WO2007129939A1

WO2007129939A1 - Increasing the accuracy of location and / or path information of a moving client in a wireless network

Info

Publication number: WO2007129939A1
Application number: PCT/SE2006/050098
Authority: WO
Inventors: Imre Koncz; Gabor Nemeth
Original assignee: Telefonaktiebolaget Lm Ericsson (Publ)
Priority date: 2006-05-04
Filing date: 2006-05-04
Publication date: 2007-11-15

Abstract

In a wireless networks a method, a client device and a server node for increasing the accuracy of location and/or path information of a moving client are provided in which an accelerometer is located in a client device (120) providing motion information that combined with the strength intensity of radio signals (RSSI) can generate sufficient input information for a server node (130) furnished with business logic to obtain geographical position. In accordance with the implementation, an iteration algorithm is provided calculating the estimated position and/or path of the client. The client device (120) and the server node (130) are also disclosed, referring to localization of a moving client.

Description

INCREASING THE ACCURACY OF LOCATION AND/OR PATH INFORMATION OF A MOVING CLIENT IN A WIRELESS NETWORK

BACKGROUND OF THE INVENTION

Technical Field of the Invention

The invention relates in general to wireless networks. In particular, and not by way of limitation, the present invention directed to a method, a client device and a server node for localization of a moving client.

Description of Related Art

There are many methods of automatically determining the location of an object in a wireless network. These methods range from Global Positioning System (GPS) to solutions of acoustic, infrared or radio-frequency (RF) systems. Each method has advantages and disadvantages in various environments (e.g., indoors, outdoors), yielding differences in metrics such as accuracy, repeatability, computational complexity, power consumption, ease of use, cost and infrastructure requirements. In addition, different methods have advantages or disadvantages in supporting requirements for client privacy and control of location information. In many applications of interest, ease of use and deployment are more important than high degrees of accuracy. Since there is no single technology that can address all environments and requirements, it is likely that many different determination methods will be needed to support various location aware applications. Some location-determination schemes can be fully implemented in an isolated client device, such as a GPS navigation device as it is described in patent specification US-6 , 865 , 453 , although there is a large cost in infrastructure and deployment of the satellites. Other systems determine the client location in system servers, such as those employed at base stations in cellular telephone systems, as it is shown in patent document US- 5,548,583 utilizing the time difference of arrival (TDOA), at three or more synchronized base stations. Location information is often considered sensitive and private and in general, a client-based scheme is deemed to offer better privacy control for the user.

A particular location determination method may or may not depend on a two-way communications infrastructure as an integral part. In many systems, a two-way communication network is already incorporated in the devices for interaction with the remote location-aware applications. This communication capability can then also be used for location determination. In these cases, there can be a cost and system complexity advantage, if there is a radio transmitter and receiver on the device and these can also be used for location measurements.

A beacon is a device at a known location that emits a signal that is used by a client to determine its location. Various techniques are used in conjunction with the beacon signal to actually obtain a precise knowledge of the client position. An example of such system is described in US- 6,567,044. These techniques can be as simple as proximity to the beacon (as with a lighthouse and a map) or as complex as estimating range to multiple beacons and then using triangulation .

As it is obvious from the description above, moving client tracking is a topic of interest in various wireless (communications, sensory, etc.) networks, in which different applications may require different quality of the tracking system in precision, responsiveness, or deployment strategy. Consequently, no universal solution is developed. E.g., GPS is a major technology and also mobile systems are able to locate the devices with some precision, but both technologies have constraints. Each technology has inherent strengths and weaknesses depending on many factors: accuracy, environment, (e.g., temperature, pressure, wind, ambient light), power, infrastructure requirements, susceptibility to noise, etc.

Indoor localisation systems usually use RF signals as primary (or only) reference. However if there is any information available about the user's movement that data can assist the localisation system and increase its precision and reliability.

In current localisation systems that use movement data (e.g., tracking moving robotsl there is an odometer built in that conveys speed information to the tracking logic. (Speed data is derived from displacement measurements output by the odometer. )

The problem with respect to the related art is that infra-red, laser, video, and ultrasound based solutions need plenty of transmitters and are more expensive than a few RF transmitters. Using exclusively RF can result in low precision. This way one looses the possibility to use movement-related information. On the other side, humans hardly can use any odometers, so odometer based sensor fusion is not feasible for moving human tracking. Triangulation can not be used indoor because of indoor radio wave propagation features (reflections, interferences).

We have set ourselves the objective with this invention to improve the solutions described above by increasing the accuracy of location and/or path information of a moving client in a wireless network, in which the user carried client device can be easily integrated into a mobile terminal, and the system can be implemented over any existing mobile communications network. This service can support several applications like interactive museum guidance, context aware shop information, airport navigation, emergency/rescue actions etc.

