of the path. The WAN/LAN concentrator factors in
the delay of the additional repeaters in the path, as before.
This process is repeated until all LAN nodes in the domain are captured. At
the
NMS, supervisory personnel can monitor the progress of the domain
registrations. The number of registration sequences employed (or the maximum
route length) can be preset by authorized personnel to a preset limit. This
allows
the propagation of the network within the domain to be monitored by expert
supervision. This approach would allow the user to adjust algorithm
parameters,
based on analysis of gathered path data, adjust power levels used and follow
up
on situations indicating inoperable or otherwise unavailable devices, to
essentially manage network propagation in an optimal manner.
In circumstances where a preferred propagation strategy is required, such as
on
the edge of WAN coverage, a manually generated domain selection could be
sent. Rather than the WAN/LAN concentrator being in the geographic center of
the domain, it could be at the edge of the domain, where WAN network coverage
exists. This method could also be used to force propagation solely to fill
gaps in
the SONA coverage, or redefine coverage at a later date.
9
CA 02308819 2000-OS-15
SONA Transmit Power Management
FCC guidelines for 902-928 FHSS radio modems allow a maximum power level
of one watt. A key aspect of the SONA strategy is that the LAN node transmits
at
either a predetermined or programmable power level. Few AMR deployments
utilize the full power allowed by the license, due to the additional cost of
the
transceiver when amplification is required. Also, the general difficulty of
managing contention and packet collisions increases in high densities of
devices
as the transmission power increases. Power output is either based on the
theoretical transmission requirements for successful acquisition of the packet
by
devices within the target distance or a value supplied by NMS based on other
influencing characteristics of the RF environment. A power level consistent
with
the known propagation characteristics and therefore the transmit power
requirements for the distance value, might be 1 to 10 milliwatts depending on
the
attenuation of the device antennae. (Note: A key consideration of FHSS is that
unregistered devices must always operate on a default FHSS hopping sequence
before final arbitration and registration into a new domain. Once registration
is
complete, and the appropriate numbers of LAN nodes have been acquired,
assigning unique hopping sequences for co-located domains, will reduce local
RF collisions and contention. This reassignment is managed through the NMS)
The registration packet incorporates the binary power setting of the transmit
power used for transmission. If the first attempt does not successfully
receive a
valid acknowledgement transmission from the LAN node contacted, the
WAN/LAN device retransmits at the next staged increment in power level, for
example 10 to 100 milliwatts. In the case where a repeater or repeaters are
involved, the same case holds true. Any repeaters involved use the previously
established power transmit level for retransmission of the packet message. It
is
important to note that this could mean that it uses different power levels for
retransmission depending on the LAN address of the recipient.
Note:
Many ISM band transceiver chip sets also incorporate a RSSI (Received Signal
Strength Indication) capability. This information can also be used to
dynamically
select power level. RSSI looks at the signal power of received transmissions.
Devices could negotiate power levels until minimum signal to noise ratios are
achieved. RSSI is also a simple way to identify signal degradation over time,
and
could be used to have the individual devices manage their own power levels
using the RSSI indication as the basis for a transmit power set point.
CA 02308819 2000-OS-15
Missing LAN node Arbitration and Domain Registration
In large volume device deployments, it must be expected that some number of
devices will exist in or surrounding WAN/LAN concentrator domains that did not
acquire domain registration. The domain sequence is then compared to the
registration information sent by the WAN/LAN concentrator. The NMS then
generates a list of "missing" nodes for supervisory intervention. NMS can also
be
used to generate alternate routing strategies to acquire missing nodes, by
comparing known GPS coordinates of missing devices with known RF paths and
vector information of registered LAN and WAN/LAN concentrators. In many
circumstances, devices along Domain boundaries may need to be reassigned to
a new domain to achieve registration. NMS can also manage redistribution of
LAN nodes to balance loading evenly among domains, by reassigning LAN
nodes to co-located domains.
SONA Maintenance
Each LAN node now has a known path back to the WAN/LAN concentrators.
