WO2007072400A2

WO2007072400A2 - A method and apparatus for determining the location of nodes in a wireless network

Info

Publication number: WO2007072400A2
Application number: PCT/IB2006/054921
Authority: WO
Inventors: Stephen Michael Pitchers; Paul Richard Simons Simons
Original assignee: Koninklijke Philips Electronics N.V.
Priority date: 2005-12-20
Filing date: 2006-12-18
Publication date: 2007-06-28
Also published as: WO2007072400A3

Abstract

A method of determining the location of nodes in a wireless network, the method comprising the steps of: 1 ) establishing wireless communication between nodes to obtain range measurements between a new node (4) and at least three reference nodes (1 , 2, 3); 2) determining the locations of points of intersection (A, B) of circles representing the range measurements, each circle being centred on an associated reference node and having a radius representative of the range measurement obtained between that reference node and the new node; 3) enumerating valid combinations of points of intersection; 4) for each enumerated combination, calculating a metric related to distances between the points of intersection in the combination; 5) selecting the combination having the lowest metric; and 6) determining the position of the new node using the selected combination.

Description

DESCRIPTION

A METHOD AND APPARATUS FOR DETERMINING THE LOCATION OF NODES IN A WIRELESS NETWORK

The present invention relates to the determination of the location of nodes in a wireless network, and in particular to the process of commissioning building service devices such as luminaires in a building.

Use of wirelessly-controlled luminaires in buildings is becoming increasingly popular, since it can substantially reduce lighting installation costs. Physical wires between the switching control nodes or actuation sensors and the luminaires are replaced by wireless (e.g. radio) links. All luminaires and switching control nodes need only be connected to an appropriate power source. Each luminaire includes a wireless receiver and each switching control node includes a wireless transmitter. During commissioning, each luminaire is identified and assigned to a particular switching control node or nodes. Typically, multiple luminaires are assigned to a particular switching control node, e.g. to operate multiple luminaires within one large room.

A significant disadvantage that remains in the prior art is that the commissioning process is time consuming and can interfere with the ability of other contractors on a building site to proceed with their work. For example, the commissioning electrician must typically selectively actuate luminaires or groups of luminaires throughout the building in order to work out which luminaires should be assigned to which switching control nodes. Other parts of the building could be in darkness while this operation continues. Another disadvantage is that the task of node assignments is a skilled job requiring the services of a lighting control specialist. It is an object of the present invention to overcome or mitigate at least some of the disadvantages indicated above.

According to the invention, there is provided a method of determining the location of nodes in a wireless network, the method comprising the steps of:

1 ) establishing wireless communication between nodes to obtain range measurements between a new node and at least three reference nodes; 2) determining the locations of points of intersection of circles representing the range measurements, each circle being centred on an associated reference node and having a radius representative of the range measurement obtained between that reference node and the new node;

3) enumerating valid combinations of points of intersection; 4) for each enumerated combination, calculating a metric related to distances between the points of intersection in the combination;

5) selecting the combination having the lowest metric; and

6) determining the position of the new node using the selected combination. By "valid combination" is meant a combination including either one or the other of the two points of intersection, but not both points, for each overlapping pair of circles of the at least three reference nodes.

The reference nodes may be constituted by some or all of the lighting and/or switching nodes in the network, and/or external devices. Each range measurement may be a single reading or, preferably, a derivation of several readings, for example a mean of numerous readings taken at different frequencies and/or using different techniques.

The techniques used to take measurements may include RF signal time-of-flight measurements and / or received signal strength indication. Each range measurement may be associated with a measure of its quality, for example a variance or standard deviation where the measurement is based on multiple readings, and/or a measure of the quality of a particular technique at a certain range. The measure of quality may be used as a weighting factor when determining the exact position of a node using a centre of gravity approach, as described below.

Thus, according to the invention, valid combinations of intersections of circles centred on the reference nodes are examined to find the combination with the best agreement in position. The variation in position is assessed for each combination and the one showing the least variation is deemed to be the correct combination.