SUMMARY OF THE INVENTION

The present invention involves a method, a client device and a server node, which solve the aforementioned problems, as well as other problems that will become apparent from an understanding of the following description.

Accordingly, it is the object of the invention to provide a method for increasing the accuracy of location and/or path information of a moving client. The object is attained by an algorithm, the core of which is a set of phases. In these phases the moving client's position is modeled by a set of abstract particles. For every particle an acceleration vector is chosen from a distribution governed by the measured acceleration length. The weights of the particles are updated according to the received RSSI data. A decision is made, in which particles with weights lower than the predefined threshold are filtered out and the weights of all remaining ones are re-normalized. Finally, the estimated position of the client device is calculated as the weighted centre of the particle set.

In another aspect, the present invention is directed to a client device, which comprises a receiver and an accelerometer, the data of which are transferred to a business logic of a server node.

In yet another aspect, the present invention refers to a server node in which data of a reference Radio Signal Strength Intensity distribution map, a data set for every client device, and a processing unit are located.

In an optimized embodiment, accelaration data for walking recognition is also processed for improving the precision, stability and reliability of the system.

According to the present invention, it was recognized that an accelerometer providing motion information of a client device combined with the strength intensity of radio signals can generate sufficient input information for a server node furnished with business logic to obtain geographical position.

The most important advantage of the invention is that using an accelerometer, having a small size of -5x5 mm, the localization process is improved by responding very quickly to sudden movements. This makes this method especially applicable in cases where precise tracking of moving individuals is needed like rescue actions, live games etc. It is also advantageous zhat the user carried client device (radio receiver and accelerometer) can be easily- integrated into a mobile terminal, and the system can be implemented over any existing mobile communications network (e.g. , GPRS, WLAN) .

A further advantage is that no prior knowledge of the radio propagation model needed - it is only required that the measured Radio Signal Strenth Intensity (RSSI) vectors differ sufficiently from position to position. The algorithm is reliable even if no RSSI data is received in some time period, e.g. no transmitter is visible or some of them malfunction.

BRIEF DESCRIPTION OF THE DRAWINGS

For a more complete understanding of the invention, reference is made to the following detailed description taken in conjunction with the accompanying drawings wherein:

Fig. 1 is the block diagram of a basic communication scheme according to the invention.

Fig. 2 is the flowchart of a possible embodiment of the localization algorithm.

Fig. 3 illustrates the block diagram of the communication scheme supported by walking recognition. DETAILED DESCRIPTION OF THE BEST MODE AND THE PREFERRED EMBODIMENTS

In Fig. 1, the area is covered by radio transmitters 101, 102, 103, ..., 1On which transmit custom data packets (beacons) at regular intervals. Beacon packets 151, 152,

153, ... 15n contain a header identifying the transmitter and the transmission power used. Transmitters are positioned in a way that in every position inside the area a radio receiver 122 of a client device 120 can identify signals from a few, e.g. two to five, transmitters. It should be noted that the system is operational if even only one transmitter can be seen from a location.

The user's client device 120 is equipped with a radio receiver 122 and an accelerometer 121. The radio receiver 122 measures the RSSI for every beacon decoded correctly, and the accelerometer 121 measures the client device's 120 acceleration.

The client device 120 communicates the measured RSSI data 142 and the acceleration data 141 to a server node 130 via a (possibly separate, e.g., WLAN, GPRS) data network periodically. As a reply the server node 130 calculates the data 143 of the client device's most probable position inside the area.

Prior to using the algorithm, a reference RSSI map needs to be built up over the whole location. The map should contain the distribution of RSSI data measured at sample points (e.g., a geographic grid) .

The server node 130 contains the data 131 of the reference RSSI distribution map, and a separate data set 132 for every user. The localization algorithm is executed independently for every client in a processing unit 133.

A client's data set 132 consists of a Particle Set (PS) containing N particles representing the probabilistic "guess" of the respective client's position. N is a parameter of the method and its optimal value depends on the required precision and the accessible computation capacity. (Typical value is N-1000...100000) .

Every particle has following parameters associated with it:

position vector ^=(r_x.r)_j ;

speed vector v,=(v_x.v_y)_j ;

weight W₁.

All these parameters are updated in iterative steps during the operation even if no measurement data is received. The time interval between two steps is a fixed parameter: Δr . The estimated position of a client at any time is calculated as the weighted centre of all particles in the PS.

Fig. 2 shows the flowchart of a possible embodiment of the localization algorithm.

At system initialization 201 all particles' position and velocity coordinates are distributed randomly, and their weights are set to 1/N. After initialization 201, an iteration 202 is applied consisting the following phases: MOVE phase 203, UPDATE phase 204, RESAMPLE phase 206 and ESTIMATE phase 207. During the first three phases the PS of the client is modified. The last one calculates the weighted centre of the particles, and outputs these coordinates as the estimation of the real position 208.