When the WAN/LAN concentrator is unable to transmit or receive to any LAN
node due to either a transceiver failure or a reduction in RF path, it
notifies the
NMS. The NMS in turn, initiates a network repair sequence. The sequence
follows a similar pattern to the SONA sequence. In most circumstances, the
first
repair attempt will simply increase the transmit power of the sending
transceiver
to see if a response can be achieved. If increasing transmit power does not
repair the network connection, then NMS can reroute to another LAN node that
represents another probable path/route in the domain. Failure to complete any
transaction generates a failure packet, which is sent to a log file and then
passed
onto the system administrator. NMS can reassign routes automatically or be
rerouted manually if required or desirable. NMS manages WAN/LAN
concentrator activity, following up on any failure and generates a service
report
recommendation for maintenance follow-up.
In the case of a device failure, where replacement is required, SONA
registration
is simplified. The new installation information is used to update the
database,
and the original registration information is used by the WAN/LAN concentrator
of
the domain to register the newly installed LAN node.
The WAN/LAN concentrator monitors network health during polling, comparing
bit error rates of the LAN nodes registered in its domain table to determine
certainty of the established routing. Circumstances where transmit power has
been increased to ensure network connection, or where rerouting has proven
necessary are updated in both the NMS database and the WAN/LAN
concentrator's routing tables as part of routine network management. NMS can
initiate a forced domain registration scan on demand where circumstances
warrant, such as a natural disaster or in a managed network maintenance
strategy.
11
CA 02308819 2000-OS-15
SONA Sub-LAN Addressing and Registration
Sub-LAN devices must also be considered in the overall network strategy. Each
LAN node is capable of managing up to 256 devices as an address layer below
the SONA LAN node using a single address byte. This second byte is part of all
LAN addresses. In circumstances where the LAN node is being polled, the
second byte is always represented as 0. Addresses 1 to 255 on the second byte
are reserved for Sub-LAN devices. The primary economic rationale for a Sub-
LAN address is to support localized associated services at very low
transmission
levels. Associated services are given a unique assignment ID. The assignment
ID limits the protocol overhead required to manage the transactions through
the
network. A more complete description of the data structure and data
management rationale for both LAN and Sub-LAN devices can be found in
Appendix "A" attached. This assignment ID also lets NMS know the ultimate
destination of data packets, which for this type of information might be a
completely different database structure in a different location or Internet
address
than primary services. The ID also identifies the construction of the packet
and
the way information is organized within the packet. In this way, the routing
of
data and the organization, structure and home destinations of the information
can be completely separate, yet efficiently defined and managed.
Gas and Water Metering as examples of Sub-LAN Services
Since power is best accessed by installation of an electric meter as a
"gateway"
device, it is most likely to be the method of entry for a LAN node. The most
likely
associated services that could be attached to the Electric meter LAN node are
other metering activities such as Gas and Water metering. Supplying a source
of
electric service for power for Gas or Water AMR is extremely cost prohibitive.
Gas and Water metering require an aggressive power conservation strategy to
be used to extend battery life. The ultimate strategy is to have battery life
be
equal to the useful service life of the installed meter. Typically, battery
life
requires the meter to keep the transceiver section turned off, except during
scheduled intervals where meter information is uploaded to a local LAN node
for
routing to a WAN/LAN node.
Installation of the sub-meter, whether Gas or Water, is performed identically
to
Electric meter installation, other than the mechanical portion of the
installations.
The installation information, including GPS coordinates, LAN ID, meter ID,
Customer ID and unique password, is downloaded to the appropriate supervisory
system and then subsequently to the NMS database. A Sub-LAN device is sent
a message from the NMS via the WAN/LAN concentrator and LAN nodes whose
coordinates most closely equate to the GPS location of the Gas or Water meter.
Since data organization and data routing are completely separate, the LAN node
next door to the Gas meter could be used rather than the LAN node of the house
where the Gas meter was installed. Managing RF efficiency and data integrity
is
best handled through the closest LAN node.
12
CA 02308819 2000-OS-15
The NMS generates routing instructions by a review of the existing path
profile
information. A Sub-LAN Address incorporating the LAN node address of the
selected LAN node is generated and sent to the WAN/LAN concentrator. Once
the WAN/LAN concentrator has acquired the routing information, it reregisters
the LAN node as a Sub-LAN master using a simplified SONA acquisition
sequence. The LAN node, acting as a Sub-LAN master then transmits a packet
that includes the Gas or Water meter LAN ID and the password. The Sub-LAN
Gas or Water meter is then given its new Sub-LAN address. As part of the
registration process, a time date stamp is sent to the meter to schedule
activation
of its receiver section. The Sub-LAN master node, which also synchronizes
itself
to the Gas or Water meter LAN node's activation schedule, sends a request for
information to the meter at the appointed time. Once the new meter data is
received, it sends an acknowledgement packet containing a new time stamp. In
this way, time synchronization is maintained between the LAN node and the Sub-
LAN node at all times, and transceiver activity is limited to a minimum,
extending
battery life. This same strategy can be used for any battery powered Sub-LAN
service.
