TECHNICAL FIELD

Wireless communication and in particular, to detection of passive intermodulation (PIM) by a machine learning system for Fifth Generation (5G) and Fourth Generation (4G) radio networks.

BACKGROUND

The Third Generation Partnership Project (3 GPP) has developed and is developing standards for Fourth Generation (4G) (also referred to as Long Term Evolution (LTE)) and Fifth Generation (5G) (also referred to as New Radio (NR)) wireless communication systems. Such systems provide, among other features, broadband communication between network nodes, such as base stations, and mobile wireless devices (WD), as well as communication between network nodes and between WDs. Sixth Generation (6G) wireless communication systems are also under development. It is noted that WDs may also be referred to as user equipment (UE) in 3GPP standards specifications.

Referring to the drawing figures, in which like elements are referred to by like reference numerals, there is shown in FIG. 1 a schematic diagram of a communication system 10, according to an embodiment, such as a 3GPP-type cellular network that may support standards such as LTE and/or NR (5G), which comprises an access network 12, such as a radio access network, and a core network 14. The access network 12 comprises a plurality of network nodes 16a, 16b, 16c (referred to collectively as network nodes 16), such as NodeBs (NBs), eNBs, gNBs or other types of wireless access points, each defining a corresponding coverage area 18a, 18b, 18c (referred to collectively as coverage areas 18). Thus, the network nodes 16 can be 4G base stations (eNBs) and/or 5G base stations (gNBs). Each network node 16a, 16b, 16c is connectable to the core network 14 over a wired or wireless connection 20. A first wireless device (WD) 22a located in coverage area 18a is configured to wirelessly connect to, or be paged by, the corresponding network node 16a. A second WD 22b in coverage area 18b is wirelessly connectable to the corresponding network node 16b. While a plurality of WDs 22a, 22b (collectively referred to as wireless devices 22) are illustrated in this example, the disclosed embodiments are equally applicable to a situation where a sole WD is in the coverage area or where a sole WD is connecting to the corresponding network node 16. Note that although only two WDs 22 and three network nodes 16 are shown for convenience, the communication system may include many more WDs 22 and network nodes 16.

Also, it is contemplated that a WD 22 can be in simultaneous communication and/or configured to separately communicate with more than one network node 16 and more than one type of network node 16. For example, a WD 22 can have dual connectivity with a network node 16 that supports LTE and the same or a different network node 16 that supports NR. As an example, WD 22 can be in communication with an eNB for LTE/E-Universal Terrestrial Radio Access Network (UTRAN) and a gNB for NR/NG-RAN.

FIG. 2 is a block a network node 16 in communication with a WD 22. The communication system 10 of FIG. 1 includes a network node 16 provided in a communication system 10 and including hardware 38 enabling it to communicate with the WD 22. The hardware 38 may include a radio interface 42 for setting up and maintaining at least a wireless connection 46 with a WD 22 located in a coverage area 18 served by the network node 16. The radio interface 42 may be formed as or may include, for example, one or more RF transmitters, one or more RF receivers, and/or one or more RF transceivers. The radio interface 42 includes an array of antennas 43 to radiate and receive signal carrying electromagnetic waves.

In the embodiment shown, the hardware 38 of the network node 16 further includes processing circuitry 48. The processing circuitry 48 may include a processor 50 and a memory 52. In particular, in addition to or instead of a processor, such as a central processing unit, and memory, the processing circuitry 48 may comprise integrated circuitry for processing and/or control, e.g., one or more processors and/or processor cores and/or FPGAs (Field Programmable Gate Array) and/or ASICs (Application Specific Integrated Circuitry) adapted to execute instructions. The processor 50 may be configured to access (e.g., write to and/or read from) the memory 52, which may comprise any kind of volatile and/or nonvolatile memory, e.g., cache and/or buffer memory and/or RAM (Random Access Memory) and/or ROM (Read-Only Memory) and/or optical memory and/or EPROM (Erasable Programmable Read-Only Memory).

