WO2009144030A2

WO2009144030A2 - Method for generating a radio map of an environment and radio communication system being controlled on the basis of a radio map generated by this method

Info

Publication number: WO2009144030A2
Application number: PCT/EP2009/003871
Authority: WO
Inventors: Alexander Baumann; Christoph Flossmann; Marian Grigoras; Octavian Sarca; Andrei Szabo; Joachim Bamberger; Hildegard Wiggenhorn
Original assignee: Siemens Aktiengesellschaft
Priority date: 2008-05-30
Filing date: 2009-05-29
Publication date: 2009-12-03
Also published as: WO2009144030A3

Abstract

A method for generating a radio map of an environment starts with an initial radio map, used as current radio map in a first step. The radio map is then stepwise adapted by amending a current radio map by help of a measurement up-date of this current radio map, yielding an amended radio map in each step, using the amended radio map of the preceding step as current radio map of the current step. The measurement update is based on measurements made available by localizing a radio device in the environment using the current radio map. The accuracy of the localization is thereby improved stepwise by using amended radio maps, and the accuracy of the radio maps is improved by using measurements made available from an improved localization of radio devices.

Description

METHOD FOR GENERATING A RADIO MAP OF AN ENVIRONMENT AND RADIO COMMUNICATION SYSTEM BEING CONTROLLED ON THE BASIS OF A RADIO MAP GENERATED BY THIS METHOD

FIELD AND BACKGROUND OF THE INVENTION

The present invention relates to a method for generating a radio map of an en- vironment and a radio communication system comprising a plurality of mobile radio devices and at least one base station, the system being controlled on the basis of a radio map generated by such a method.

Today's urban as well as indoor radio communication systems such as cellular mobile radio networks or wireless local area networks (WLANs) have to man- aged efficiently in order to achieve an optimal use of available resources in heterogeneous environments with a plurality of interfering access points. In existing radio communication systems the radio resource optimization or channel assignment is typically performed manually by service technicians or manually or automatically by using a priori knowledge about the signal propagation in the environment, e.g. radio modeling using knowledge about the structures within the environment, or by use of a signal propagation model, which s calibrated by site survey measurements, or using measurements from the communication network, e.g. received signal strength, interference, packet errors or traffic information.

Details about such prior art approaches have been published in articles of

[1] T. Vanhatupa, M. Hannikainen, T. D. Hamalainen, "Evaluation of throughput estimation models and algorithms for WLAN frequency planning", The International Journal of Computer and telecommunications Networking, Vol. 51 , Issue 11 (August 2007) [2] Arunesh Mishra, Suman Banerjee, William Arbaugh, "Weighted Coloring based Channel Assignment for WLANs", ACM SIGMOBILE Mobile Computing and Communications review, Vol. 9, Issue 3 (July 2005)

[3] Rodrigues, R.C., Mateus, R.R.; Loureiro, A.A.F., "On the design and capacity planning of a wireless local area network", Network operations and Management Symposium, 2000. NOMS 2000. 2000 IEEE/IFIP.

SUMMARY OF THE INVENTION

The present invention aims at further improving the resource management of such radio communication systems. This end, the present invention teaches a method for generating a radio map of an environment and a radio communication system comprising a plurality of mobile radio devices and at least one base station, the system being controlled on the basis of a radio map generated by such a method according to one of the independent claims. Preferred embodi- ments of the invention shall be subject to one of the dependent claims.

According to the present invention, a method for generating a radio map of an environment starts with an initial radio map, used as current radio map in a first step. The radio map is then stepwise adapted by amending a current radio map by help of a measurement update of this current radio map, yielding an amended radio map in each step, using the amended radio map of the preceding step as current radio map of the current step. The measurement update is based on measurements made available by localizing a radio device in the environment using the current radio map. The accuracy of the localization is thereby improved stepwise by using amended radio maps, and the accuracy of the radio maps is improved by using measurements made available from an improved localization of radio devices.

For the purpose of a description of the present invention, some terms used in this description are subsequently explained to clarify their meaning in the context of the description of the invention. A radio map is a data collection or database of measured received signal strengths, which may be used for localization of mobile radio devices in the network environment described by the radio map. The measurements may be acquired at so called reference points during a calibration phase and/or at arbi- trary points during operation of the network. The granularity of these reference points may depend on the environment and/or the algorithms used during localization. For each reference point, the coordinates of a position are stored in the radio map of the environment together with received signal strengths of the access points.