Other Sub-LAN Services
There will be two other types of Sub-LAN services. Fixed Sub-LAN services
such as phone cut-line detection or load regulation such as A/C, water
heating,
etc. would be good examples of the first. The second would be the support of
mobile transceivers operating as Sub-LAN subscribers. Fixed Sub-LAN services
would operate identically to sub metering, except that a power conservation
strategy would not be employed. Conceptually, randomly adopted Sub-LAN
services would be difficult to integrate as GPS located nodes. This type of
services node would operate within the SONA architecture as a locally assigned
and registered device. .To limit overhead and contention this registration
would
add the service as a Sub-LAN domain participant. NMS will manage this
strategy in a similar fashion to standard SONA, using theoretical diversity
based
on known RF paths and LAN node locations as the grouping strategy rather than
deterministic GPS radius vector strategy to provide routing alternatives.
The concept of mobile services offers many exciting opportunities. Courtesy of
the network coverage created by SONA, asynchronous attachment of mobile
devices will be accomplished with certainty throughout the SONA services area.
The local LAN node is constantly scanning whenever it is not being directed to
act as a router or is providing meter information. Mobile service nodes can
send
out a transmission asking for custom assignment to the SONA network. The
mobile node would have its own GPS transceiver. Mobile nodes would act as a
Quasi-WAN/LAN master, initiating a truncated SONA sequence requesting
devices within the target radius to respond to its custom attachment inquiry.
The
device is given temporary registration on the network by the coordinating
node.
Once the mobile device acquires a network connection it can send and receive
data over the network. Upon completion of the transactions the device
13
CA 02308819 2000-OS-15
deregisters from the network. This strategy has enormous implications for
wireless credit card verification by delivery people, as a key example.
There is a strong potential for 2.45 GHz or what some are generally calling
"Bluetooth" adoption over the next few years. Assuming that there is a need
for
attached services identified by the owner of the SONA network, LAN nodes could
be deployed with Bluetooth modules. There is a great deal of rationale for
this
practice, both from the standpoint of the cost of the Bluetooth modules as
well as
the probability of a number of associative services opportunities. Bluetooth
will
have a maximum range of 100 meters, and manages arbitration between itself
and other Bluetooth devices. The LAN module only participates as a network
connection when data arrives at the LAN node Bluetooth module that requires
network transport for a transaction. Many of these new application devices are
being referred to as IP appliances. One of the most logical reasons for a
utility to
want to consider this option is the fact that many manufacturers are now
tamping
up to use Bluetooth hardware and protocols as their standard device interface
for
a wide variety of products. Broad commercial acceptance of a single standard
for 2.45 GHz protocol by consumer product manufacturers will make a $5.00
wireless modem a possibility if not a certainty.
Wireless Data Economics
There are a lot of data services currently being managed over wired networks,
especially large volumes of Internet traffic using TCPIIP protocol. Most data
services are being converted to a TCPIIP format to take advantage of the low
cost and increased accessibility of Internet transport. For Wireless
transport, in
the interest of accessibility, Wireless Internet Protocols or Wireless
Application
Protocol (WAP) as some are calling them have been developed to provide a
level of transparency to the Internet transport layer. While simplifying
accessibility, a considerable amount of protocol overhead is introduced. WAP
is
gaining a lot of attention in media and engineering circles as a panacea for
Wireless traffic, but has little value as part of an efficiently structured
high
volume, small packet Wireless data transport for WAN/LAN strategies. On the
Internet, where use is practically free, package efficiency is much less of a
consideration.
