GB2408162A - Means for monitoring the location of a person within a building - Google Patents

Abstract

A means for monitoring the location of a person within a building is disclosed. A person to be monitored is provided with a mobile radio wave transmitter MN. A plurality of radio transceivers SN adapted to be mounted at fixed positions in or around a building monitor the receipt of radio waves emitted by the mobile transmitter MN. The pattern of the signal levels received at the radio transceivers SN are compared with pre-stored patterns of signal levels received by the radio transceivers SN. A location estimator selects the pre-stored pattern which is closest to the measured pattern to provide an estimate of the location of the person. The radio transceivers SN may communicate with gateway transceivers GN which may further communicate with the location estimator by way of radio signals. The pre-stored pattern of signal levels may be obtained by a manual calibration method. Multiple people, each having their own mobile transmitter MN may be independently located using an appropriate multiple access method such as CSMA, where the radio transceivers SN can differentiate between different mobile transmitters MN. The monitoring means may be used for locating firemen inside a burning building.

Description

c ese..

REMOTE LOCATION OF PEOPLE IN A BUILDING

The present invention relates to a method and apparatus for remotely sensing the location of a person within a building or part of a building.

Often it is desirable for a person external to a building, or part of a building, to know the location of a person inside a building or part of a building. For example, for safety reasons it is desirable for the Senior Fire Officer to know the location of firemen inside a burning building.

This is difficult with known technology, which is unable to provide an estimate of the location with sufficient accuracy to be useful on such a small scale. The present invention aims to facilitate a more accurate alternative.

According to one aspect of the invention we provide a method of monitoring the location of a person within a building comprising providing the person with a portable radio wave transmitter (hereinafter called a mobile node), monitoring the receipt of radio waves emitted by the mobile node at a plurality of radio transceivers mounted in fixed positions relative to the building (hereinafter called static nodes), and comparing the pattern of the signal levels of the signals received at the static nodes with pre-stored patterns of signal levels received by the static nodes and selecting in a location estimator the pre-stored pattern or patterns which is closest to the measured pattern to provide an estimate or estimates of the location of the person.

It will be appreciated that the invention can relate to the location of any person or mobile object within a limited area such as a building and its grounds, or a section of a large building or complex.

:. ;:e te. .e .' ..e According to another aspect, multiple people in a building, each with their own mobile node, may have their positions independently estimated.

Preferably a person's mobile node will only transmit a radio signal burst when no other person is transmitting their signal burst. This may be achieved using the well known carrier sense multiple access method.

Preferably the static nodes and the mobile nodes form an adhoc network that uses peer-to-peer communications to forward their measurements of received signal levels from each person's mobile node, usually via a gateway node, to a location estimator.

The invention will now be further described, by way of example only, with reference to the accompanying diagrams: Figure 1 is a schematic plan view of one floor of a building showing the deployment of static nodes, gateway nodes and a mobile node in and around the floor of the building; Figure 2 shows the similarity in construction in one embodiment between a mobile node, a static node and a gateway node; Figure 3 shows a plurality of static nodes deployed internally of a building and the direct radio transmission paths between them; and Figure 4 shows a virtual grid of the type which may be produced if self-calibration is employed by the network of static nodes depicted in Figure 3.

Although the example used relates specifically to the location of firemen within a burning building it will be appreciated that the invention relates :e.e tee; t:- .e to many other situations, for example the location of prisoners or warders in a prison complex.