An initial radio map is a starting point for an iterative method to obtain an improved radio map from an initial radio map by iterating certain steps. Such an initial radio map may be calculated from theoretical models of signal propagation in the environment. Initial radio maps are usually rather inaccurate estimates of the true signal propagation situation in the environment, especially when the environment has a complicated structure.

An environment is a special arrangement of radio devices, e.g. base stations or access points and structural elements like buildings, obstacles, etc. influencing the propagation characteristics of signals in the regarded space.

A measurement update is an improved estimate of the actual radio map, which is obtained by improving a prior estimate by taking into account further measurement data.

During the measurement phase of the method according to the present invention, signal information is taken from mobile clients (mobile radio devices) and/or from access points (or base stations) and the measurements are stored in a database. The measurements are preferably taken in an unsupervised manner, so that preferably no human interaction is necessary to specify the positions at which the measurements are taken. The measurement phase may preferably take place either in a setup phase on the network or directly in the runtime period of the network.

In a pre-processing phase, that may preferably take place after the measurement phase, a data clustering process can be applied to reduce the amount of measurement data and to reduce noise, preferably by averaging over a plurality of measurement data put into the same cluster corresponding to (essentially) the same position, which is usually or frequently unknown during this step.

In a learning phase, the radio frequency signal propagation model is learned from the acquired measurement data, which have been taken during the preferably unsupervised measurements. Details about such procedures may be taken e.g. from

[4] B. Betoni Parodi, H. Lenz, A. Szabo, Hui Wang, J. Horn, J. Bamberger, D. Obradovic, "Initialization and Online-Learning of RSS Maps for Indoor / Campus Localization". In PLANS 2006, page 164-172, San Diego - CA, USA, April 2006

[5] Bruno Betoni Parodi, Henning Lenz, Andrej Szabo, Joachim Bamberger and Joachim Horn, "Algebraic and statistical conditions for the use of SLL", in ECC 07, Kos, Greece, July 2007.

The measurement update is based on measurements made available by local- izing a radio device in the environment using the current radio map, thereby stepwise improving the accuracy of the localization by using amended radio maps and improving the accuracy of the radio maps by using measurements made available from an improved localization of radio devices.

These steps may be performed or repeated any time without human interaction.

The present invention is characterized by a subsequent step of radio-resource- optimization based on the learned signal propagation model and desired properties of a communication system that is to be controlled on the basis of the generated radio map. The optimization of radio resources may be performed cost efficiently without any operator cost. The advantage is a reduction of serv- ice effort during system installation and during system operation and an increased network performance due to possible compensation of changes in the system within the physical limits.

The present invention allows for an automated and unsupervised measurement of radio-frequency-data without any human interaction, a pre-processing using a data clustering method, the use of a learning approach for learning radio-fre- quency-models from unsupervised measurements and an automated optimization of radio resources based on the radio-frequency models.

Previous similar algorithms for radio-frequency-optimization were missing the online learning step, the unsupervised measurement collection and learning of propagation environment, and were based either on manual calibration efforts or on theoretical propagation models.

Consequently, the new proposed method has much less operator costs, while no manual effort is required, and it is more accurate, while online measurements from the network and clients are used. A further advantage may be seen in the fact that the optimization can run unsupervised during the entire runtime of the network, keeping the information about the radio propagation up to date, while continuously optimizing radio resources.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 shows a schematic representation of the radio-frequency-learning-process according to the present invention.

Figs. 2 and 3 show schematic representations of two typical problems of radio- frequency-optimization. Fig. 3 illustrates the so-called "Hidden Node Problem" while Fig. 2 illustrates the so-called "Exposed Node Problem".

DETAILED DESCRIPTION OF THE INVENTION

It is a goal of the present invention to estimate radio filed distributions online from automated measurements without manual effort, to provide actual, accurate and cost free knowledge of the radio fields and to achieve an improvement, e.g. a cost reduction, of applications like radio channel optimization, positioning, intrusion detection and network diagnosis.

With respect to radio-frequency-optimization, it is a goal to achieve channel and/or transmit-power optimization based on learned filed distributions, in order to reduce interferences and to improve network performance, while keeping coverage.