Optimizing the manner in which data is managed and transported is extremely
important to overall Wireless network functionality and wireless economics in
general. Inside a WAN/LAN network architecture, especially when the WAN is a
public Wireless network, overall efficiency will be an essential component of
long-
term viability and profitability. This is especially true when the network is
packet
switched. The LAN portion of the network architecture is also a beneficiary of
efficient data management. There are a number of fundamental ways to promote
overall data efficiency that enhance both the economics and operational
characteristics of the overall network architecture. Typical protocol
structures,
data structures, and in particular data organization for both WAN and LAN
strategies offer little opportunity to increase the efficiency of AMR data
management and transport.
14
CA 02308819 2000-OS-15
WAN protocols in particular are rigidly defined. The only opportunity for
optimization is in the management and organization of the data through the LAN
and at the WAN interface to increase the informational value of data packets
that
are sent in the protocol wrapper used by the WAN. The difficulties imposed by
uniform protocol standards for disparate information requirements, create
challenges to the optimization of WAN/LANs for AMR purposes. The even
greater challenge is to deliver optimal AMR performance, while also providing
an
optimal structure for the organization and transport of other value added
services
information.
AMR Packet Structure and Data Organization
Constructing an optimal Packet Structure and Data Organization for AMR
purposes offers a unique opportunity to dramatically increase the efficiency
of
AMR transactions within WAN/LAN architectures. This will require the use of a
different set of organizing theories and principles that more reliably support
the
real economics of wireless transactions. At the most fundamental level, all
wireless transaction economics could be described by this formula:
(Transaction Value) - (Transaction Cost) _ (Transaction Margin)
(Where Transaction Value is the Informational value of the transaction and
data
packet size equates to the Transaction Cost of moving the data over the Public
Network.)
Conceptually, all transactions can be valued, whether for AMR or any other
type
of wireless services looking to displace a wired service or to augment a wired
service. Given the huge difference in cost between wired and wireless
transport,
either Transaction Values must increase or Transaction Costs must be
reduced to provide an economic rationale for adoption. Of the two scenarios,
for
virtually any application, the most secure way to drive adoption is to provide
a
method for reducing cost. I.e. cost displacement. Therefore the most practical
method of enhancing Transaction Margin is best pursued through the reduction
of Transaction Cost. This can be best described as the formula:
(Information Content) divided by (Data Overhead) _ (Transaction Efficiency)
(Where Information Content represents the desired portion of the data package
and Data Overhead represents the balance of the data package sent over the
network.)
A review of Transaction Efficiency across a broad set of Transaction
requirements for Wireless Service quickly uncovers opportunities for a
dramatic
increase in Transaction Efficiency, simply by organizing and structuring the
data into two primary categories: Static and Dynamic.
Static Data is the descriptor portion of the data package that describes the
Dynamic Data. These descriptors are items such as normal protocol overheads
and other associative data that get bundled with Dynamic Data. Examples of
CA 02308819 2000-OS-15
Static Data from an AMR context would be Customer ID, Meter ID, LAN ID,
WAN ID, Time/Date Stamp, etc.
Dynamic Data is the data that changes over time, and typically represents the
portion of any data package that truly represents the informational content of
the
Transaction. Examples of Dynamic Data from an AMR context would be the
meter reading itself. Each LAN device has this Static Data (Customer ID, Meter
ID, LAN ID, WAN ID, Time/Date Stamp, etc.) stored in memory. The Static Data
is then attached to the Dynamic Data portion (the current meter reading
value),
and then inserted into a protocol wrapper for transport through the LAN to a
WAN
concentrator.
16
CA 02308819 2000-OS-15
Typical scheduled AMR transactions of this type passed from LAN based meters
through WAN concentrators result in packet sizes that range from 60 to 300
bytes depending on packet construction, protocol overhead, etc. Movements of
information in packets of this size do not appreciably dilute LAN economics,
but
rather rely on a reasonable level of LAN performance. However, since a single
residential meter reading can be represented in 4 bytes, the efficiencies of
these
transactions over a WAN are 1.66% (300 bytes) and 6.66% (60 bytes)
respectively. Different types of transactions will have wide variations in
their
Static and Dynamic Data requirements, independent of LAN or WAN protocol
overheads, resulting in wide ranging Transaction Efficiencies.
Organizational Structure and Data Packaging for Transaction Efficiency
The key to efficient management of transactions over a WAN/LAN network is the
precise way in which data is organized and structured at the WAN/LAN
interface.