The mobile node has a processor, some data memory, a radio transceiver (i.e. a transmitter and receiver) and an antenna. Its minimum requirement is to transmit a radio signal at a known frequency and at a known power. The transmissions could be continuous, but in order to conserve battery power, and for multi-mobile node operation, the transmissions are in a burst format. The transmissions may also contain rudimentary information, such as the name or identification number of the fireman carrying the mobile node, and perhaps the seriousness of the situation the fireman is encountering, e.g. a fireman could press one of a set of buttons on his mobile node corresponding to his judgement of the status of the fire. A basic static node consists of an antenna, a radio receiver capable of receiving the transmissions from the mobile node; a means of monitoring the received signal level; an ability to recover any information i.e., the name of the fireman; and a means of transmitting the knowledge of this signal level and other information to other static nodes who will relay the signal until a gateway node is reached. The gateway node is a static node having a radio transceiver that has radio communications with the location estimator. The static nodes have the role of accepting and forwarding information from the mobile node, and in doing so they are providing the information that will enable the location estimator, in the command and control centre (CCC), with the means of knowing the position of the fireman within the building.

In this example the location estimator is external to the building. In other situations, such as the location of a prisoner within a prison complex, the location estimator may be internal to the building, but external to the part of the building in which the location of the prisoner is being monitored.

:. . ce hi; ce tIt.

Figure I is an example showing the ground floor of a burning building, the deployment of static nodes, gateway nodes, and a mobile node (attached to a fireman). The dotted lines inside the building represent internal walls. Only static node 1, static node 2, static node 3, static node 4, and gateway node 1, are able to detect a signal from the mobile node that is above the receiver noise-floor. The data gathered by these four static nodes is relayed to gateway node 1, which passes it on to the location estimator. Static node 3 may send its information to gateway node 1 directly or via static node 4, depending on the range of static node transmissions. The dotted lines in Figure 1 show the radio routes from static node 3 and static node 4 to gateway node 1. Information from static node 1 and static node 2 might be conveyed via mobile node to gateway node 1, without the fireman necessarily being aware of the process. Without the assistance of the mobile node the information from static node 1 and static node 2 might not reach gateway node 1.

However, the information from static node 1, and static node 2, could be relayed around the building via other static nodes until gateway node 2 is reached. The dashed lines in Figure 1 show this route of the radio signals from static node 1 and static node 2 to gateway node 2. Gateway node 2 conveys the information from static node 1 and static node 2 to the location estimator. The location estimator works out the location of the fireman, displaying his position on an electronic screen. The display would be similar to that of Figure 1. Notice that the nodes can be made with off-the-shelf technology, e.g. simple flash memory chips, processor chips, and modems based on the IEEE 802.11b standard. The location estimator would need to be specially built, but this is within the current state-of-the-art.

If there is more than one fireman in the building then only one fireman's mobile node should transmit at any one time. The IEEE 802.11b has a e. :. c:e:: ::.

. . . . particularly suitable method of multiple access to the radio channel for this situation. It is called carrier sense multiple access (CSMA), where each mobile node only transmits a burst of data after sensing that currently there are no other nodes transmitting. Should there be many firemen in the building all the firemen's mobile nodes should transmit frequently using CSMA. Although there will be occasional collisions when more than one mobile node senses that the channel is free and they transmit at the same time, these collisions will be relatively rare. Each mobile node that was involved in a packet collision would wait a short random time before retransmitting. Each transmission burst should be short, but this restriction is not a problem as the packet bursts are of sub- second duration.

If necessary more that one carrier frequency can be used. For example, a group of mobile nodes may be assigned to specific carriers to decrease the probability of packet collisions as the number of nodes to be located increases. The static nodes must be aware of the carrier frequencies used by each mobile mode.

More complex static nodes and mobile nodes would, in addition to providing the location of the firemen, support two-way speech communications between the Senior Fire Officer (SFO) in the CCC, or elsewhere and the fireman in the burning building. The nodes could also be designed to handle a raft of multimedia information that the fireman might need. It may also be a requirement for the fireman to convey multimedia information from inside the building to the CCC. As the static nodes may only be required to communicate multimedia information between the mobile node and the CCC, then an approach for constructing the nodes is for the mobile node and the static node to be identical, and arrange for a separate multimedia unit to be connected to :e;:. ë e.e if.

the mobile node if required. This unit would consist of display facilities, e.g. a head-up video display and a speech loud speaker; and their associated transducers, e.g. a camera and a microphone, with an encoding ability to digitise or decode the multimedia signals as appropriate. Figure 2 shows that the basic node is the static node, the mobile node is a static node interfaced with a multimedia unit if required (which could come with different levels of complexity depending on the requirements), while the gateway node is a static node with a gateway transceiver for communicating with the CCC. The CCC has a gateway node, and in addition a multimedia unit, a location estimator complete with display, plus information relating to other personnel and equipment that is at the site of the fire. The site of the fire, i.e. the burning buildings and the area cordoned off from the public is sometimes referred to as the fireground.