Thus, the network node 16 further has software 44 stored internally in, for example, memory 52, or stored in external memory (e.g., database, storage array, network storage device, etc.) accessible by the network node 16 via an external connection. The software 44 may be executable by the processing circuitry 48. The processing circuitry 48 may be configured to control any of the methods and/or processes described herein and/or to cause such methods, and/or processes to be performed, e.g., by network node 16. Processor 50 corresponds to one or more processors 50 for performing network node 16 functions described herein. The memory 52 is configured to store data, programmatic software code and/or other information described herein. In some embodiments, the software 44 may include instructions that, when executed by the processor 50 and/or processing circuitry 48, causes the processor 50 and/or processing circuitry 48 to perform the processes described herein with respect to network node 16.

The processing circuitry 48 is in communication with a transceiver 56 of the radio interface 42. The transceiver 56 includes a transmitter for transmitting signals to WDs 22 and a receiver for receiving signals from WDs 22, via the antennas 43.

The communication system 10 further includes the WD 22 already referred to. The WD 22 may have hardware 60 that may include a radio interface 62 configured to set up and maintain a wireless connection 46 with a network node 16 serving a coverage area 18 in which the WD 22 is currently located. The radio interface 62 may be formed as or may include, for example, one or more RF transmitters, one or more RF receivers, and/or one or more RF transceivers. The radio interface 62 includes an antenna or an array of antennas 63 to radiate and receive signal carrying electromagnetic waves.

The hardware 60 of the WD 22 further includes processing circuitry 64. The processing circuitry 64 may include a processor 66 and memory 68. In particular, in addition to or instead of a processor, such as a central processing unit, and memory, the processing circuitry 64 may comprise integrated circuitry for processing and/or control, e.g., one or more processors and/or processor cores and/or FPGAs (Field Programmable Gate Array) and/or ASICs (Application Specific Integrated Circuitry) adapted to execute instructions. The processor 66 may be configured to access (e.g., write to and/or read from) memory 68, which may comprise any kind of volatile and/or nonvolatile memory, e.g., cache and/or buffer memory and/or RAM (Random Access Memory) and/or ROM (Read-Only Memory) and/or optical memory and/or EPROM (Erasable Programmable Read-Only Memory).

Thus, the WD 22 may further comprise software 70, which is stored in, for example, memory 68 at the WD 22, or stored in external memory (e.g., database, storage array, network storage device, etc.) accessible by the WD 22. The software 70 may be executable by the processing circuitry 64. The software 70 may include a client application 72. The client application 72 may be operable to provide a service to a human or non-human user via the WD 22.

The processing circuitry 64 may be configured to control any of the methods and/or processes described herein and/or to cause such methods, and/or processes to be performed, e.g., by WD 22. The processor 66 corresponds to one or more processors 66 for performing WD 22 functions described herein. The WD 22 includes memory 68 that is configured to store data, programmatic software code and/or other information described herein. In some embodiments, the software 70 and/or the client application 72 may include instructions that, when executed by the processor 66 and/or processing circuitry 64, causes the processor 66 and/or processing circuitry 64 to perform the processes described herein with respect to WD 22.

In some embodiments, the inner workings of the network node 16 and WD 22 may be as shown in FIG. 2 and independently, the surrounding network topology may be that of FIG. 1.

The transceiver 56 of the radio interface 42 of the network node 16 may be subjected to passive intermodulation (PIM). Passive Intermodulation (PIM) may be described as the distortion resulting from nonlinear mixing of two or more signals with different frequencies. PIM can arise from the interaction of passive components or elements that have non-linear responses with signals generated by the transmitter of the transceiver 56. PIM can be generated by a variety of components and objects including coaxial cable connectors, rusty bolts or joints between dissimilar metals, for example. Even some normally ‘linear’ components may generate PIM. PIM can produce interference with a desired signal received by the receiver of the transceiver 56 from one or more WDs 22, thereby desensitizing the receiver of the transceiver 56.