The invention provides an efficient means to determine which combination of points of intersection should be used to place a new node. Once this has been done, the selected combination can be used to determine a likely position of the new node, in the manner described below.

Using the invention, the correct combination of intersections can be identified reliably, and it becomes possible to solve situations that would not otherwise be solvable. Accurate placement of nodes is highly desirable as the derived topology will be used to determine the positions of further nodes. The invention ensures that the resulting topology is more accurately determined thus and more likely to provide an acceptable basis for such further node positioning. Each metric may comprise a sum, an average or a variance of the distances between the points of intersection in a said combination.

However, each metric preferably comprises an overall score formed by summing a variance in x-coordinates of each point of intersection in a said combination with a variance in y-coordinates of each point of intersection. The metric may further comprise a variance in z-coordinates of each point of intersection (A, B) in a said combination summed to the overall score.

Step 6) may include positioning the new node at a centre of gravity of the points of intersection in the selected combination. A preferred embodiment includes using a weighted average to determine the centre of gravity.

Advantageously, the method may include, between steps 4) and 5), the step of 4a) comparing a second lowest metric to the lowest metric and, if the two metrics are similar, postponing steps 5) and 6) until the position of at least one further node has been determined. In this way, the invention supports the recognition of bad geometry, so that placement can be delayed if the current information is not yet adequate.

In the interests of efficiency, steps 3) and 4) may be performed for the combinations one at a time, saving on memory space. The method may include maintaining and updating a list of at least the lowest metric and the second lowest metric.

According to the invention, there is also provided apparatus for determining the location of nodes in a wireless network, the apparatus comprising: 1 ) means for establishing wireless communication between nodes to obtain range measurements between a new node and at least three reference nodes;

2) means for determining the locations of points of intersection of circles representing the range measurements, each circle being centred on an associated reference node and having a radius representative of the range measurement obtained between that reference node and the new node;

3) means for enumerating valid combinations of points of intersection; 4) means for calculating, for each enumerated combination, a metric related to distances between the points of intersection in the combination;

5) means for selecting the combination having the lowest metric; and

6) means for determining the position of the new node using the selected combination.

The means for determining may be operable to position the new node at a centre of gravity of the points of intersection in the selected combination.

The means for determining may be operable to position the new node using a weighted average to determine the centre of gravity.

The apparatus may be operable to compare a second lowest metric to the lowest metric and, if the two metrics are similar, postpone the selection of the combination having the lowest metric and the determination of the position of the new node until the position of at least one further node has been determined.

The apparatus may be operable to perform the enumerating and calculating for the combinations one at a time. The apparatus may be operable to maintain and update a list of at least the lowest metric and the second lowest metric.

The invention provides an effective means for automatically determining the location of nodes in a wireless network, e.g. in a 'plug-and- play' type mode, where nodes can be added to a network and their position automatically determined within the network. The automatic determination can be performed by a centralised system or by a new node itself. In order that the invention may more readily be understood, a description is now given, by way of example only, reference being made to the accompanying drawings, in which:-

Figure 1 is a schematic diagram of an exemplary situation including three nodes, the three circles representing range measurements taken between the respective nodes and a fourth node the position of which is unknown;

Figure 2 is a schematic diagram of the situation of Figure 1 showing a likely position of the fourth node; Figure 3 is a schematic diagram of apparatus according to the invention.

Figure 1 shows three nodes 1 , 2, 3, the positions of which are known. The circle around each node 1 , 2, 3 represents a range measurement taken between that node and a fourth node (not shown), the position of which is unknown. Each node 1 , 2, 3 defines a reference point. The circles are centred on the reference point in respect of which they are measured and have a radius equal to, or proportional to, the range measurement. Each circle intersects an overlapping circle at two points, A and B.

At least three reference points are required to determine the position of a new node. The first three nodes selected are any of those within range of the new node to be positioned, and preferably ones at a range at which the measurement accuracy is good. In some circumstances, only the relative topology of the array is important; the resulting topology may be a reflected or rotated representation of the actual topology without affecting the commissioning process. In other arrangements, where it is necessary or desirable to consider absolute topology with respect to a reference coordinate system, the absolute positions of the first three nodes may be used. Given the positions of the three nodes and range measurements between each node and the fourth node, the position of the fourth node can be established using a process of trilateration.