The static node, and perhaps the other nodes, can be further enhanced by giving them additional processing power and memory, a middleware capable of supporting agent technology, an adaptive routing algorithm to re-route signals to the CCC should a static node, mobile node, or gateway node fail, and most important of all, an ability to implement pervasive computing to assist in bringing the fire situation to a quick and

satisfactory conclusion.

The key aim of this application is the remote location and tracking of a person within a building by means of an ad hoc network of nodes. For this system to work a specific type of electromagnetic behaviour needs to be ascertained, and this involves the use of a calibration process, which we will now describe.

In order to realise the mobile node location within a building the building must be calibrated in terms of radio signals prior to the subsequent :: :e:: I: .. ... . . location and tracking of the mobile node. Members of the fire service, usually inspect buildings that pose a significant fire risk, or buildings where a fire could have serious environmental consequences or large loss of life, annually. It is at this time of inspection that the building could be calibrated for the purpose of locating firemen in the building should it be necessary for them to enter the building in the event of it being on fire.

A manual calibration procedure is described below.

The possession of an architectural drawing of the plans of the building(s) is a prerequisite for the calibration process. The plans must show the rooms, corridors, entrances, etc. which we will refer to as enclosures.

There is no need to know either what the building is made of or its contents for the purposes of calibration.

In an actual operational scenario for a fire in a small to moderate size building the static nodes may be located around the outside of the building. The advantage of having static nodes outside the building is that they are not destroyed by the fire, and that the nodes can be removed after the fire for use at another fire. This means that the number of nodes the Fire Service need to have are relatively few, the number depending on how many fires they expect to fight at any one time.

The actual number of the static nodes needed for acceptable location accuracy, and the location of these static nodes is building specific. In order to ascertain the number and position of the static nodes for a specific building we may proceed as follows.

(a) In the calibration process only one mobile node is used. This mobile node, which is a small piece of electronic equipment, e.g. small :: c:. :e:: se . enough to be clipped to a fireman's belt or located in his helmet, say, is placed in a selected position in the building, and remains stationary.

(b) A static node, of similar size to the mobile node, is then carried at a constant convenient height around the building by a person. The route taken by this person is noted, as are the specific points along the route where the person dwells while the received signal levels at the static node are noted. The position of the mobile node is recorded.

(c) The mobile node next moves a short distance, i.e. either remaining within the enclosure or moving to an adjacent one, and remains stationary in this new position. The procedure in (b) is repeated with the static node taking the same path as before around the building.

(d) The procedure described above is repeated as the mobile node moves around the building in discrete positions. Thus for every recorded mobile node location there is a recorded set of static node received signal levels for the route around the building.

(e) The recorded signal levels are examined to identify where the static nodes need to be positioned for a required spatial resolution, identified as the distance between discrete adjacent locations of the mobile node in the calibration process. Then for any of these discrete positions of the mobile node within the building there exists a unique set of static node received signal levels at positions along the route around the building.

From this data we identify how many static nodes are needed, and where they should be positioned, such that the received signal levels at the static nodes is unique to each mobile node location that was used in the calibration process. Notice that for any one of the discrete mobile node positions it is likely that only some static nodes will have values above the receiver noise-floor, i.e. only these nodes will be able to detect the :e;:e sese ceele. .' e. ..e mobile node's transmitted signal. Thus for the particular route used by the static node during the calibration process, the number of static nodes and exactly where they should be deployed can be decided.

