WO2001087279A2

WO2001087279A2 - Self-organizing network architecture

Info

Publication number: WO2001087279A2
Application number: PCT/CA2001/000691
Authority: WO
Inventors: Roderick Michael Johnson; Wilfred Mueller
Original assignee: Smartsynch Ltd.
Priority date: 2000-05-15
Filing date: 2001-05-15
Publication date: 2001-11-22
Also published as: WO2001087279A3; AU2001259981A1; CA2308819A1

Abstract

A method for forming a routing table for an automated meter reading system in which a plurality of meters are assigned to a primary concentrator that in turn forms part of a wide area network for transmitting data collected from the meters to a central location for further processing. An attempt is first made to transmit data from the primary concentrator to each meter assigned to the primary concentrator by starting at the minimum possible signal strength or a signal strength calculated based upon estimated distance. If the transmission is not successful, the signal strength is increased until the data is transmitted successfully to the meter or the signal strength can be increased no further. If the data is successfully transmitted to the meter, the meter is registered in the routing table with an indication that direct transmission between the primary concentrator and the meter is possible. If any meter remains unregistered after registration from the primary concentrator is attempted, then an attempt is made to register unregistered meters using a previously registered meter as a secondary concentrator. A meter registered using a secondary concentrator is listed in the routing table with an indication that the secondary concentrator acts as a relay for transmission of data between the primary concentrator and the meter. Registered meters may be selected as possible secondary concentrators based upon distance from the unregistered meter determined from GPS coordinates. Registered meters on or near a line extending from the primary concentrator to the unregistered meter may be avoided in the selection of possible secondary concentrators. The method may be applied to the selection of tertiary and deeper levels of concentrators, wherein more than one hop is used to transmit data between the primary concentrator and a meter.

Description

SELF-ORGANIZING NETWORK ARCHITECTURE

Field

The present invention relates to a computer method and system for creating a routing table for a communications network and, more particularly, to a method and system for creating a routing table for a wireless automated meter reading system.

Background

One form of large-scale wireless automated meter reading ("AMR") system that has been proposed uses both a wireless local area network ("LAN) and a wide area network ("WAN"), which is not necessarily wireless, to obtain readings from meters and transmit the readings to a central server computer for billing and other purposes. Each meter is provided with a wireless LAN transceiver. The WAN is used to collect meter readings from the wireless LAN transceivers and transmit them on to the central server. The WAN is composed of a number of transceivers that need to be placed so that each LAN transceiver can reliably communicate with at least one WAN transceiver.

Because of obstructions such as vegetation and buildings, a LAN transceiver at a meter may not necessarily be able to communicate with the nearest or any of the WAN transceivers if the WAN transceivers are simply spread out uniformly over the area in which the meters are located. In the past, skilled technicians have had to place the WAN transceivers in locations based upon studies of the local RF environment. Such studies and the time required by skilled technicians to set up the AMR system adds significantly to the cost of a wireless AMR systems, so that in many cases such systems have not been feasible due to the low value of meter readings.

A better method is needed for setting up and maintaining AMR systems in which the placement of the WAN transceivers in not critical and which can be done in an automated or semi-automated manner. Summary

In one aspect of the invention, a method is provided for forming a routing table for an automated meter reading system in which a plurality of meters are assigned to a primary concentrator that in turn forms part of a wide area network for transmitting data collected from the meters to a central location for further processing. For each meter assigned to the primary concentrator, an attempt is made to transmit data from the primary concentrator to the meter, starting at an initial signal strength and increasing the signal strength until the data is transmitted successfully to the meter or the signal strength can be increased no further. If the data was successfully transmitted to the meter, the meter is registered in the routing table with an indication that direct transmission between the primary concentrator and the meter is possible. If any meter remains unregistered, a registered meter is selected from a list of possible secondary concentrators comprised of those meters registered in the routing table with an indication that direct transmission between the primary concentrator and the meter is possible. An attempt is made to transmit data from the selected registered meter to the unregistered meter, starting at an initial signal strength and increasing the signal strength until the data is transmitted successfully or the signal strength can be increased no further. If data cannot be transmitted to the unregistered meter by the selected registered meter, then another registered meter is selected from the list of possible secondary concentrators and an attempt is to made to transmit data to the unregistered meter. This process continues until a registered meter is found that can transmit data to the unregistered meter or until all such registered meters on the list of possible secondary concentrators have been tried. If data was successfully transmitted to the unregistered meter by a selected registered meter, the previously unregistered meter is registered in the routing table with an indication that transmission between the primary concentrator and the previously unregistered meter is possible by using the selected registered meter as a secondary concentrator.