In Figure 1 , each pair of points of intersection for an overlapping pair of circles is labelled A, B. Therefore, there are a total of six intersection points A, B between the range circles from the three reference points 1 , 2, 3. Ideally, three of the intersections A, B should be aligned exactly at the position of the new node. However, errors in the range measurements cause inaccuracies in the diameter of the range circles, and there also may be errors in the positions of the reference points 1 , 2, 3 due to the effects of earlier placement inaccuracies.

The first step is to enumerate all the possible combinations of intersection. Table 1 below shows an enumeration for the example shown in Figure 1.

Table 1

A binary number with the same number of bits as the number of pairs of reference points represents each combination. Each bit represents either A or B for one of the pairs. This is convenient, as a valid combination must not include both A and B from the same pair of reference points.

Next, a metric is generated for each combination in order to decide whether it is the most satisfactory solution. The metric is related to the distances between the points of intersection in the combination. It will be understood that the combination having the tightest cluster of points of intersection is likely to indicate the best estimate of position of the fourth (unknown) node.

Thus, the metric could include a sum of the distances between intersections, for example, in the case of the combination 000, the distance between Ai₂, and A₂3, plus the distance between A₂3 and Ai₃, plus the distance between A₁3 and Ai₂. Alternatively, the metric could include an average distance between intersections or an average variance in distance between intersections. The lowest value of the metric would then represent the best estimate.

However, a preferred way is to measure the statistical variance (or standard deviation) of x and y values for the different points of intersection in each combination. Adding the variance in x to the variance in y gives an overall score for each combination. Table 2 below shows the variances in x and y and the resulting overall score for each combination in the example of Figure 1 .

Table 2

In this example, the best combination, the one with the lowest overall variance score, is BAB.

The final step is to place the new node at a position indicated by the tightest cluster of intersections. In a preferred embodiment, the position could be the centre of gravity of the selected intersections determined using a weighted average of the positions of the intersections. The location of each point of intersection is associated with a weighting factor w of the form w = R(range) x V(variance), where R is a function reflecting the quality of a measurement technique at a certain range, and V is a function converting a variance into a measure of quality. The weighting factor also accounts for the number of points of intersection in the combination.

The x- and y-coordinates of the new node are found by combining the locations of the points of intersection in the selected combination according to their weighting factors, using the following equations. V n

Figure 2 shows the position of the new node, designated as node 4, as determined using the above method.

In addition to finding the best combination, it is also desirable to find the second best combination (in this case BBA) and compare the scores of the best and second best combinations. If the scores are similar, this is indicative of bad geometry. In this case, a solution for the new node is preferably reattempted at a later time, when more reference points have become available. In general, there are two different cases of bad geometry: bad pair geometry, and bad overall geometry.

Bad pair geometry arises when the circles associated with a pair of reference nodes intersect at a small angle. In this case, a small deviation or error in one of the range measurements results in a large deviation in the locations of the intersections and, therefore, the estimated position of the new node.

Bad overall geometry arises when it is not clear from the circle of a third reference node which of the two points of intersection of the circles of a pair of nodes is indicative of the position of the new node. In one embodiment, the algorithm of the invention enumerates and evaluates the possible combinations one at a time, saving on memory space. When running through the scores for every combination, if a combination scores better than the current best, it becomes the current best and the previous best becomes the second best. Additionally, even if the current combination is not better than the current best, its score is checked against the current second best: if it is better than the current second best, it becomes the new second best. In a variant of this embodiment, the possible combinations are enumerated using a Gray code. A Gray code has the property that only one bit changes each time the code advances. An example three bit Gray code is shown in Erreur ! Source du renvoi introuvable.. With 0 corresponding to intersection A and 1 to intersection B in each pair we can represent all possible combinations of intersection, where exactly one intersection has been chosen from each pair. The pairs in this example are combinations of circles 1 , 2 and 3 from Erreur ! Source du renvoi introuvable..