If the received static node signal pattern for each discrete mobile node position is not sufficiently different, then another route can be tried in order to enhance the differences in the received static node signal pattern.

An alternative approach to the method described above is as follows.

(a) In the calibration process only one mobile node is used. This mobile node, which is a small piece of electronic equipment, e.g. small enough to be clipped to a fireman's belt or located in his helmet, say, is placed in a selected position in the building, and remains stationary.

(b) A number of static nodes are placed around the building, and then their readings noted for this specific position of the mobile node. The positions of the static nodes are recorded.

(c) The mobile node is moved a short distance within the current enclosure, or into an adjacent enclosure. Observe if the positions of the static nodes give a significantly different set of readings to those previously recorded. If this is true, record the information as before. If this is untrue, move one or more static nodes until a set of received signal levels at the static nodes is obtained that significantly differs from the set in (b). Note this information and the static node locations, and ensure that the uniqueness of the pattern of received signal levels at the static nodes is also true for the previous positions of the mobile node.

: . de: # , (d) This procedure is repeated as the mobile node enters every enclosure within the building. By this procedure a unique set of received signal levels, and the positions of these static nodes where these signal levels were received, is recorded for every discrete position of the mobile node.

This calibration method requires more static nodes than that of the method firstly described. However, a person experienced in this method of calibration is able to have a good idea where the static nodes should be sited and may be able to complete the calibration process quickly.

One method to speed up the calibration process is to employ a radio prediction planning tool, which is used to make predictions of the received signal levels outside the building for each discrete position of the mobile node. The static nodes could then be placed in the locations indicated from the predictions, as this would give the person calibrating the building a useful start as to where to place the static nodes in the calibration method described above. Observe that the current accuracy of indoor radio planning tools is not good enough to be used exclusively for calibrating the building. Hence the measurement calibration process still has to be done.

In the calibration process the discrete positions of the mobile node must be known. If low location accuracy of the mobile node is acceptable, then it might be sufficient for the mobile node to use a few easy-to-define locations in the calibration process, e.g. the mobile node could be positioned at two metres, say, from a door or window in each enclosure.

If greater accuracy of the mobile node's location is required, then a laser light-gun could be used to measure the location of the mobile node from a number of walls in each enclosure.

a.. .:. esee;.e ce.e 46.

When the position of the static nodes has been decided in the calibration process, and when these static nodes will be exclusively positioned outside of the building and therefore not left permanently in place, then these static node positions must be marked in some way so that they can be quickly identified in the event of a fire. As the height of all static nodes must be the same when in use as during calibration, then we need to identify their North/South, East/West coordinates. In some cases a simple paint mark at ground level might suffice, or a marker embedded in the ground could be used, or an electronic tag could be embedded into the ground surface that would be in sleep-mode until electronically switched on by members of the fire service when they arrive at the site of the fire. When the fire brigade arrives at the fireground and locates the markers, the firemen must place the static nodes on poles, or tripods, corresponding to the height they were at during the calibration process, and then locate the static nodes above the markers.

It may be that the building has a very large cross-sectional area, or is made with materials that highly absorb electromagnetic energy, e.g. ultraviolet light absorbing glass windows, or has some other property that prevents the mobile node transmitted signal from being sufficiently large to be detected by the static node receivers, and hence the unique pattern required for mobile node location is not obtained. In these circumstances the transmitted power from the mobile node, and/or its carrier frequency, should be increased and decreased, respectively. (An alternative is to use internally placed nodes, as described later).

When handling high-rise buildings alternative calibration procedures to those described above might be more suitable.

c; teee; (a) The static nodes could be placed on the outside of adjacent high-rise buildings so that they can effect unique mobile node location on every floor. This implies placing static nodes on adjacent buildings, and that might involve their placement at every floor height, or every few floors.

These static nodes would be left permanently in place after the calibration procedure had identified their locations. Placing static nodes on other owner's building implies the cooperation of the owners of these adjacent buildings.