In order to achieve these goals, the present invention aims at optimizing access point channels and transmit powers to reduce the area of "overlapping" regions with the constraint of keeping coverage. To avoid overlapping, on a channel of the strongest access point, the power of other access points has to be kept below the clear channel assessment (CCA) threshold, which is e.g. typically -89 dBm. To ensure coverage, at least one access point has to send with a power above the coverage threshold of typically -68 dBm.

The optimization is done by a calculation of access point transmit powers (so called "heat maps" or "radio maps") based on learning with online measurements. Heat map is another term for radio map, and both of them represent the received signal properties (typically received signal strength) distribution over an area. Then, the optimization of transmit powers is done by successive reduction of transmit powers until no overlapping is achieved, while coverage is preserved. The channels are preferably optimized by employing genetic or other discrete optimization algorithms to solve the discrete optimizations problem associated with channel optimization.

Fig. 1 shows a schematic representation of the radio-frequency-learning-proc- ess according to the present invention. Depending on a chosen user position 11 , the radio frequency-system RFS produces a received signal strength (RSS) "finger-print" 12 that is compared (D) with a signal estimate 13, generated from a radio-frequency model (RFM) of the system. The radio-frequency-model RFM not only generates a signal estimate 13, but also a position estimate 14. Both estimates 13, 14 are used by an update rule UR to generate a field update (15), which is used to update 16 the radio-frequency-model RFM.

The update rule UR is an adaptation algorithms designed to reduce differences 17 between the actual measured signal strengths 12 and the estimated signal strength 13 at the chosen position 14. Figs. 2 and 3 show schematic representations of two typical problems of radio- frequency-optimization. Fig. 3 illustrates the so-called "Hidden Node Problem" while Fig. 2 illustrates the so-called "Exposed Node Problem".

In wireless networking, the hidden node problem (Fig. 3) occurs when a node N1 is visible from a wireless access point (AP), but not from other nodes N2 communicating with said AP. This leads to difficulties in media access control. Hidden nodes in a wireless network refer to nodes that are out of range of other nodes or a collection of nodes. Take a physical star topology with an access point AP with many nodes N1 , N2 surrounding it, e.g. in a circular fashion: Each node N1 , N2 is within the communication range 33 of the AP, as the access point AP is within the communication range 31 , 34 of each Node N1 , N2, but the nodes cannot communicate directly with each other, as they do not have a direct radio connection to each other.

In a wireless network, it is likely that a node N1 at the far edge of the access point's range 33, can see the access point AP, but it is unlikely that the same node N1 can see a node N2 on the opposite end of the access point's range 33. These nodes N1 , N2 are known as "hidden". The problem is when nodes N1 and N2 start to send packets simultaneously to the access point. Since node N1 and N2 cannot sense their respective carriers, Carrier sense multiple access with collision avoidance (CSMA/CA) does not work. Both nodes N1 , N2 only sense the carrier of the access point, but not a carrier of another node.

Consequently, there will be packet collisions or bandwidth / channel sharing when RTS/CTS (Request to Send/Clear to Send) procedures of the Carrier Sense Multiple Access (CSMA) protocol is active. Frequently, these problems will induce reductions in traffic capacity of the network. To overcome this problem, handshaking is frequently implemented in conjunction with the CSMA/CA (collision avoidance) scheme.

As shown in Fig. 3, the communication ranges 31 , 34 of nodes N1 and N respectively with radii 32 and 35 respectively, slightly overlap but are to small for both nodes to mutually sense their respective carriers, because both nodes N1 , N2 are too far apart 36. Both nodes N1 , n2 lie, however, inside the range 33 of the access point AP, and can therefore communicate 37, 38 with this access point.

In wireless networks, the exposed node problem (Fig. 2) occurs when a node is prevented from sending packets to other nodes due to a neighboring transmit- ter. Consider an example of 4 nodes labeled R1 , S1 , S2, and R2, where the two receivers R1 , R2 are out of range 21 , 26 of each other, while the two transmitters S1 , S2 in the middle are in range 23, 25 of each other. Here, if a transmission T between S1 and R1 is taking place, node S2 is prevented 27 from transmitting to R2, as it concludes - after carrier sense - that it will interfere with the transmission by its neighbor S1. However, in fact, R2 could still receive the transmission of S2 without interference because it is out of range from S1. This problem also frequently causes reductions in traffic capacity of a network.