10

The benefit of the Gray code technique is that we can reuse the accumulator values from the previous calculation to calculate the new variance. For the first combination we operate in the same way as the

15 above presented technique: we reset the accumulator and perform a normal calculation using the distance between Ai₂, and A₂₃, plus the distance between A₂₃ and Ai₃, plus the distance between Ai₃ and Ai₂, noted respectively 1 +2A, 2+3A and 3+1 A.

For all subsequent operations, we do not reset the accumulator but

20 simply remove ("decumulate") one value and add ("accumulate") a different one. For example, we derive the variance for 001 by decumulating 3+1 A and accumulating 3+1 B, resulting in 1 +2A, 2+3A and 3+1 B. Next, the variance for 011 is calculated by decumulating 2+3A and accumulating 2+3B, resulting in 1 +2A, 2+3B and 3+1 B. The benefit is that, for each axis, we need only N + 2x(2^N -V) accumulate or decumulate operations rather than Nx 2^N accumulations, where N is the number of bits. Savings increase as N increases, as shown in Erreur ! Source du renvoi introuvable..

After tests, it has been obtained optimum accuracy from a multi- lateration implementation by using around five or six reference points (in two dimensions) for each placement, showing that the variant technique should therefore take less time to deliver the results (about one third of the time).

This technique depends on using a suitable implementation for the variance calculation, where the accumulator values are retained between operations. We need to retain the number of values accumulated, the sum of the values, and the sum of the square of the values. Following the accumulate and decumulate operations, the variance is given by the expression:

Once the best combination has been chosen, the desired position is estimated by the mean value, which is given by: x = Σ n=ϋ*.

N

Rather than enumerating all the possible combinations at once before calculating variances, it is possible to generate each combination individually and calculate its variance before generating the next combination. This avoids the need to store the Gray code sequence. During the algorithm we keep track of the current "best combination" and its variance and mean, avoiding the need to store the results of every calculation. Actually, we typically store both the best and the second best results, according to the principles of the first embodiment.

Moreover, it is possible to adapt this method also for three- dimensional multi-lateration, where up to four intersections per pair of spheres are to be handled. To handle this, it is possible to use a 4-ary Gray code, for three-dimensional positioning, which employs the numbers between 0 and 3 instead of just 0 and 1. Erreur ! Source du renvoi introuvable. shows a Gray code being used to enumerate all the possible combinations for two pairs of spheres (spheres 1 and 2, and spheres 2 and 3).

Figure 3 shows apparatus according to the invention, including a building management system 10 including a processor 12 and memory 14 storing a map 16.

The building management system 10 is able to communicate wirelessly (or by other communication channel such as mains-borne signalling) with the nodes 1 , 2, 3, in order to obtain range measurements, and with the node 4 in order to communicate the position of that node when calculated. The processor 12 is operable to process the range measurements according to the above method in order to calculate the position of the fourth node 4, and accordingly to update the map 16 stored in memory 14.

The building management system 10 may comprise a temporary computer implementing the commissioning process. Alternatively, the building management system 10 could be a permanent feature in the building and may have routine management and maintenance functions outside of the commissioning process. In a further variant, the building management system 10 is constituted by lighting and/or switching nodes including distributed processing and storage capability, which perform the above method without the need for a centralised system. Thus, each new node may have processing power to determine its own position using the algorithm described above, once it has established communication with at least three reference nodes that can broadcast their own positions. The new node, having established its own position could then redesignate itself as a reference node.

Although the invention has been described using three reference points 1 , 2, 3, the invention is able to deal with more than three reference points at the same time, allowing the invention to use as many reference points as may be available. In practice, the quality of the range measurements, as described above, may be taken into account when deciding which nodes to use as reference nodes.

Although the invention has been described in relation to deriving the topology of a wireless lighting array, the invention is generally applicable to any positioning application where a topology must be established based on range measurements that are subject to error.