(b) Another approach for achieving the location of mobile nodes on every floor, is to place the static nodes on the outside of the building that that the mobile nodes are in. The nodes would be either attached to the outside of the building or spaced a small distance from the outside wall of the building, say, less than a metre. The calibration process would involve identifying the positions on the outside of the building of the static nodes to give the unique set of readings for each discrete position of the mobile node.

(c) If the building is both tall and has a large cross-sectional area, e. g. like the former World Trade Centre building, then the external static nodes would have to be augmented by many internal static nodes, or internal static nodes may be exclusively deployed for without them the location of every fireman could not be known in every discrete location within the building complex. Thus during the calibration process every discrete mobile node location must result in a unique set of received signal levels, where the set of static nodes may include both the external and the internal static nodes, or only the internal or only the external static nodes. Both the internal and the external static nodes will participate in the ad hoc network for firemen location and, if required, other purposes such as communications.

I. : : .e ë:: :: . be. c. ee. : In the case of a set of buildings being fitted with a firemen location system, e.g. a university campus, or a hospital complex, the nodes would be placed around the buildings, and the calibration procedures implemented (and only supplemented with internal nodes if essential).

The received signal at the nodes would specify the position of a mobile node in any of the buildings. It would be an operational improvement if these static nodes were permanently deployed, forming an ad hoc network used as a firemen location system in the event of a fire, while in the absence of a fire it could be used for other purposes, such as supporting an intranet.

An automatic calibration procedure is now described.

Although the calibration procedure can be done by a non-expert, it does need to be up-dated regularly, for example annually. An automatic calibration procedure that is performed by the ad hoc network is therefore desirable.

Keenan and Motley (reference: J.M. Keenan and A.J. Motley, 'Radio coverage in buildings', British Telecom Technology J. Vole, No. 1, pp.19- 24, Jan.1990) derived an equation of path loss from measurements in a multi-storey office building. Their path loss (PL) equation is: PL = L(v) + 2010gd + nf. af + nw. aw Where L(v) is the PL at a distance of lm from the transmitter for the carrier frequency used, the logarithm is to the base 10, Of, af, nw, and aw, are the number of floors, the attenuation of the radio signal per floor, the number of walls, and the attenuation of the radio signal per wall, respectively, along a straight line between the positions of the transmitter and receiver.

: .F c: : e. ce. :: : We utilise a form of this equation as follows. Each static node in turn sends a signal burst of known power to every other static node, allowing the PL between each static node and every other static node to be measured. Because the map of the building is available, and may be introduced into the processor of every static node, the straight line distances between each node is known, and so is the knowledge of how many floors, internal walls and external walls are along the straight line path. The ad hoc network of nodes has many measurements of PL, as if there are p static nodes then there are (p-1) + (p-2) + ... + 1, PL measurements, e.g. if there are 8 static nodes then there are 28 PL measurements, and this number must be adequate to determine the loss per internal wall, external wall and floor, it being appreciated that the nature of the floors, internal and external walls may differ in the building and each floor and wall must be independently evaluated for the parameters af and aw. For the case of all internal and external walls and floors being the same in the building, nw. aw becomes nwi.awi+nwo.awo, where nwi is the number of internal walls, awl is the attenuation per internal wall, nwo is the number of outside walls, and awo is the attenuation per outside wall. The nodes should be distributed around the building, either completely inside, completely outside, or both inside and outside, the criterion being that from any part of the building a number of static nodes can communicate with each other and that at least one signal path goes through an enclosure. Figure 3 shows the single floor building of Figure 1 with a set of internal static nodes and the numerous straight line paths through the building.