Optionally, possible secondary concentrators may be selected in order of distance to the unregistered meter. Further, to avoid obstructions whose presence can be inferred, possible secondary concentrators may be excluded or tried last if they are on or near a line extending from the primary concentrator to the unregistered meter. The method may be extended to further levels. For example, if no secondary concentrator can be found for a particular unregistered meter, meters that were registered by secondary concentrators can be tried for use as a tertiary concentrator. If successful, the meter that was not registered by a secondary concentrator would be reached by three hops, using meters acting as a secondary and a tertiary concentrator as relays.

Brief Description of Drawings

Figure 1 is a block diagram that shows a portion of a WAN and a LAN that may be organized by the system and method described herein.

Figure 2 is a block diagram illustrating a portion of one domain of LAN nodes associated with a WAN node of the WAN of Figure 1.

Detailed Description

An exemplary AMR system that can be organized in the manner proposed by the inventors is shown schematically in Figure 1 and is indicated by reference numeral 10. The exemplary system 10 includes a central server computer 12, a WAN comprised of WAN nodes, three of which labeled with reference numerals 14, 16, and 18 are shown as squares in Figure 1, and a LAN comprised of LAN nodes, a few of which are shown as circles in Figure 1. Representative LAN nodes are labeled with reference numerals 20, 22, 24, 26, and 28. The meters to be read may be located at LAN nodes or at WAN nodes, although in the following discussion it will be assumed that the meters are only located at LAN nodes. In a typical installation for the automated reading of residential electrical power meters there might be as many as or more 1,000,000 LAN nodes and 5,000 WAN nodes. A ratio of one WAN node to a maximum of 200 to 240 or so LAN nodes is considered by the inventors to be preferable, but other ratios could clearly be used.

The server 12 may be connected to the WAN nodes 14, 16, 18 by any form of wired or wireless network connection that provides reliable communication. The LAN, however, is assumed to be a short-range wireless communications network in which not all LAN nodes can necessarily communicate reliably with each other. Each WAN node 14, 16, 18, in addition to the hardware needed to communicate with the server 12, is provided with hardware, such as a LAN transceiver, to communicate wirelessly with nearby LAN nodes.

Two necessary features of the LAN nodes are that (1) each node must be capable of relaying a transmission received by it on to nearby LAN nodes and (2) the signal strength used by a LAN node's transceiver can be varied by external commands received via the LAN. The latter feature is also necessary in the LAN node transceiver located at the WAN node. These hardware features are commonly found in commercially available LAN transceiver boards. It will be assumed that the signal strength used by the LAN transceivers used in the exemplary system can be set by external command to any one of 256 discrete values, so that an indication of the signal strength may be stored in one byte.

The LAN nodes and WAN nodes each contain, in addition to the necessary transceiver or transceivers, the necessary hardware to carry out the functions described below. That hardware can be provided by microcontrollers having onboard memory in a manner well known to those skilled in the art.

Ultimately, for the AMR system 10 to operate, the WAN nodes must be located so that each LAN node can reliably communicate with at least one WAN node. One way to accomplish this is to undertake a comprehensive study of the radio frequency environment in which the LAN nodes are located. This would likely require testing by skilled personnel prior to the installation of the WAN nodes. However, even if such a study were conducted, the radio frequency environment could change with time. For example, new buildings could be built that interfere with RF transmission between the LAN nodes and their assigned WAN node.

Rather than studying the RF environment in a conventional manner to determine a suitable location for each WAN node, the inventors propose to use the relaying capability and variable signal strength capability of the LAN transceivers to provide a system that adapts dynamically to the local RF environment. Figure 2 illustrates how this may be done. Figure 2 shows a portion of Figure 1 that includes WAN node 14 and a number of nearby LAN nodes, each LAN node shown as a circle containing a unique number ranging from 0x01 to 0x18 ("Ox" preceding a number indicates that the number is in hexadecimal notation). The LAN nodes will be referred to by those hexadecimal numbers in the following discussion, but in Figure 1 the "Ox" is omitted.