Claims

1. A method of determining the location of nodes in a wireless network, the method comprising the steps of:

1 ) establishing wireless communication between nodes to obtain range measurements between a new node (4) and at least three reference nodes (1 , 2, 3);

2) determining the locations of points of intersection (A, B) of circles representing the range measurements, each circle being centred on an associated reference node and having a radius representative of the range measurement obtained between that reference node and the new node;

3) enumerating valid combinations of points of intersection;

4) for each enumerated combination, calculating a metric related to distances between the points of intersection in the combination;

5) selecting the combination having the lowest metric; and 6) determining the position of the new node using the selected combination.

2. The method of Claim 1 wherein each metric comprises a sum, an average or a variance of the distances between the points of intersection (A, B) in a said combination.

3. The method of Claim 1 wherein each metric comprises an overall score formed by summing a variance in x-coordinates of each point of intersection (A, B) in a said combination with a variance in y-coordinates of each point of intersection.

4. The method of Claim 3 wherein the metric further comprises a variance in z-coordinates of each point of intersection (A, B) in a said combination summed to the overall score.

5. The method of Claim 1 wherein step 6) includes positioning the new node (4) at a centre of gravity of the points of intersection (A, B) in the selected combination.

6. The method of Claim 5 including using a weighted average to determine the centre of gravity.

7. The method of Claim 1 including, between steps 4) and 5), the step of 4a) comparing a second lowest metric to the lowest metric and, if the two metrics are similar, postponing steps 5) and 6) until the position of at least one further node has been determined.

8. The method of Claim 1 including performing steps 3) and 4) for the combinations one at a time.

9. The method of Claim 8 including maintaining and updating a list of at least the lowest metric and the second lowest metric.

10. The method of any preceding claim, wherein step 3) comprises listing the possible combinations by using a Gray code.

1 1. The method of claim 10, wherein step 4) comprises, for each considered combination, the sub steps of: - comparing the considered combination with a previous combination, for identifying an amended part of the combination; removing from the result of the previous combination metric a first amount based on the previous combination corresponding to the amended part; adding a second amount based on the considered combination corresponding to the amended part.

12. Apparatus (10) for determining the location of nodes in a wireless network, the apparatus comprising:

1 ) means for establishing wireless communication between nodes to obtain range measurements between a new node (4) and at least three reference nodes (1 , 2, 3);

2) means (12) for determining the locations of points of intersection (A, B) of circles representing the range measurements, each circle being centred on an associated reference node and having a radius representative of the range measurement obtained between that reference node and the new node;

3) means (12) for enumerating valid combinations of points of intersection;

4) means (12) for calculating, for each enumerated combination, a metric related to distances between the points of intersection in the combination;

5) means (12) for selecting the combination having the lowest metric; and

6) means (12) for determining the position of the new node using the selected combination.

13. The apparatus of Claim 12 wherein each metric comprises a sum, an average or a variance of the distances between the points of intersection (A, B) in a said combination.

14. The apparatus of Claim 13 wherein the metric further comprises a variance in z-coordinates of each point of intersection (A, B) in a said combination summed to the overall score.

15. The apparatus of Claim 12 wherein each metric comprises an overall score formed by summing a variance in x-coordinates of each point of intersection (A, B) in a said combination with a variance in y- coordinates of each point of intersection.

16. The apparatus of Claim 12 wherein the means (12) for determining is operable to position the new node (4) at a centre of gravity of the points of intersection (A, B) in the selected combination.

17. The apparatus of Claim 16 wherein the means for determining (12) is operable to position the new node (4) using a weighted average to determine the centre of gravity.

18. The apparatus of Claim 12 operable to compare a second lowest metric to the lowest metric and, if the two metrics are similar, postpone the selection of the combination having the lowest metric and the determination of the position of the new node (4) until the position of at least one further node has been determined.

19. The apparatus of Claim 12 operable to perform the enumerating and calculating for the combinations one at a time.

20. The apparatus of Claim 19 operable to maintain and update a list of at least the lowest metric and the second lowest metric.