Armed with the attenuation losses for each floor and wall, the ad hoc network of static nodes considers a virtual grid to be placed over each floor. An example of a virtual grid is displayed in Figure 4. The spacing of this grid depends on the accuracy that the system seeks when c: I. . ::: cees e:e ë ee.

locating a mobile node moving about the building, e.g. a spacing of a few metres, or which enclosure might be appropriate. For each place on the virtual grid the ad hoc network computes the straight line distance to every static node, the number of internal and external walls, and the number of floors. It then deduces what would be the received power at each static node if a mobile node was at a specific virtual grid position and was transmitting a packet at a known power. It derives this received power using the PL equation described above as the received power is the transmitted power minus the PL. This procedure is performed for every position in the virtual grid. The building is now calibrated.

The previous comments for locating a fireman carrying a mobile node in a building now apply. A mobile node when entering the building will cause the static nodes to receive its transmissions. The vector of received static node values is compared with their store of vectors corresponding to their calculations for the virtual grid. The closest store vector to the measured vector is the first estimate of the mobile node's position.

The method by which the location of a person is estimated will now be described.

At the beginning of the operation to extinguish a fire in a building where only outside static nodes were used in the calibration procedure then the static nodes are rapidly deployed in the positions identified during the calibration procedure. Should the building have only internal static nodes then they will already be in place. All firemen who enter the building (s) are equipped with a mobile node, and their personal identification number entered into the memory of their mobile node.

Each mobile node will transmit the same constant power signal bursts as used in the calibration process, but these transmissions will not occur at ::: .:. ::: ec es A: * . . . * the same time as previously described. On entering the building(s) some of the static nodes will receive, although not at the same time, the signals transmitted from each mobile node. The static nodes can distinguish the different mobile nodes as their transmitted signals, e.g. in the form of Internet Protocol (IP) packetised data, will identify each fireman uniquely.

Let us now discuss the location procedure for one mobile node more closely, appreciating that this procedure applies for multiple mobile nodes, although only one mobile node will transmit at a time. The unique received signal levels at the static nodes for a discrete mobile node position may be viewed as a vector. The coefficients of this vector: al, a2, ad, ak, are the received signal levels at static nodes: static node 1,static node 2, static node 3, static node k, respectively, there being k static node deployed, where k is an integer number. If a mobile node is exactly at one of the discrete positions used in the calibration procedure, that the mobile node antenna is in the same position, that no changes have been made to either the building, its contents, and to a lesser extent neighbouring buildings, etc then the vector will identify the position of the mobile node.

The number of vectors available to the location estimator corresponds to the number of discrete positions used in the calibration process. If a mobile node enters the building and is not at a mobile node position used in the calibration process, and/or that there have been changes in the content or the position of objects within the building, or that the angle of the mobile node antenna relative to some reference is different, then the location system will decide which of the vectors used in the calibration process most closely corresponds to the static node vector currently obtained from the mobile node transmissions. This closest calibration : Be ce: : He eve cee:: :.

be. e.. ce. ace: vector is deemed, in the basic location method, to be the position of the mobile node.

However, this location estimation based on matching the received vector with the closest fitting stored calibration vector, ie a form of vector quantization has an error, and on the same basis, so will the previous and subsequent location estimates. This situation of having a set of variables with an unknown error can have this error decreased by processing a vector comprising the latest estimated position and the previous Q- 1 estimated positions. The greater the value of Q. the greater is the accuracy of the final estimate of the position of the mobile node, but the greater is the delay in making the estimate. Also, positions of the mobile node considerably back in time are less valuable in making a prediction as to the current position of the mobile node.

Should the location estimator produce a location of a mobile node that is unexpected, e.g. due to a calibration error or a change in the local environment that made the calibration out of date, then a movement predictor (in software) may be used to check the location estimator's location of the mobile node. This predictor takes into consideration the previous positions that the location estimator deemed the mobile node to have had, and predicts the next most probable positions in two or three dimensions. If these predictions are significantly different compared to the position given by the location estimator then the location given by the location estimator is deemed to be unreliable, and is either rejected, or weighted in favour of the movement predictor. The movement predictor continually checks its predictions with those of the location estimator, and only when they are in relatively close agreement will the location estimator adopt the location given by the vectors as the location of the fireman. However, except when the differences between the movement :e;e *e ee.' ed:e .

predictor and the location estimator are very large, e.g. when the two predictors give the fireman's location as more than one enclosure apart, the location estimator would do well to accept the locations given by the quantized vectors. The movement predictor, i.e. a predictor that estimates a mobile node's position based on it previous positions, is a well-established technology and a number of existing algorithms are

suitable.

