Methods for recording and determining the probability of failure of a radio network and central computer

Abstract

Method for recording and determining a failure probability (pA) of a radio network (4), characterized in that a. operating parameters (2) are cyclically received from devices (3) of the radio network (4) by a central computer (1) and made available via an access point, where Each transmitted operating parameter (2) includes an operating value (5) of the respective device (3) and a recording time t of the operating value (5) (P1), b. the central computer (1) cyclically stores the operating values ​​(5) of all operating parameters (2) recorded within a specified time interval (T) as the respective operating constellation (8) and checks the radio network (4) for the presence of a failure (A) (P2), c. the central computer (1) cyclically creates a model (12) based on the stored operating constellations (8) and failures (A), which assigns a failure probability (pA) (P3) to each possible operating constellation (8), d. that the central computer (1) calculates the probability of failure (pA) for each new stored operating configuration (8), e. the central computer (1) compares the probability of failure (pA) with a predetermined warning threshold (pAc) (P5) and generates a warning signal (W) (PW) if the warning threshold (pAc) is exceeded (pW), and f. for each current operating parameter (2) a difference (2') is output (P6), which would lead to an exceedance of the warning threshold (pAc) with otherwise constant operating values ​​(5) of other operating parameters (2).

Description

[0001] The invention relates to a method for detecting and determining the probability of failure of a wireless network and to a central computer configured to implement the method. In particular, this refers to a WLAN network (WLAN - Wireless Local Area Network).

[0002] Extensive wireless network infrastructures, which include multiple access points and are intended for a multitude of end devices, require special design and monitoring for their reliability. Design errors or overloading of the network infrastructure can lead to functional limitations or a complete failure of the wireless network. Particularly in industrial plants whose components communicate via wireless networks, wireless network failures can lead to interruptions in production processes, which can result in significant economic losses.

[0003] In order to ensure the reliability of radio networks, methods and devices have been developed that can optimize the planning and control of a radio network infrastructure.

[0004] DE 10 2006 016 800 A1 describes a measurement-supported iterative method for setting up an optimal wireless network in a building using a 3D model of the building and empirical models to determine the relevant performance parameters. Within the framework of the method, the arrangement of WLAN access points in a building is optimized using a 3D model and experimentally collected data. A further development of the method also provides for the repeated and iterative collection of the experimental data in real data traffic.

[0005] US 2002 / 0 168 993 A1 describes a path loss estimation method for power control and link adaptation for an IEEE 802.11h WLAN. The method measures the path loss between a receiver and a transmitter, and based on the measurement results, a power and link adaptation is performed for the receiver.

[0006] DE 10 2015 204 246 A1 describes a device and method for adjusting the power of a WLAN network. The transmission power of a WLAN module is adjusted such that the received signal strength at a reference WLAN module is less than a specified threshold.

[0007] US 2016 / 0 057 639 A1 describes a method for dynamically providing real-time services to a mobile device in a decentralized wireless network. This involves predicting a coverage area and the probability of a network error occurring in the coverage area.

[0008] US Patent No. 9,531,522 B2 describes a method for predictive resource allocation in a network. This method assigns a time window for responding to a request from a device in the network.

[0009] US 2015 / 0 148 040 A1 describes an anomaly correlation mechanism for analyzing handover processes in a communications network. This mechanism evaluates time series relating to errors during handover processes between radio cells.

[0010] The aforementioned inventions describe methods that enable the planning of a radio network or power adjustments to ensure radio network operation. In a radio network comprising a large number of participants, operation may not be maintained by adjusting transmission power alone. In such cases, it is necessary to be able to take appropriate measures before an impending failure in order to ensure the operation of the radio network.

[0011] It is an object of the invention to provide a method which makes it possible to determine the utilization of a radio network.

[0012] The problem is solved by the subject matter of the independent patent claims. Advantageous developments of the invention are disclosed by the features of the dependent patent claims, the following description, and the figure.