We begin by assuming that the LAN nodes have been placed at the locations of all meters in the area and that the WAN node 14 has been installed at a reasonably central location that would under ideal conditions allow direct wireless communication between the WAN node 14 and each of the LAN nodes shown in Figure 2. During the installation of those LAN nodes and the WAN node 14, GPS coordinates of each LAN node and of the WAN node 14 are obtained for use in the process that will now be described. When a LAN node or the WAN node 14 is installed, its GPS coordinates and a password are written into its onboard memory and its LAN transceiver is turned on and configured to listen for and respond only to transmissions containing its GPS coordinates and password. Also, the WAN hardware of the WAN node 14 is turned on as soon as its GPS coordinates are known so that it can communicate with the central server 12. The WAN node 14 is shown in Figure 2 as connected to the server 12 by a wired connection 30, but any form of reliable communication network connection may be used. The details of the hardware used to provide connection 30 and the server 12 is beyond the scope of this invention.

As part of the installation, the GPS coordinates of each LAN node and the WAN node 14 are saved by the installer and subsequently downloaded into the server 12. Additional data collected by the installer may also be downloaded into the server 12, such as each LAN transceiver's unique media access control ("MAC") layer hardware address and physical location (such as its street address).

Once the LAN nodes and the WAN node 14 have all been located and their GPS coordinates are loaded into the server 12, the server 12 constructs a routing table for a domain to be associated with the WAN node 14. The routing table contains a line for each LAN node. If the RF environment turns out to be ideal, each LAN node will be within range for reliable communication directly with the WAN node 14 and the routing table entry for each LAN node in the domain will simply comprise data needed for composing a direct transmission to the WAN node 14. The server 12 can then poll a LAN node for meter data simply by sending an appropriate message to the WAN node 14, which in turn will poll the LAN node directly. One advantage of the system proposed by the inventors arises if the RF environment is not ideal or if it changes with time so that direct wireless communication between the WAN node 14 and one or more of the LAN nodes becomes more difficult or impossible.

In the following discussion, the process of establishing a route from the WAN node

14 to a LAN node and adding that route to the routing table is referred to as "registering" the LAN node. Registration of a LAN node is complete when the LAN node is listed in the routing table of WAN node 14 with a path to the WAN nod 14, including possibly one or more other LAN nodes in the path. The relaying of a message for a particular LAN node through another LAN node is referred to as a "hop" in the following discussion. Therefore, the entry for a LAN node in a routing table may show a hop to another LAN node. A LAN node that relays a message is referred to in the following discussion as a "concentrator LAN node". A LAN node that relays data from a meter directly to the WAN node 14 is referred to as a "secondary concentrator LAN node", a LAN node that relays data from a meter to a secondary concentrator LAN node is referred to as a "tertiary concentrator LAN node", and so forth. The WAN node 14 can then be considered as a "primary concentrator" as it is the first level of concentrator for the server 12.

The goal is that each LAN node in the entire AMR system 10 will be registered to a discrete WAN node with as little human intervention as possible using as few hops and as little signal power as possible. In the following discussion, only the WAN node 14 and the

LAN nodes shown surrounding it in Figure 2 will be considered. However, the process described will clearly be applicable to a large AMR system containing many WAN nodes.

The inventors prefer to order the LAN nodes listed in the routing table and to attempt to register the LAN nodes in order of increasing distance from the WAN node 14. Hence, the first step in constructing a routing table is to calculate the distances between the WAN node 14 and each LAN node. The server 12 performs this calculation using the GPS coordinates downloaded into it by the installer.

While not essential, the server 12 also calculates an initial signal strength for each LAN node to be used by the WAN node 12 to attempt to establish commuiήcation with that LAN node. Assuming that the RF environment is ideal, a calculation can then be made, using methods well known to those skilled in the art, of the minimum signal strength expected to be needed to reliably communicate between the WAN node 14 and the LAN node via the LAN. If initial signal strengths are not calculated, the time needed to register the LAN nodes will increase somewhat, as will be evident from the following discussion, because the WAN node will have to start at the lowest power and work its way up.

After the initial signal strengths are calculated, an initial routing table for the domain associated with WAN node 14 is filled with the LAN nodes listed in order of increasing distance of the LAN nodes from WAN node 14 (which is of course proportional to the estimated signal strength needed) and with a unique one byte LAN node address assigned to each LAN node. For example, the LAN node closest to the WAN node could be assigned LAN node address 01, the next closest assigned LAN node address 02, etc. As mentioned above, the LAN node addresses are given in hexadecimal notation, so that up to 255 LAN nodes can be addressed in one domain using a single byte address. The use of a single byte address is preferred by the inventors, but clearly two-byte (or larger) addresses could be used if the designer so desired.