In addition to using well-known prediction techniques, there are factors that relate to the construction of the building that can enhance the lO estimation of the location of the mobile node. It should be noted that prior to the internal structure of a building collapsing, the SFO will withdraw the firemen as their lives are of primary importance.

Accordingly the disclosed scheme is designed to function with the proviso that the internal walls and floors are in place. The scheme would continue to function in those parts of the building where the floors and internal walls remained, although the location accuracy may be degraded.

Consequently the location estimation procedure can also use information relating to the known positions of the internal floors, ceilings, walls, stairwells, e.g., it can use the knowledge that the fireman cannot be on one floor at one instant and on another an instant later. Thus the location estimator can use the knowledge of the internal construction of the building and how a fireman will move about the building in dealing with the saving of life and dealing with the fire in addition to the vector location quantization and the movement predictor.

The well established neural network technique can also be employed, in isolation or in conjunction with prediction techniques and knowledge of how people move around a building, that are based on the Q vectors.

The disclosed system is able to perform the following tasks. : :

Perform the location and tracking of the movement of a person or multiple persons within a building or part of a building. This location of a person within a building is achieved by each person's mobile node transmitting packetised data at a known power level, and the received signal levels from these mobile node transmissions at a set of static nodes being unique for each person's location within the building. In order to locate multiple people in the same building each mobile node transmits at a different time. This procedure is only made possible by calibrating the building prior to the location estimation.

lO The location of a mobile node is found by processing Q vectors, where each vector has k coefficients corresponding to the number of static nodes and gateway nodes used in the calibration process, and the values of the coefficients are the received signal levels at the static nodes and the gateway nodes. The vectors used are the current values and the previous Q-1 values. There are a number of refinements to achieve an improved location accuracy, some of which have been described above.

Communications are supported between static nodes, mobile nodes, gateway nodes, and the CCC. These communications may be single or multi-hopped over the ad hoc network formed by all types of nodes and the CCC.

For most small size buildings, as found in cities such as London and Paris, static nodes and gateway nodes can be outside the building and not connected to it. However, for tall buildings, say in excess of six to eight stories, it may be necessary to fit some static nodes and gateway nodes to the outside of the building.

For tall buildings that have large cross sectional areas, such as skyscrapers, external static nodes and gateway nodes may be deployed, e but it may be more convenient to position them within the building. The criterion is that this set of static nodes and gateway nodes will function as a single ad hoc network, and will be able to uniquely identify the location of each person carrying a mobile node within the building, using the methodology previously described.

The static nodes and gateway nodes can be located solely within the building, where they permanently reside. The calibration process is done as for the case when all the static nodes and gateway nodes are outside of the building, or it may be use the automatic calibration procedure previously described. The static nodes and gateway nodes in the absence of a fire act as an ad hoc network providing other services, such as a building intranet, and multimedia communications. Although this network may be more costly because of the number of static nodes permanently deployed, the additional services, the reduction in fire insurance premiums, and the likely increase in location accuracy of a mobile node, may make it a good option compared to when the static nodes and the gateway nodes are outside of the building.

The static nodes, mobile nodes and gateway nodes are technologically similar as shown in Figure 2. The mobile node is a static node plus an optional multimedia unit; the gateway node is a static node with a gateway modem. The CCC has a gateway node to communicate with the ad hoc network as well as a multimedia unit.

If all nodes have sufficient intelligence they can combine to act as a pervasive computing network, making recommendations, and real-time decisions on how to combat the fire, how to optimise the resources on site, quick and effective site clear up, and so on. The SFO would usually have the right to over-ride the decisions of the pervasive computing network if he/she so desires.