Thus, some network nodes 16 employ methods to reduce PIM. Such methods may include application of PIM cancellation (PIMC) algorithms. Such PIMC algorithms include adaptive filtering and least squares or least mean square algorithms. These algorithms may involve detecting PIM in the received signal. A problem with PIMC algorithms may include difficulty in model application to higher order PIM modes.

In a separate technical field, neural networks that can be trained to perform a task or solve a problem have developed in parallel with, and independently of, wireless communication systems.

A conceptual illustration of one example of a neural network is shown in FIG. 3. In this example, the neural network has an input layer 74 having a first number of input components, labeled “i”, followed by a number of hidden layers 75, each hidden layer having a number of hidden components, labeled “h”, and finally, an output layer 76 with a number of output components, labeled “o”. Note that in any given implementation of a neural network, the number of input components in the input layer may differ from the number of hidden components in the first hidden layer. Also, the number of output components in the output layer may differ from the number of hidden components in a hidden layer and may further differ from the number of input components in the input layer.

In the example neural network of FIG. 3, each component in a hidden layer receives as an input, the output from each component of a previous layer. Each such input is multiplied by a weight or number that is derived from training the neural network. Note that although three hidden layers are shown in FIG. 3, fewer or greater than three hidden layers may be implemented. The neural network of FIG. 3 may be implemented in software and/or hardware. For example, the weights may be stored in a computer memory. In other examples of neural networks, an output of one component may be fed back to its own input or to the input of a component in a preceding layer. Such feedback may reduce the overall complexity of the neural network in some applications. Many topologies have been studied, including the Jordan network, the Elman network and the Hopfield network. A convolutional neural network (CNN) is a type of neural network that employs mathematical convolution in at least one layer to convolve the input with a convolutional kernel to produce an output. CNNs have been used in image recognition applications to detect features of an image, for example.

FIG. 4 is a conceptual illustration of one example of a component k of a layer of a neural network. The example component k is shown receiving 4 inputs from a previous layer. For example, a j* component in a preceding layer may produce an output yj that is an input to the component k. Note that a component in a neural network may have fewer or greater than four inputs. Each input to the component k is received from a different component of the preceding layer and weighted by a different weight w. For example, y ₇ is weighted by Wjk- In the example of FIG. 4, the weighted inputs are all added together and then optionally added to a bias 0k . The contribution to the sum for positive weight Wjk may be called an excitation, whereas the contribution to the sum for negative weight Wjk may be called an inhibition. More complex rules for combining the weighted inputs may be implemented.

Once the weighted inputs are combined to produce the value Sk , the value Sk is input to an activation function f(sk) which determines the output yk of the component k, based on Sk- Two examples of activation functions are shown in FIGS. 5 A and 5B. In the graph of FIG. 5A, a signum function is shown, which acts as a threshold function. The signum may be defined as -1 below a threshold and +1 above the threshold. Thus, when the weighted sum Sk is less than a threshold, the output yk is low (-1), whereas when Sk is greater than a threshold, the output yk is high (+1). Other known functions such as a sigmoid function, shown in the graph of FIG. 5B, may be employed as the activation function f(sk). The sigmoid function may be represented _1+e-s - Note that in some neural networks, the output of a component can be a stochastic function of the total of the weighted inputs of the component. In such case, the component inputs may be interpreted as determining a probability p that the activation value (output) of the component is high.

In a learning or training stage, the weights and biases by which the inputs to the various components of the neural network are multiplied are iteratively adapted (learned) until some learning criteria is satisfied. For example, the weights and biases may be iteratively adapted until a difference between an expected output for a given input and an actual output obtained in response to the given input is arbitrarily close to zero. FIG. 6 is a flowchart of an example learning or training process. A set of input signals are input to the neural network (Block S10). The outputs of the neural network in response to the set of input signals are compared to expected outputs to determine an error function (Block S12). The error function is compared to a threshold (Block S 14). If the error function is greater than the threshold, the neural network is not yet sufficiently trained and so the weights and biases of the neural network are adjusted (Block S16), and the process returns to Block S10. Otherwise, when the error is less than the threshold, the current weights and biases are stored (Block S 18), and the training process ends.