[0013] The invention provides a method for detecting and determining the probability of failure of a radio network. It is provided that devices of the radio network cyclically transmit operating parameters to a central computer, wherein each transmitted operating parameter comprises an operating value of the respective device and a detection time of the operating value. The central computer cyclically stores the operating values ​​of all operating parameters detected within a predetermined time interval as a respective operating constellation. Based on the stored operating constellations, the central computer cyclically creates a model that assigns a probability of failure to each possible operating constellation. In other words, the devices of the radio network regularly transmit operating parameters, each comprising an operating value and a detection time, to a central computer.The values ​​recorded within a specific time interval are summarized and saved as a respective operating constellation. These can be viewed at any time via access to the central computer. The central computer checks the radio network for any failures within the specific time interval. The stored operating constellations and detected failures are used by the central computer to create a model that assigns a failure probability to each possible operating constellation. A probability value can be calculated, for example, by counting how often a failure occurred in a given operating constellation. The ratio of counted failures to the total number of observations for this operating constellation results in a probability. A failure can, for example, consist of at least one of the devices being disconnected or cut off from the radio network.

[0014] A wireless network can, for example, be a Wi-Fi network. Devices in the wireless network include intermediary devices or infrastructure devices, which can include, for example, access points and / or routers and / or repeaters, and end devices such as devices with a wireless module compatible with the wireless network.

[0015] Operating parameters can be parameters that describe the state of a device and, for example, relate to its connection to the radio network. An operating parameter comprises an operating value and a recording time of the operating value. An operating parameter can, for example, comprise a transmission and / or reception power, the number of other devices in the radio network detected by the respective device, a transmission frequency, a number of transmitted, corrected, faulty or received packets, error messages or signal quality. The operating parameters that relate to the device itself can include, for example, power consumption, an operating state, a system version or the temperature or its memory utilization or a geographical position. A central computer can be a server or computer with at least one microprocessor.A respective operating constellation is a compilation of all operating parameters which were recorded according to their recording time in a respective predetermined time interval and is to be understood as a snapshot of the radio network in the respective time interval. A time interval can, for example, be a period of time from 1 minute to 1 hour. A failure of the radio network is a state of the radio network in which it does not fulfill a predetermined function or no longer fulfills it to a predetermined extent. This does not only mean a complete failure of all devices, but also, for example, a state in which a single device cannot interact with the radio network to the required extent. A model can, for example, be a mathematical model based on empirical values ​​or is a simulation.The model can be designed to output a failure probability for an operating configuration. With each additional stored operating configuration and information about the presence of a failure, the model can be adapted with each cycle. In other words, it is improved iteratively. The model can not only calculate the failure probability of a real operating configuration, but can also output at least one operating parameter and / or a failure probability for a changed operating configuration. Thus, it may be possible, as part of an improvement measure, to forecast the resulting reception performance when expanding the radio network by a repeater at a specific spatial location. A failure probability can therefore also be forecast or extrapolated for a planned operating configuration.

[0016] The central computer is designed to cyclically calculate a probability of a failure occurring for each new stored operating configuration. In other words, after each successful transmission of an operating configuration, the central computer uses the model to calculate the current failure probability. This has the advantage of allowing a trend of increasing failure probability in the radio network to be identified.

[0017] The current failure probability is compared with a predetermined warning threshold, and exceeding the warning threshold triggers the generation of a warning signal. In other words, a warning threshold can be determined with respect to the failure probability, and exceeding this threshold triggers the issuance of a warning signal. For example, if the calculated failure probability exceeds the warning threshold, e.g., 2%, a warning signal can be transmitted or sent by the central computer in the form of an email. This has the advantage that countermeasures can be initiated if the failure probability becomes too high, i.e., if it exceeds the warning threshold.

[0018] For each current operating parameter, a difference is output which, according to the probability distribution, would lead to an exceedance of the warning threshold if all other operating values ​​remain constant. In other words, the central computer determines the difference for a current operating value of an operating parameter that would lead to an exceedance of the warning threshold. This provides the advantage of allowing users to see how large individual operating parameter reserves are or which values ​​need to be adjusted to reduce the probability of failure. This allows critical operating parameters to be identified.