Claims (22)

:e;e.e.e.:e.. CLAIMS

1. A method of monitoring the location of a person within a building comprising providing the person with a portable radio wave transmitter (hereinafter called a mobile node), monitoring the receipt of radio waves emitted by the mobile node at a plurality of radio transceivers mounted in fixed positions relative to the building (hereinafter called static nodes), and comparing the pattern of the signal levels of the signals received at the static nodes with pre-stored patterns of signal levels received by the static nodes and selecting in a location estimator the pre-stored pattern or patterns which is closest to the measured pattern to provide an estimate or estimates of the location of the person.

2. The method of claim 1 in which the mobile node is a radio wave transceiver.

3. The method of claim 1 or claim 2 in which substantially all of the static nodes are sited externally of the building.

4. The method of claim 1 or claim 2 in which substantially all of the static nodes are sited internally of the building.

5. The method of any one of the preceding claims in which the static nodes communicate with a gateway node which communicates with the location estimator by way of radio signals.

6. The method of claim 5 in which some of the static nodes communicate with the gateway node by relaying a signal via other static nodes.

e;e ë ceete e ë

7. The method of claim 6 in which some of the static nodes communicate with the gateway node by relaying a signal via the mobile node.

8. The method of any one of the preceding claims in which the location estimator is arranged to make use of the recent history of the estimated positions of the person to select between a plurality of possible combinations of stored patterns of signal levels corresponding to the current position of the person.

9. The method of any one of the preceding claims in which a movement predictor is adapted to compare a plurality of probable positions of the mobile node with the estimated position of the mobile node selected by the location estimator to produce an estimate of the reliability of the position of the mobile node selected by the location estimator.

10. The method of any one of the preceding claims in which the mobile node transmits data bursts at a predetermined signal level.

11. The method of claim 6 or claim 7 as appended to claim 6, in which the static nodes are arranged as an ad hoc network for communication with the locations estimators.

12. The method of claim 10 in which there is a plurality of mobile nodes.

13. The method of claim 12 in which each mobile node emits a data burst only when it perceives that no other mobile node is emitting. c c

e::: ce.:. ee. ee.

14. The method of any previous claim in which the signal levels received by the static nodes (and gateway nodes) from a mobile node are viewed as a vector.

15. A system for monitoring the location of a person within a building comprising a portable radio wave transmitter (mobile node) for carrying by the person whose position is to be monitored, a plurality of radio transceivers (static nodes) adapted to be mounted in fixed positions in or around a building and configured for receiving radio waves emitted by the mobile node and transmitted to the respective static nodes, the mobile nodes being adapted to constitute an ad hoc network by which the level of signals received in use by the static nodes from the mobile nodes are communicated to a location estimator unit, the location estimator unit being configured to store sets of signal levels of signals received at the static nodes, and to compare sets of signal levels received by the static nodes with said stored sets of signal levels and to provide an estimate of the location of a person carrying the mobile node at least in part in dependence upon the results of said comparison.

16. The system of claim 15 comprising a plurality of mobile nodes for carrying by different people.

17. The system of claim 15 or 16 in which the stored set of signal levels is obtained by a manual calibration method.

18. The system of claim 15 or 16 in which substantially all of the static nodes are sited within the building and the arrangement is such that the stored set of signal levels is obtained by a self-calibration method.

19. The system of claims 15 to 18 in which the mobile node comprises a multimedia unit. :.:e

20. The system of claim 19 in which the mobile nodes being adapted to provide voice communication.

21. A mobile node for use in the system of claim 15 comprising of a processor, data memory, a radio transceiver and an antenna and able to transmit in use a radio signal at a known frequency and a known power.

22. A static node for use in the system of claim 15 comprising of an antenna, a radio transceiver, a means of monitoring the received signal level and providing a signal level output and a means of recovering information from the signal level output.