The order of LAN nodes in the table could also be determined in some other manner. For example, the order could be determined so that the LAN nodes are on a spiral path that has several LAN nodes on each turn of the spiral and that originates at the closest LAN node to the WAN node 14 and spirals outward with increasing LAN node address. It should be noted that in a spiral pattern, the radial distances from the WAN node 14 to the successive LAN nodes in the table may not increase monotonically if the constraint that there be a minimum number of LAN nodes per turn is enforced.

The routing table stored in the server 12 for a domain associated with a specific WAN node 14 then includes a LAN node address, GPS coordinates, initial recommended signal strength, and password entry for each LAN node in the domain. The server 12 sends as that table to the WAN node 14. The following is a sample table for WAN 14 and the LAN nodes shown in Figure 2 (GPS coordinates and passwords are not filled in this sample, but are needed for the WAN to communicate with the LAN nodes):

Table 1

Alternatively, if the WAN node 14 has a sufficiently powerful microcontroller and enough memory, the server 12 could simply provide the WAN node with the GPS coordinates and passwords of the LAN nodes to be registered by the WAN node and the WAN node could prepare the initial routing table shown above itself. In any case, routing table shown in Table 1 is stored at the WAN node 14 and additional columns added to store, for each LAN node, a "pass/fail" flag, a bit error rate, the actual acquisition signal strength needed to communicate with the LAN node, and a domain access address, whose purpose will be explained below.

Once the WAN node 14 has received or prepared a routing table such as Table 1, the WAN node 14 attempts to establish communication with each LAN node in routing table, starting with the first LAN node in the routing table and progressing through the routing table sequentially. For each LAN node, the WAN node 14 starts by sending a request to respond to the LAN node using the initial signal strength in the routing table. The request includes the GPS coordinates, password, and the signal strength to be used by the LAN node for its response. The signal strength requested is the same as the signal strength that the WAN node 14 is using to send the request. If the LAN node does not response to a request at the initial signal strength in the routing table, then the WAN node 14 increases the signal strength by steps until the LAN node responds. If the LAN node does respond, the WAN node 14 sets a "pass" flag for the LAN node and stores in its copy of the routing table the bit error rate and signal strength ("acquired signal strength") needed to obtain the response from the LAN node. If a LAN node does not respond even at the highest available signal strength, then a "fail" flag is set in the routing table and the process continues with the next LAN node in the routing table. For example, the following routing table is an example based upon Figure 2 and Table 1 of the routing table stored in the WAN node 14 following completion of its first attempt to register the LAN nodes shown in Figure 2 (the columns for GPS coordinates and passwords have been omitted):

0x18 70 Pass 80

Table 2

After the WAN node 14 has attempted to register all of the LAN nodes in the routing table, it sends a table comprised of the LAN node addresses, bit error rates, and acquired signal strengths for the LAN nodes that were registered back to the server 12. For example, based upon the results shown in Table 2, the following table is returned to the server 12:

Table 3

Because the server 12 has a stored copy of the initial routing table, it can determine by a process of elimination which LAN nodes were not registered (in this example, those having LAN node addresses 0x13, 0x14, and OxOC). The server 12 updates its copy of the routing table with the data in the table returned by the WAN node 14. For example, based upon Table 3, the server 12 will update its routing table for the domain of the WAN node 14 to the following (the columns for GPS coordinates and passwords have been omitted):

Table 4

For each LAN node that did not register, the server 12 selects a registered LAN node to act as a secondary concentrator LAN node for the unregistered LAN node. If more than one LAN node is unregistered, then it is preferable to attempt to assign more than one unregistered LAN node to each secondary concentrator LAN node. [Is this so? Why?]

Many strategies could be used by the server 12 to select secondary concentrator LAN nodes and assign unregistered LAN nodes to them. Those skilled in the art will have no difficulty devising a variety of strategies. For example, the server 12 could select the registered LAN node closest to the unregistered LAN node to act as a secondary concentrator LAN node of the unregistered LAN node. This might work, but it may be preferable to take into account that a line-of-sight transmission from the WAN node 14 to the unregistered LAN node was unsuccessful in eliciting a response from the unregistered LAN node. For that reason it may be preferable to select a registered LAN node that is one of the closest registered LAN nodes to the unregistered LAN node, but not on or near the line-of-sight between the WAN node 14 and the unregistered LAN node. This may be done using the GPS coordinates of the WAN node 14 and the LAN nodes to calculate distances between registered LAN nodes and the unregistered LAN node as well as the angular coordinates of the LAN nodes in a polar coordinate system in which the WAN node 14 is at the origin. In the following discussion, "angular coordinate" should be understood to mean the angular coordinate in a polar coordinate system in which the WAN node 14 is at the origin. The angular coordinates of the nearest registered LAN nodes to the unregistered LAN nodes may then be compared to the angular coordinate of the unregistered LAN node and the registered LAN node that has an angular coordinate that differs the most (or more than a predefined angle) from that of the unregistered LAN node could be selected as the secondary concentrator LAN node for the unregistered LAN node.