[0019] A further development of the invention provides that a failure comprises at least one occurrence of an indication value of at least one indication parameter within a predetermined value range. An indication parameter can comprise an operating parameter whose operating value represents a statement about the functioning of the radio network. This can include, for example, the received power and / or the number of faulty packets received. The predetermined value range can be selected such that proper functioning of the radio network is no longer ensured if a respective indication parameter has an indication value within this value range. For example, a failure can be defined as a device falling below a certain received power.

[0020] A further development of the invention provides that the model for assigning a failure probability is generated by correlating the operating configurations with the respective failures that have occurred using a pattern recognition algorithm. In other words, the operating configurations and the failures of the radio network are used as input values ​​for a pattern recognition algorithm, which then creates a model. A pattern recognition algorithm can be the program code of a program for evaluating large amounts of data (big data tool). In particular, machine learning methods or artificial neural networks can be used. This offers the advantage that there is no need to manually define causal relationships, which cannot be grasped by a user due to the complexity and volume of the data.The algorithm can thus create a model for the probability of failure by recognizing patterns between operating configurations and failures.

[0021] A further development of the invention provides that the central computer determines a forecast operating constellation using an extrapolation method and generates a warning signal if the warning threshold is exceeded for the forecast operating constellation. In other words, a forecast or trend for an operating constellation at a future point in time is created from the stored operating constellations, and a warning signal is generated if a failure probability is calculated for the forecast operating constellation that exceeds the warning threshold. In a simple example, a constantly decreasing received power over a period of hours can be linearly extrapolated, and a warning signal can be generated if the extrapolation predicts an operating constellation for the next hour that has a failure probability above the warning threshold.This allows adjustments to the wireless network to be made at an early stage.

[0022] A further development of the invention provides that boundary parameters external to the radio network are determined for a specific period of time and assigned to the operating configuration, and the boundary parameters are taken into account by the central computer when creating the model. In other words, in addition to a failure and an operating configuration, boundary parameters external to the radio network can also be assigned to a specific period of time. Boundary parameters external to the radio network can include values ​​that describe boundary conditions under which the radio network operates. These can include, for example, voltage fluctuations in a supply voltage or maintenance work. This offers the advantage that boundary conditions that are not directly related to the radio network but can influence the probability of failure can also be recorded.For example, a failure may be due to maintenance work performed on a device. Using the boundary parameter, the model can associate the failure with this rather than with the operating configuration of the time interval.

[0023] The invention also includes a central computer which is configured to carry out one of the methods.

[0024] The invention also includes further developments of the central computer according to the invention that have features already described in connection with the further developments of the method according to the invention. For this reason, the corresponding further developments of the central computer according to the invention are not described again here.

[0025] Exemplary embodiments of the invention are described below. These show: Fig. 1 a schematic representation of a central computer; Fig. 2 a procedure as it can be carried out by the central computer; and Fig. 3 a possible model for determining a probability of failure.

[0026] The exemplary embodiments explained below are preferred embodiments of the invention. In the exemplary embodiments, the described components of the embodiments each represent individual, independently considered features of the invention, which also further develop the invention independently of one another and are thus also to be considered as components of the invention, either individually or in a combination other than that shown. Furthermore, the described embodiments can also be supplemented by further features of the invention already described.

[0027] In the figures, functionally identical elements are provided with the same reference numerals.

[0028] Fig. 1 shows a central computer 1 to which operating parameters 2 from devices 3 of a radio network 4 are cyclically transmitted. Operating parameters 2 can include operating values ​​5 and respective recording times t of the operating values ​​5. A radio network 4 can be, for example, a GSM radio network or a WLAN radio network. The devices 3 can be, for example, access points or repeaters that function as infrastructure devices and end devices, such as microcomputers or microcontrollers, that are connected to the radio network 4 via a suitable radio module. Operating parameters 2 can be data that can relate to a respective device 3 itself or to the interaction of the device 3 with the radio network 4. This can include, for example, the received power and / or the transmitted power. The central computer 1 can also receive boundary parameters 6 that were recorded by an external sensor unit 7.The central computer 1 can assign the operating parameters 2 and boundary parameters 6 recorded in a time interval T to a respective operating constellation 8. Thus, an operating constellation 8 represents a snapshot of the radio network 4 in a specific time interval T. Information about the presence of a failure A is also assigned to such a time interval T. A failure A can, for example, comprise the occurrence of an indication value 9 of an indication parameter 10 in a predetermined value range 11. An indication parameter 10 can be an operating parameter 2 which must have specific values ​​for the proper functioning of the radio network 4. The central computer 1 can be configured to create a model 12 based on the stored data, which model assigns a failure probability pA to a specific operating constellation 8.