In MOVE phase 203, for every particle an acceleration vector is chosen from a distribution governed by the measured acceleration length: the direction of the acceleration is randomly generated from a uniform distribution from (0:2;r) and the amplitude is drawn from a Gaussian distribution centred around the measurement data with σ as a parameter.

First, the velocity of every particle is updated using the acceleration vector, then the position is updated using the new velocity:

During the UPDATE phase 204 the weights are updated according to the received RSSI data:

w.(k + l)= H- (jfc)* P(RSSI I };) , that is the weight of each particle is multiplied by the conditional probability that the given RSSI values are measured at the actual position of the given particle.

The conditional probability is calculated using the RSSI map 131 stored by the server node 130.

The DECISION phase 205 decides if the effective number of particles falls below a given threshold. The RESAMPLE phase 206 filters out particles with weights lower than a predefined threshold, and re-normalizes the weights of all remaining ones.

This phase is only executed if the effective number of particles falls below a given threshold. The effective number of particles can be calculated as:

eff ;V

/=1 and the threshold is a parameter to the algorithm.

Then, a cumulated distribution function (CDF) is calculated:

CDF₁ =£_{Wj .}

J=I

A uniformly distributed random number β is drawn from the interval [0;l/N] . After the new PS is generated as follows. The m-th particle in the new particle set will be replaced by the j-th particle of the old PS if:

CDF_j≤p +^≤CDF_j+].

The replacement particle's parameters are:

and all the weights are set to:

. In ESTIMATE phase 207, the estimated position 208 is calculated as the weighted centre of the particle set:

It shoud be noted that other strategies to return an estimate based on particle coordinates and weights can be used, e.g., returning the mass centre of particles with the 10 largest weights.

In Fig. 3, an optimized embodiment of the localization algorithm is shown in which elements indicated by the reference numbers are the same as in Fig. 1, but acceleration data for walking recognition 144 is transferred to a walking recognition unit 134 and information of moving type 145 is utilized by the processing unit 133 for localization algorithm.

The performance (precision, stability, reliability) can be increased by tuning the parameters of the algorithm, like the number of particles.

In case at least the vertical component of the acceleration can be separately measured, the following improvement can be added to the method.

The vertical acceleration contains characteristic patterns indicating the way the client is moving (e.g., walking or running) . This information 145 can suggest restriction of possible speed values, which information can be added to the filter's input. The vertical acceleration time-series measured by the accelerometer 121 can be characterized by a classification training algorithm. The classes are typical movement patterns with given speed intervals.

Every movement pattern has its own probabilistic density function ^•'w^¹' , where MT is the Movement Type classifier. E.g. standing is MT=O, slow walk is MT=I, and fast run is

MT=6. The one possible choice to •'•wr^'-' is a Gaussian density function with ^vuτ mean typical walking speed in class MT, and deviation parameter^σ*τ . The algorithm will be the same as described above but the UPDATE phase is complemented by the following step where the weights are updated according to the speed of the particles:

w,{k + l)= w,(A-)*P(MT I |v,|) = w,(k)*f_m (V₁.) .

In the basic case the acceleration directions are drawn randomly. In a further advantageous embodiment the algorithm may incorporate an improvement of acceleration correlation.

The angle of the acceleration vector is used identical to the previous iteration, and it is increased with a random angle: φ(k + \) = φ(k) + Δφ .

where Δφ is a uniform random number from [-oc,+a] , and oris a tunable parameter from [0.π\ .

The correlation is never applied if the length of the acceleration vector changes suddenly. An indicator to decide whether to use it or not is:

If ind ≥ind_trc%lhM , the correlation is not applied for the particle.

The method can be straightforwardly used in a 3D application (e.g. there are floors in the building), only the reference map and the particle data needs to include the 3rd axis.

Although preferred embodiments of the present invention have been illustrated in the accompanying drawings and described in the foregoing detailed description, it is understood that the invention is not limited to embodiments disclosed for wireless networks, but is capable of numerous rearrangements, modifications, and substitutions for localization of a moving client without departing from the spirit of the invention, based on an accelerometer providing motion information combined with the strength intensity of radio signals that can generate sufficient input information for a node furnished with business logic to obtain geographical position, as realized and defined by the following claims.