In the example shown in Figure 2 and the Tables above, the server 12 might select LAN node OxOA rather than LAN node OxOB as a secondary concentrator LAN node for LAN node 0x13, because LAN node OxOA is the closest registered LAN node that is not close to the line-of-sight between WAN node 14 and LAN node 0x13. For example, LAN node OxOB, which is closer to LAN node 0x13 than LAN node OxOA, is close to that line of sight, suggesting that there is some obstruction between LAN node OxOB and LAN node 0x13. For example, an obstruction 32 is shown in Figure 2, but the server 12 has no direct knowledge of it. We will assume, for the purposes of illustration, that the obstruction 32 blocks direct RF communication between LAN node OxOB and LAN nodes 0x13, 0x14, and OxOC, as well as between LAN node OxOC and LAN node 0x14 and between LAN node 0x15 and LAN node 0x14. In other words, we, who have a viewpoint that the server 12 does not have, know that of the LAN nodes that are closest, only LAN nodes OxOD and 0x15 can communicate with unregistered LAN node OxOC, only LAN node OxOA can communicate with LAN node 0x13, and only unregistered LAN node 0x13 can communicate with LAN node 0x14 (none of the initially registered LANs being capable of so doing). If more than one LAN node remains unregistered, the server 12 should preferably attempt to assign more than one unregistered LAN node to each second concentrator LAN node. One method for doing so is for the server 12, when considering which of the registered LAN nodes that are nearest to an unregistered LAN node to select as a secondary concentrator LAN node, to give preference to any registered LAN node that is already designated as a secondary concentrator LAN node, provided that it is not near the line-of- sight between the WAN node 14 and the unregistered LAN node that is being assigned to a secondary concentrator LAN node and provided that the calculated signal strength necessary for the already registered secondary concentrator LAN node is within some predefined fraction of the maximum signal strength possible. If there is no nearby registered LAN node that is already designated as secondary concentrator LAN node, the server may then use the process described above to select another secondary concentrator LAN node.

In the example discussed here, the closest registered LAN nodes to unregistered LAN node 0x14 appear to be LAN nodes OxOA, OxOB, 0x04, 0x15, and OxOD. Applying the method suggested above, the server 12 might select LAN node OxOA as a secondary concentrator LAN node for LAN node 0x14 even though other LAN nodes are closer, because LAN node OxOA has already been designated as a secondary concentrator LAN node. On the other hand, LAN node OxOA might be considered too far from OxOA and LAN node OxOB might be selected as a second secondary concentrator LAN node for LAN node

0x14. LAN node 0x04 might not be selected because it is near the line-of-sight between the

WAN node 14 and LAN node 0x14. Note that server 12 does not "know" that obstruction 32 blocks LAN nodes OxOA, OxOB, OxOD, and 0x15 from registering LAN node 0x14. For purposes of illustration, we assume that LAN node OxOB is made a second secondary concentrator LAN node and assigned to attempt to register LAN node 0x14.

As to unregistered LAN node OxOC, server 12 may attempt to assign one of the two secondary concentrator LAN nodes OxOA and OxOB to the task of registering OxOC. As secondary concentrator LAN node OxOB is closer to unregistered LAN node OxOC, secondary concentrator LAN node OxOB is assigned to attempt to register unregistered LAN node OxOC. Hence concentrator LAN node OxOB has two unregistered LAN nodes (0x14 and OxOC) to attempt to register. For each registered LAN node selected to act as a secondary concentrator LAN node, the server 12 forms a secondary routing table of unregistered LAN nodes such that each unregistered LAN node appears only in the secondary routing table associated with one secondary concentrator LAN node. Each secondary routing table also includes, for each LAN node in the table, a calculated initial signal strength for the secondary concentrator LAN node to start with in attempting to register that LAN node. The initial signal strength for an unregistered LAN node is calculated by the server from the distance between the secondary concentrator LAN node and the unregistered LAN node.