[0029] Fig. 2 shows a possible sequence of a method according to the invention for determining a failure probability pA of a radio network 4. In a step P1, devices 3 of the radio network 4 cyclically transmit operating parameters 2 to the central computer 1. In step P2, the central computer 1 stores the operating values ​​5 of all operating parameters 2, which were recorded in a predetermined time interval T, as a respective operating constellation 8 and checks the radio network 4 for the presence of a failure A. On the basis of the stored operating constellations 8, the central computer 1 can create P3 a model 12 which assigns a failure probability pA to each possible operating constellation 8. The central computer 1 can be set up to automatically calculate P4 the failure probability pA for each new stored operating constellation 8. It is possible that a warning threshold pAc is defined with regard to the failure probability pA.It can be provided that the central computer 1 compares P5 the calculated failure probability pA for a new stored operating constellation 8 with the warning threshold pAc and generates a warning signal W PW if the warning threshold pAc is exceeded. It can be that the central computer 1 outputs a difference 2' for each current operating parameter 2, which would lead to the warning threshold pAc being exceeded with the operating values ​​5 of other operating parameters 2 remaining the same P6. It can be that the central computer 1 determines P7 a forecast operating constellation 8' by means of an extrapolation method and, if the failure probability pA is exceeded for the forecast operating constellation 8', generates a warning signal W PW.

[0030] Fig.3. shows a possible model 12 for determining a failure probability pA of a radio network 4 Shown are three areas pA1, pA2, pA3, which describe three different failure probabilities depending on two operating values ​​5',5''.

[0031] Overall, the examples show how the invention provides a method for monitoring a failure probability of a radio network.

Claims

[1] Method for detecting and determining a failure probability (pA) of a radio network (4), characterized by , that a. operating parameters (2) are cyclically received from devices (3) of the radio network (4) by a central computer (1) and are provided via an access, where each transmitted operating parameter (2) comprises an operating value (5) of the respective device (3) and a recording time t of the operating value (5) (P1), b. the central computer (1) cyclically stores the operating values ​​(5) of all operating parameters (2) recorded within a given time interval (T) as the respective operating constellation (8) and checks (P2) the radio network (4) for the presence of a failure (A), c. the central computer (1) cyclically creates a model (12) on the basis of the stored operating constellations (8) and failures (A), which assigns a failure probability (pA) to each possible operating constellation (8) (P3), d. that the central computer (1) calculates the failure probability (pA) for each new stored operating constellation (8), e. the central computer (1) compares (P5) the failure probability (pA) with a predetermined warning threshold (pAc) and generates (PW) a warning signal (W) if the warning threshold (pAc) is exceeded, and f. for each current operating parameter (2) a difference (2') is output (P6), which would lead to an exceedance of the warning threshold value (pAc) with otherwise constant operating values ​​(5) of other operating parameters (2). [2] Method according to claim 1, characterized bythat a failure (A) comprises at least one occurrence of an indication value (9) of at least one indication parameter (10). [3] Method according to one of the preceding claims, characterized by that the model (12) for assigning a failure probability is generated by correlating the operating constellations (8) with the respective failures (A) that have occurred by means of a pattern recognition algorithm. [4] Method according to one of the preceding claims, characterized by that the central computer (1) determines (P7) a predicted operating constellation (8') by means of an extrapolation method and generates a warning signal (W) (PW) if the warning threshold value (pAc) is exceeded in the predicted operating constellation (8'). [5] Method according to one of the preceding claims, characterized bythat a specific period of time (T) is assigned boundary parameters (6) external to the radio network, which are taken into account by the central computer (1) when creating the model (12) and are recorded by external sensors (7) (PE). [6] Central computer (1) arranged to carry out one of the preceding methods.

Images