As shown in Fig. 2, although the communication ranges of most radio devices overlap and only the ranges 26 and 21 of stations R2 and R1 do not overlap, R1 is within the range 23 of S1 and S1 is in the range 22 of R1 , as S2 is in the range 26 of R2 like R2 is within the range 24 of S2, since, e.g. R1 and S2 are too far apart 27. Consequently, S1 , transmitting T to R1 prohibits S2 to send 27 to R2, because S2 senses the carrier of S1 and concludes there might be a collision, although, in fact, the transmission 27 of S2 to R2 would not interfere with the transmission T from S1 to R1 , because R1 would not receive the signals 27 transmitted by S2.

Radio-frequency-optimization according to the present invention aims at minimizing performance-losses induced by these other similar phenomena. This is achieved e.g. by a radio channel optimization, i.e. e.g. by minimizing the over- lapping regions on the same channel by optimally adjusting the signal strengths and thresholds in the system, or by a channel/transmit-power optimization based on the learned field distributions. This requires an accurate knowledge of the radio map, which is achieved by the method according to the invention. The invention therefore reduces interferences to improve network performance while keeping coverage.

In order to achieve these goals, the present invention preferably optimizes access point channels and transmit powers to reduce the area of "overlapping" regions with the constraint of keeping coverage. To avoid overlapping, on a channel of the strongest access point, the power of other access points has to be kept below the clear channel assessment (CCA) threshold, which is e.g. typically -89 dBm. To ensure coverage, at least one access point has to send with a power above the coverage threshold of typically -68 dBm.

The optimization is preferably done by a calculation of access point transmit powers (so called "heat maps") based on learning with online measurements. Then, the optimization of transmit powers is done by successive reduction of transmit powers until no overlapping is achieved, while coverage is preserved. The channels may be optimized by employing genetic or other discrete optimization algorithms to solve the discrete optimizations problem associated with channel optimization.

The optimization of the radio management can be done among others by changing the transmit channels of each access point or base station and/or by the variation of the transmit power of each access point or base station. The computation of the communication performance, e.g. throughput, peak data rate, interference, packet delay, is performed based on the learned radio map and on the current radio management setup, e.g. channel and transmit power of each base station or access point. The optimization can be performed jointly over the transmit power and channels, but this is computationally more complex. A suboptimal but simpler method can first optimize the transmit powers, adapting then to minimize the overlap between access points or base stations, with the constraint to satisfy the coverage requirement. In a second stage, the channel can be optimized.

Claims

What is claimed is:

1. Method for optimization of the radio management of a radio communications system (RFS) based on an automatically adaptive radio map (RFM), the method comprising the following steps:

a) Starting with an initial radio map, used as current radio map (RFM) in a first step;

b) Stepwise adapting (16) the radio map by amending a current radio map by help of a measurement update (UR) of this current radio map, yielding an amended radio map in each step;

c) Using the amended radio map of the preceding step as current radio map of the current step;

Characterized in that

d) The measurement update is based on measurements (14) made available by localizing a radio device in the environment using the current radio map;

e) Thereby stepwise improving the accuracy of the localization by using amended radio maps and improving the accuracy of the radio maps by using measurements made available from an improved localization of radio devices, and by a subsequent

f) Radio-resource-optimization based on the learned signal propagation model and desired properties of a communication system, which is to be controlled on the basis of the generated radio map.

2. Method according to claim 1 , characterized in that the initial radio map is generated on the basis of at least one theoretical model of signal propagation in the environment.

3. Method according to one of the preceding claims, characterized in that the initial radio map is generated on the basis of self-calibration with mutual base station measurements of at least one radio device in the environment.

4. Method according to one of the preceding claims, characterized by a

a) Measurement phase, during which signal information is taken from at least one mobile radio device and/or from the access points and stored in a database;

b) Learning phase, during which a signal propagation model is learned from he measurement data;

c) Radio-resource-optimization based on the learned signal propagation model and desired properties of a communication system, which is to be controlled on the basis of the generated radio map.

5. Method according to claim 4, characterized by a pre-processing phase, carried out after the measurement phase and before the learning phase, during which pre-processing phase a data clustering is applied to reduce the amount of measurement data and to average the noise of multiple measurements from the same position.

6. Radio communication system comprising a plurality of mobile radio devices and at least one base station, the system being controlled on the basis of a radio map, which has been generated by a method according to one of the preceding claims.