Once the neural network is trained, it may be used in a subsequent inference stage to produce an output for a new input signal (for which the output is not known beforehand) based on the final set of weights and biases obtained during the learning stage. For example, neural networks have been used for character recognition. During training, known characters are input to the neural network and a measure of how well the neural network has correctly identified the character is used to determine a new set of weights and biases. The process is repeated until a satisfactory level of recognition is achieved. Once trained, the neural network can be used to recognize characters for which it was trained, even if the characters to be recognized are in a different font or size. This is but one example of machine learning (ML) using neural networks. Often, it is not apparent what set of training signals should be employed to train a neural network to solve a particular problem.

SUMMARY

Some embodiments advantageously provide a method and system for detection of passive intermodulation (PIM) by a machine learning system for radio networks, e.g., Fifth Generation (5G) and Fourth Generation (4G) radio networks.

According to one aspect, a method for detecting passive intermodulation, PIM, in a signal received at a network node is provided. The method includes detecting a presence of PIM in the received signal by inputting signal statistics of a Fourier transform of the received signal to a neural network trained to determine a presence of PIM.

According to this aspect, in some embodiments, the method further includes providing a determination of PIM presence to a PIM cancellation unit configured to cancel PIM from the received signal. In some embodiments, the method also includes switching between (a) PIM cancellation based on detecting a PIM presence by applying the neural network to the received signal and (b) PIM cancellation based on detecting a PIM presence by applying a non-neural network based detection algorithm to the received signal. In some embodiments, the method also includes simultaneously performing (a) PIM detection by applying the neural network to the received signal and (b) PIM detection by applying a non-neural network based detection algorithm to the received signal. In some embodiments, the Fourier transform is a short time Fourier transform, STFT, and the signal statistics input to the neural network are based at least in part on a magnitude squared of the STFT. In some embodiments, the signal statistics include at least one of a mean value, a standard deviation, a signal maximum and a signal minimum, of a sub band of the Fourier transform of the received signal. In some embodiments, the neural network is trained by a plurality of test signals, at least some of the test signals having PIM in a frequency band encompassing a frequency of the received signal. In some embodiments, the neural network is trained by a plurality of test signals, at least some of the test signals having PIM of a highest order sought to be cancelled from the received signal. In some embodiments, the neural network is trained by a plurality of test signals, at least some of the test signals having PIM of a highest order sought to be detected in the received signal. In some embodiments, the neural network is configured to apply a first set of predetermined weights and biases to the inputted signal statistics, the predetermined weights and biases being determined by training the neural network.

According to another aspect, a network node is configured to detect passive intermodulation, PIM, in a signal received by the network node, the network node comprising processing circuitry configured to detect a presence of PIM in the received signal by inputting signal statistics of a Fourier transform of the received signal to a neural network trained to determine a presence of PIM. According to this aspect, in some embodiments, the processing circuitry is further configured to provide a determination of PIM presence to a PIM cancellation unit configured to cancel PIM from the received signal. In some embodiments, the processing circuitry is further configured to switch between (a) PIM cancellation based on detecting a PIM presence by applying the neural network to the received signal and (b) PIM cancellation based on detecting a PIM presence by applying a non- neural network based detection algorithm to the received signal. In some embodiments, the processing circuitry is further configured to simultaneously perform (a) PIM detection by applying the neural network to the received signal and (b) PIM detection by applying a non-neural network based detection algorithm based at least in part on a transmit signal corresponding to the received signal. In some embodiments, the Fourier transform is a short time Fourier transform, STFT, and the signal statistics input to the neural network are based at least in part on a magnitude squared of the STFT. In some embodiments, the signal statistics include at least one of a mean value, a standard deviation, a signal maximum and a signal minimum, of a sub band of the Fourier transform of the received signal. In some embodiments, the neural network is trained by a plurality of test signals, at least some of the test signals having PIM in a frequency band encompassing a frequency of the received signal. In some embodiments, the neural network is trained by a plurality of test signals, at least some of the test signals having PIM of a highest order sought to be cancelled from the received signal. In some embodiments, the neural network is trained by a plurality of test signals, at least some of the test signals having PIM of a highest order sought to be detected in the received signal. In some embodiments, the neural network is configured to apply a first set of predetermined weights and biases to the inputted signal statistics, the predetermined weights and biases being determined by training the neural network.