Certain types of transactions lend themselves well to optimization. In
particular,
scheduled transactions such as AMR can be repackaged to achieve Transaction
Efficiencies of up to 98% in WAN/LAN strategies with LAN to Wan Ratios of 200
to 1 or more. In general, all meter services fall into this scheduled
category. This
type of packaging also lends itself well to a variety of other Assigned
Services
while still providing full functionality for Unassigned or Custom Services.
Transaction Services Data Management (TSDM)
Assigned Services, such as Scheduled Meter Reading benefit greatly from
TSDM. Primary services such as reading utility meters (electric, gas and water
meters) require delivery of consumption readings on a predefined schedule.
Custom Services are those transactions that require immediate processing and
delivery, such as demand readings for metering, remote alarms, credit card
verification, or other transactions that are time/value sensitive that must be
managed separately from Assigned Services.
Utility providers and most Commercial Enterprises construct their Enterprise
Databases, Billing Systems and Customer Information Systems to Organize a
considerable amount of Static Data, such as the Customer's name, phone
number, address, Meter ID, etc. Dynamic Data Transactions such as Monthly
consumption information, Service charges, Bill payments, etc. are inserted
into
this structure on an as required basis. Typically these occur once per month
for
a specific client's billing purposes. The data is organized within a vast
table
structure that defines the locations for insertion of Dynamic Data values
within
the Static Table Structure. Every Enterprise organizes and manages their data
in this fashion. This structure holds true for Utilities, Banking, and any
other
Commercial Enterprise.
17
CA 02308819 2000-OS-15
To insert Dynamic Data into this Static Table Structure, a transfer table must
be populated with the Dynamic Data that will load this information in a
precise
field within the Database. All Transactions, whether managed over wired or
wireless mechanisms ultimately must be repackaged in a manner that allows this
type of insertion. This activity is managed by a Transaction Server, which
collects individual Dynamic Data packages, takes them out of their individual
protocol wrappers, and repackages the data into defined fields within the
transfer table for insertion into the enterprise's database.
People within the Enterprise can then build queries that allow them to access
the
precise information they require from the database for their various purposes.
Whether Billing or Customer Services, etc. this is also accomplished through
transfer tables constructed to suit the unique requirements of the
applications.
The data structure is logically organized to maximize its usefulness for
enterprise
activities and transfer tables are logically constructed to maximize the
insertion
and retrieval of information by a wide variety of users.
18
CA 02308819 2000-OS-15
Assigned Services Table (AST) Structure
Today, typical LAN/WAN strategies send individual transactions through the
network to a transaction server, where they are repackaged in a transfer file
for
insertion into the Enterprise Databases. This requires the individual
transactions
to have enough descriptive Static Data included with the Dynamic Data in the
individual transaction data packages to facilitate population and delivery of
the
appropriate transfer files to the Enterprise Databases. Using an AST that
resides at the WAN/LAN interface and a Duplicate of the AST resident at the
Transaction Server dramatically reduces Static Data overhead, while still
providing all the functionality required for the management of Transaction
services, dramatically increasing Transaction Efficiency.
The diagram below depicts a typical Assigned Services Table. (Table 1)
LAN GPS Device EM GM WM Assigned Custom
Services
ID# ID# Data Data Data Services
1 YYYYYY XXXXXXX YES YES NIA NIA Phone-Cut
X Detection
2 YYYYYY XXXXXXX YES N/A YES NIA NIA
X
3 YYYYYY XXXXXXX YES YES N/A Demand Security
Alarm
X
Profiling
... ... ... ... ... ... N/A NIA
239 YYYYYY XXXXXXX YES NIA YES NIA Flood Alarm
X
240 YYYYYY XXXXXXX YES NIA YES NIA NIA
X
LAN ID# = LAN Address assigned by the WANILAN concentrator
Device ID# = Unique Address of LAN Device GPS = GPS coordinates of LAN ID
EM = ELECTRIC METER Data GM = GAS METER Data WM = WATER METER Data
Assigned Services = Other Assigned Services Data
Custom Services = Custom Services Data
Having direct knowledge of the table structure at the transaction server
level,
allows information to be packaged sequentially. Through the table above, it is
known which fields are active and populated, the structure and purpose of the
information in each field. Each column describes a type of information, and
each
row describes the information being collected through the individual devices.
19
CA 02308819 2000-OS-15
When LAN nodes are deployed, they each have a completely unique Device ID.