A sample secondary routing table for LAN node OxOA constructed by the server and based upon the situation shown in Figure 2 is:

Table 5

A sample secondary routing table for LAN node OxOB constructed by the server 12 and based upon the situation shown in Figure 2 is:

Table 6

The server 12 then sends the secondary routing table or tables to the WAN node 14. Each table is sent with the LAN node address of the LAN node that the server 12 has picked to act as the secondary concentrator LAN node for that secondary routing table and with an instruction instructing the WAN node 14 to revise the copy of routing table stored by the WAN node 14 to reflect the contents of the secondary routing table. The WAN node 14 contacts each of the secondary concentrator LAN nodes and transmits the appropriate list of unregistered LAN node addresses, initial signal strengths, and passwords as well as a command instructing the secondary concentrator LAN nodes to attempt to register each of the unregistered LAN nodes on its list. The secondary concentrator LAN nodes follow the same process as was followed by the WAN node 14 in the registration of the LAN nodes already registered, namely, increasing the signal strength until each unregistered LAN node in its secondary routing table responds or until the maximum signal strength is reached without a response. Each secondary concentrator LAN node updates the secondary routing table received by it from the WAN node 14 so that the resulting secondary routing table corresponds to that formed by the WAN node 14 in the initial registration of LAN nodes by the WAN node 14, except that the secondary concentrator LAN node's routing table lists only the unregistered LAN nodes in the list sent to it by the WAN node 14 for registration, and contains slots for a LAN node address, initial signal strengths, bit error rates, a pass/fail flag, acquired signal strength, and domain access for each imregistered LAN node on the secondary concentrator LAN node's list.

In the example shown in Figure 2, we have assumed that LAN node OxOA can communicate with LAN node 0x13, but that LAN node OxOB cannot communicate with LAN node 0x14 or LAN node OxOC. Hence the sample updated secondary routing table for LAN node OxOA based upon the situation shown in Figure 2 is:

Table 7

and the sample updated secondary routing table for LAN node OxOB based upon the situation shown in Figure 2 is:

Table 8

After a secondary concentrator LAN node has completed its attempt to register the unregistered LAN nodes in its list, it returns a copy of its updated routing table to the WAN node 14. For each previously unregistered LAN node that the secondary concentrator LAN node registers, the secondary concentrator LAN node has filled in the appropriate slots in its routing table. The domain access slot is filled in with the LAN node address of the secondary concentrator LAN node. The WAN node 14 in turn revises its routing table. The acquired signal strength for LAN nodes registered by the secondary concentrator LAN node are entered as the acquired signal strength for the secondary concentrator LAN node itself, because the WAN node 14 will use that signal strength to send transmissions to all of the LAN nodes registered by the secondary concentrator LAN node via the secondary concentrator LAN node. Also, the domain access slot is filled in with the LAN node address of the secondary concentrator LAN node, to indicate that the newly registered LAN node is accessed via the secondary concentrator LAN node. The routing table of the WAN node 14 based upon the situation shown in Figure 2 will now be (changes from Table 2 are in bold type):

Table 9

The WAN node 14 also forwards an abbreviated version of the secondary concentrator LAN node's routing table to the server 12 that contains the LAN node addresses of the LAN nodes registered by the secondary concentrator LAN node, the bit error rates, and the acquired signal strengths. An exemplary abbreviated secondary concentrator LAN node's routing table based upon the situation shown in Figure 2 is:

Table 10

The server 12 then updates its routing table according. For example, in the example, Table 4 would be updated to:

Table 11

For each LAN node that the secondary concentrator LAN node or nodes did not register, the server 12 may select an alternative secondary concentrator LAN node using the same process described above or may select a LAN node registered by a secondary concentrator LAN node to act as a tertiary concentrator LAN node for the unregistered LAN node. The server 12 may be configured to attempt one or more alternative secondary concentrators for each LAN node that remains unregistered or may immediately proceed to select a tertiary concentrator LAN node for each unregistered LAN node. The inventors consider that it preferable to minimize the latency involved in polling LAN nodes by attempting to use secondary concentrator LAN nodes wherever possible. The same process as described above may be used to selecting alternative secondary concentrator LAN nodes by simply not selecting as secondary concentrator LAN nodes those LAN nodes that have already attempted to register a particular unregistered LAN node. For example, LAN node OxOB will not be used as a secondary concentrator LAN node to attempt to register unregistered LAN node 14. In addition, the person designing the strategy for selecting secondary concentrator LAN nodes may wish to add a preference for selecting a new secondary concentrator LAN node that has an angular coordinate that differs as much as possible from the angular coordinate of the secondary concentrator LAN node or nodes that were unsuccessful in registering the unregistered LAN node. For this reason, it may be advantageous for the server 12 to keep track of LAN nodes that did not register a particular unregistered LAN node.