Claims

WHAT IS CLAIMED IS:

1. A method for increasing the accuracy of location and/or path information of a moving client in a wireless network, in which beacon packets (151, 152, 153, ..., 15n) of transmitters (101, 102, 103, ...,10n) are received by radio receiver (122) of a client device (120), Radio Signal Strenth Intensity (RSSI) vectors are measured and communication is established to a business logic of a server node (130) calculating and transfering the estimated location and/or path information for the client, characterized by the steps of

a.) initializing (201) the system by distributing position and velocity coordinates of particles representing the probabilistic estimation of the respective client's position and setting their weights in the business logic of the server node (130) ;

b.) iterating (202) the location and/or the path information of the client device (120) in the business logic of the server node (130) at phases of

i. MOVE (203), in which for every particle an acceleration vector is chosen from a distribution governed by measured acceleration amplitude,

ii. UPDATE (204), in which the weights of the particles are updated according to the received RSSI data, iii. DECISION (205), in which it is decided if the effective number of particles falls below a given threshold,

iv. RESAMPLE (206) , in which particles with weights lower than the predefined threshold are filtered out and the weights of all remaining ones are re-normalized,

v. ESTIMATION (207), in which the estimated position (208) of the client device (120) is calculated as the weighted centre of the particle set,

c.) communicating the estimated position (208) of the client device (120) from the business logic of the server node (130) to the the client.

2. The method of claim 1, characterized in that, in the step of initializing (201) the position and velocity- coordinates of particles are distributed randomly.

3. The method of claim 1, characterized in that, in the step of initializing (201) the weights of the particles are set to 1/N, where N is the number of the particles.

4. The method of claim 1, characterized in that, in the MOVE phase (203) of the step of iteration (202) the direction of the acceleration vector is generated randomly from a uniform distribution from (0:2;r) , and the length is drawn from a Gaussian distribution centred around the measurement data with <τ as a parameter.

5. The method of claim 1, characterized in that, in the MOVE phase (203) of the step of iteration (202) the direction of the acceleration vector is calculated from the angle of the previous iteration increasing it with a random angle chosen from the range of [θ:/r] .

δ. The method of claim 1, characterized in that, in the MOVE phase (203) of the step of iteration (202) acceleration data for walking recognition is provided and information of moving type (145) is processed.

7. The method of claim 6, characterized in that, at least a vertical component of the acceleration vector is separately measured and utilized as a restriction of possible speed values, which information can be added to the filter's input of the RESAMPLE phase.

8. The method of claim 7, characterized in that, the vertical acceleration time-series are classified into classes having typical movement patterns with given speed intervals .

9. The method of claim 1, characterized in that, in the UPDATE phase (203) of the step of iteration (202) the weight of each particle is multiplied by the conditional probability that the given RSSI values are measured at the actual position of the given particle, and the conditional probability is calculated using a reference map.

10. A client device (120) for increasing the accuracy of location and/or path information of a moving client in a wireless network, which client device (120) icludes a radio receiver (122) receiving beacon packets (151, 152, 153, ..., 15n) of transmitters (101, 102, 103, ...,10n) and measuring Radio Signal Strenth Intensity (RSSI) vectors at the location thereof, characterized in that, the client device (120)

a.) further comprises an accelerometer (121) measuring acceleration data (141) of the client device (120),

b.) transmits the acceleration data (141) together with the RSSI data to a business logic of a server node (130) , and

c.) receives the estimated location and/or path information for the client.

11. The client device (120) of claim 10, characterized in that, it is integrated into a mobile terminal of a telecommunication network.

12. The client device (120) of claim 10, characterized in that, the accelerometer (121) provides acceleration data for walking recognition (144) which is transferred to the business logic of the server node (130) .

13. A server node (130) for increasing the accuracy of location and/or path information of a moving client in a wireless network, comprises a business logic communicating with a client device (120), characterized in that, the business logic comprises

a.) a processing unit (133) receiving acceleration data (141) of the client device (120) together with data (142) of Radio Signal Strength Intensity (RSSI) of beacon packets (151, 152, 153, ..., 15n) , and sending data (143) of the most probable position of the client device (120) inside the area through a wireless telecommunication network, b. ) data (131) of a reference RSSI distribution map, communicating to the processing unit (133),

c.) a data set (132) of particles for every client device (120), also communicating with the processing unit (133) for localization algorithm executed independently for every client device (120).

14. The server node (130) of claim 13, characterized in that, the server node (130) further comprises a walking recognition unit (134) receiving accelaration data for walking recognition (144) of the client device (120), and generating information of moving type (145) for the processing unit (133).

15. The server node (130) of claim 13, characterized in that, the processing unit (133) comprises operational instructions of

a.) distributing position and velocity coordinates of particles and setting their weights,

b.) iterating the estimated location and/or path information by combining accelaration and RSSI data, and

c.) communicating with radio receiver (122) and accelerometer (121) of the client device (120), and the data set (132) of particles and the data (131) of the reference RSSI distribution map.

16. The server node (130) of claim 13, characterized in that, the data set (132) of particles comprises position, speed and weight information for each particles.