Once they are attached to a WAN/LAN concentrator, they are assigned a unique
LAN Address. Each service, such as Electric, Gas or Water Meter reading, as
described in Table 1 is assigned a unique 16-bit Service ID. Every message
received from the individual LAN nodes, incorporates the Service ID to
identify
the type of data being received by the WAN/LAN concentrator for incorporation
into the table. In this way over 64,000 individual Assigned Services can be
uniquely described. Service ID's describe the unique characteristics of the
data
in the table field, how the information is packaged, the manner in which it is
to be
repackaged for transport at the Transaction Server, and its ultimate
destination
address. Custom Services also use a similar ID assignment. Custom Service
data packets are sent directly to the Transaction Server for processing. Each
WAN address is a unique 64 Bit address on the network. This WAN address
guarantees that the Transaction Server recognizes individual AST's uniquely.
As an example, assuming that 200 LAN nodes are managed per WAN/LAN
concentrator, and assuming 1,000,000 LAN nodes resident within Electric Meters
deployed, 5000 WAN/LAN concentrators would be required. (5000 WAN/LAN
concentrators X 200 LAN nodes = 1,000,000) Therefore there would be 5000
unique AST's. The Transaction Server would store a copy of each AST in
memory. In Table 1, Assigned LAN ID#'s are stored sequentially 1 -255. The
Device ID column and GPS values column identify the unique LAN device and
location of the LAN device of the assigned addresses. In each row of the LAN
ID
column, a variable set of services is described, that the LAN ID communicates
with locally. All LAN ID's will have an electric meter. Some will have Gas
and/or
Water meters, while some others will additionally manage other services such
as
Phone-Cut Line Detection, Flood monitoring, etc.
Structuring the EM Data columnar data as a sequential string of four byte
packages can then be used to send all electric Meter readings. This string
represents the sequence of addresses 1-XXX with a Service ID header, since
the Transaction Server can place these four byte packages back into their
table
positions into the already known table. In circumstances where the electric
meter
data is not available for the LAN ID, through the failure of the LAN device,
the
meter or other difficulties, Null Characters are sent as placeholders to
ensure the
integrity of the packet sequence. In this way 200, 4 byte meter readings can
be
sent with approximately 10 bytes of packet overhead over the WAN, achieving
Transaction Efficiencies of (800/810) over 98% in comparison to normal
efficiencies of 1.5% to 6.6%.
CA 02308819 2000-OS-15
Gas Meter readings are sent in a similar fashion. In the case of GM Data
however, there are locations shown in Table 1 that do not have Gas Meters
present at indeterminate locations. The table identifies the locations as NlA.
Since there is no data associated with these locations, it is unnecessary to
send
Null Characters to fill data gaps. The GM Data in the column is strung
together
sequentially, ignoring the NIA fields in the GM Data table column, putting in
Null
Characters only when data is missing. In a circumstance where 120 Gas Meters
are co-located at LAN ID locations, a sequential string is built of 120, 4byte
packets yielding a Transaction Efficiency of over 97%(480/490).
All Assigned Services are assigned columns in the table structure. All packets
received by the WAN/LAN concentrator are opened, and have their Service ID
compared to those Service ID's in the table. If the Service ID is the same as
one in the table, then the field corresponding to the LAN ID and Service ID is
populated with the Dynamic Data in the package. If the Service ID is not
contained in the table, then the package is considered to be a Custom Service
and is sent to the Transaction Server for processing.
At the Transaction Server, packets received over the Internet are opened.
Each package contains the unique WAN address that allows the transaction
server to find the associated AST located in memory. The Transaction Server
then compares the Service ID contained in the packet to the Service ID's of
the
appropriate AST to determine whether the data contained is to be processed as
an Assigned Service or Custom Service. If it is for Custom Service, then the
information is repackaged in the appropriate Transfer file format and sent to
its
home location. If the Service ID corresponds to one contained in the AST, then
the information is placed in the appropriate column of the AST. Since the
Transaction Server knows that the 4 byte packets contained are to be put into
the
fields in the exact sequence received, it can use the now complete AST
information to build a Transfer files that is/are suitable for insertion at
the
Enterprise Database locally, or repackage the information for Transport by
TCPIIP protocol to anywhere with an Internet connection without significantly
altering the Transaction Economics.
21