In the example shown in Figure 2, LAN nodes 0x14 and OxOC remain unregistered. Using the strategy suggested above, the server 12 would attempt to use registered LAN node OxOD as a secondary concentrator LAN node to register LAN nodes 0x14 and OxOC. LAN node 0x04 would not be used as it is close to the line-of-sight from the WAN node 14 to both unregistered LAN nodes. Of course, the server 12 does not know that obstruction 32 blocks communication between LAN node OxOD and 0x14, so LAN node OxOD will only succeed in registering LAN node OxOC. The appropriate tables will be sent and updated as above, resulting in the routing table on server 12 for the domain for WAN node 14 that might look like (added route to LAN node OxOC shown in bold):

Table 12

As LAN node 0x14 will remain unregistered, server 12 will try using LAN node 0x15 as a secondary concentrator LAN node to register LAN node 0x14, but that also will fail due to obstruction 32. Having exhausted all possible registered LAN nodes as secondary concentrator LAN nodes for LAN node 0x14, the server 12 could report this to supervisory personnel for manual intervention, attempt to register LAN node 0x14 using another WAN node, or attempt to register LAN node 0x14 using a tertiary concentrator LAN node. The inventors consider that it is preferable to attempt to find a tertiary concentrator LAN node before trying other alternatives.

The process for selecting tertiary concentrator LAN nodes may be similar to that used to select secondary concentrator LAN nodes, except that only LAN nodes that were registered by a secondary concentrator LAN node are considered as candidates and candidate LAN nodes that are not on or near a line-of-sight from a secondary concentrator LAN node to the unregistered LAN node nor on a line-of-sight from the WAN node to the unregistered LAN node may be given preference. The latter criterion is suggested as the failure to register the unregistered LAN node directly from the WAN node or from any secondary concentrator LAN node suggests that there must be an intervening obstruction. For each registered LAN node selected to act as a tertiary concentrator LAN node, the server 12 forms a tertiary routing table of unregistered LAN nodes, such that each unregistered LAN node appears in only in the tertiary routing table associated with one tertiary concentrator LAN node. Each tertiary routing table also includes, for each LAN node in the table, an initial signal strength for the tertiary concentrator LAN node to start with in attempting to register that LAN node. The initial signal strength for an unregistered LAN node is calculated by the server 12 from the distance between the tertiary concentrator LAN node and the unregistered LAN node. The attempted registration of unregistered LAN nodes by a tertiary concentrator LAN node and the reporting back to the secondary concentrator LAN node, the WAN node, and the server proceeds in a manner analogous to that used for a secondary concentrator LAN node. The domain access slot in the tertiary concentrator LAN node's routing table for each LAN node that is registered by the tertiary concentrator LAN node is filled in with the LAN node address of the secondary concentrator LAN node to indicate that transmissions to the newly registered LAN node pass through both the secondary and tertiary concentrator LAN nodes.

In the example shown in Figure 2, the only possible tertiary concentrator LAN node is

LAN node 0x13 because LAN node OxOC is near the line-of-sight between WAN node 14 and unregistered LAN node 14, suggesting that there is an obstruction (which we can see is obstruction 32). As we assumed that communication is possible between LAN node 0x13 and LAN node 0x14, the final routing table for the domain for WAN node 14 will be (added data is in bold):

Table 13

Table 13 is now complete and all LAN nodes in the domain for WAN node 14 have been registered.

The process described above for registering LAN nodes may clearly be extended to further levels of concentrator LAN nodes, if latency is not considered to be a problem. For example, quaternary concentrator LAN nodes may be used. However, do the increased time needed for the server to poll LAN nodes that are several hops from the WAN, it is preferable to design the process for selecting concentrator LAN nodes so that several strategies are used to attempt to find successful secondary concentrator LAN nodes before attempting to find tertiary or quaternary concentrator LAN nodes.

The process described above can be used for maintaining the system 10. For example, suppose a building 34 erected is that prevents direct communication between LAN node 0x09 and WAN node 14. Server 12 would not "know" that the problem is a building when it tries to poll LAN node 0x09 for data and obtains no response. Therefore, server 12 would essentially go back to the point in the registration process described above at which it first asked WAN node 14 to register the LAN nodes, except that this time it would request WAN node 14 to register only LAN node 0x09 at the next higher signal strength above that presently in the routing table for LAN node 0x09. The registration process would then proceed as before. Since we have assumed that no signal strength would succeed, server 12 would next check its list of concentrator LAN nodes to determine whether there is already a concentrator LAN node within range of LAN node 0x09 and not on or near a line-of-sight between WAN node 14 and LAN node 0x09. In this case, if would find that secondary concentrator LAN node OxOA satisfies those criteria and would proceed as above to cause WAN node 14 to issue the appropriate instructions to secondary concentrator LAN node

OxOA to attempt to register LAN node 0x09. Of course, if secondary concentrator LAN node

OxOA then failed to register LAN node 0x09, the server 12 would attempt to find another registered LAN node near to LAN node 0x09 that could act as a secondary concentrator LAN node.

While the method disclosed above has been presented as a method for use with meters organized into a wireless LAN, it will be clear to those skilled in the art that the method may be applied to communications systems in which the LAN nodes are devices other than meters and in systems in which the LAN nodes communicate among themselves and with the WAN by other means. For example, they might communicate using power line carrier modulation.

Claims

Claims:

1. A method for forming a routing table for an automated meter reading system in which a plurality of meters are assigned to a primary concentrator that in turn forms part of a wide area network for transmitting data collected from the meters to a central location for further processing, the method comprising:

for each meter assigned to the primary concentrator,

attempting to transmit data from the primary concentrator to the meter, starting at an initial signal strength and increasing the signal strength until the data is transmitted successfully to the meter or the signal strength can be increased no further, and

if the data was successfully transmitted to the meter, registering the meter in the routing table with an indication that direct transmission between the primary concentrator and the meter is possible;

and then for any meter remaining unregistered,

selecting a registered meter from a list of possible secondary concentrators comprised of those meters registered in the routing table with an indication that direct transmission between the primary concentrator and the meter is possible,

attempting to transmit data from the selected registered meter to the unregistered meter, starting at an initial signal strength and increasing the signal strength until the data is transmitted successfully or the signal strength can be increased no further,

if data cannot be transmitted to the unregistered meter by the selected registered meter, continuing to select registered meters from the list of possible secondary concentrators and attempt to transmit data until a registered meter is found that can transmit data to the unregistered meter or until all such registered meters on the list of possible secondary concentrators have been tried, and if data was successfully transmitted to the unregistered meter by the selected registered meter, registering the previously unregistered meter in the routing table with an indication that transmission between the primary concentrator and the previously unregistered meter is possible by using the selected registered meter as a secondary concentrator.

2. The method of claim 1, wherein data is transmitted between the primary concentrator and the meters and among the meters by radio frequency transmissions.

3. The method of claim 2, wherein registered meters are selected from the list of possible secondary concentrators in order of increasing distance from the unregistered meter.

4. The method of claim 3, wherein the list of possible secondary concentrators contains all previously registered meters for which direct transmission with the primary concentrator is possible.

5. The method of claim 3 or claim 4, wherein GPS coordinates of the primary concentrator and the meters are used to calculate distances.

6. The method of claim 3, wherein the list of possible secondary concentrators contains all previously registered meters for which direct transmission with the primary concentrator is possible other than those previously registered meters that are on or near a radio frequency path extending from the primary concentrator to the unregistered meter.

7. The method of claim 6, wherein GPS coordinates of the primary concentrator and the meters are used to calculate distances and proximity to a radio frequency path.

8. The method of claim 3, wherein the initial signal strength at which data is transmitted from the primary concentrator is the minimum signal strength at which data can be transmitted from the primary concentrator and the initial signal strength at which data is transmitted from the secondary concentrator is the minimum signal strength at which data can be transmitted from the secondary concentrator, thereby to minimize interference with adjacent radio frequency devices.

9. The method of claim 3, wherein the initial signal strength at which data is transmitted from the primary concentrator is calculated as the minimum theoretical signal strength needed to cover the distance from the primary concentrator to the unregistered meter and the initial signal strength at which data is transmitted from a possible secondary concentrator is calculated as the minimum theoretical signal strength needed to cover the distance from the possible secondary concentrator to the unregistered meter, thereby to minimize interference with adjacent radio frequency devices.

10. The method of claim 9, wherein GPS coordinates of the primary concentrator and the meters are used to calculate distances.

11. The method of any of the preceding claims, wherein each transmission to a meter from the primary concentrator or any previously registered meter contains data representing the signal strength to be used by the meter receiving the transmission to respond to the transmission, the signal strength to be used to respond being equal to the signal strength at which the primary concentrator or previously registered meter transmitted the transmission.