BRIEF DESCRIPTION OF THE DRAWINGS

A more complete understanding of the present embodiments, and the attendant advantages and features thereof, will be more readily understood by reference to the following detailed description when considered in conjunction with the accompanying drawings wherein: FIG. 1 is an illustration of one example of a wireless communication system;

FIG. 2 is a block diagram of a network node and a wireless device;

FIG. 3 is a conceptual illustration of one example of a neural network;

FIG. 4 is a conceptual illustration of one example of a component of a hidden or output layer of a neural network;

FIG. 5A is a graph of a first example of an activation function;

FIG. 5B is a graph of a second example of an activation function;

FIG. 6 is flowchart of one example of a neural network training process;

FIG. 7 is a block diagram of an example radio interface having a neural network for detecting PIM;

FIG. 8 is a flowchart of one example of a process for determining whether to issue an indication of PIM presence;

FIG. 9 is a flowchart of one example of a process for preparing and applying test data sets to train a neural network;

FIG. 10 is a graph of one example of a Fourier transform of a test signal;

FIG. 11 is a flowchart of one example of a process for applying a received signal to a neural network trained to detect PIM; and

FIG. 12 is a flowchart of one example of a process for detecting PIM.

DETAILED DESCRIPTION

Before describing in detail exemplary embodiments, it is noted that the embodiments reside primarily in combinations of apparatus components and processing steps related to detection of passive intermodulation (PIM) by a machine learning system for Fifth Generation (5G) and Fourth Generation (4G) radio networks. Accordingly, components have been represented where appropriate by conventional symbols in the drawings, showing only those specific details that are pertinent to understanding the embodiments so as not to obscure the disclosure with details that will be readily apparent to those of ordinary skill in the art having the benefit of the description herein.

As used herein, relational terms, such as “first” and “second,” “top” and “bottom,” and the like, may be used solely to distinguish one entity or element from another entity or element without necessarily requiring or implying any physical or logical relationship or order between such entities or elements. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the concepts described herein. As used herein, the singular forms “a”, “an” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms “comprises,” “comprising,” “includes” and/or “including” when used herein, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

In embodiments described herein, the joining term, “in communication with” and the like, may be used to indicate electrical or data communication, which may be accomplished by physical contact, induction, electromagnetic radiation, radio signaling, infrared signaling or optical signaling, for example. One having ordinary skill in the art will appreciate that multiple components may interoperate and modifications and variations are possible of achieving the electrical and data communication.

In some embodiments described herein, the term “coupled,” “connected,” and the like, may be used herein to indicate a connection, although not necessarily directly, and may include wired and/or wireless connections.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the concepts described herein. As used herein, the singular forms “a”, “an” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms “comprises,” “comprising,” “includes” and/or “including” when used herein, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

The term “network node” used herein can be any kind of network node comprised in a radio network which may further comprise any of base station (BS), radio base station, base transceiver station (BTS), base station controller (BSC), radio network controller (RNC), g Node B (gNB), evolved Node B (eNB or eNodeB), Node B, multi- standard radio (MSR) radio node such as MSR BS, multi-cell/multicast coordination entity (MCE), relay node, donor node controlling relay, radio access point (AP), transmission points, transmission nodes, Remote Radio Unit (RRU) Remote Radio Head (RRH), a core network node (e.g., mobile management entity (MME), self-organizing network (SON) node, a coordinating node, positioning node, MDT node, etc.), an external node (e.g., 3rd party node, a node external to the current network), nodes in distributed antenna system (DAS), a spectrum access system (SAS) node, an element management system (EMS), etc. The network node may also comprise test equipment. The term “radio node” used herein may be used to also denote a wireless device (WD) such as a wireless device (WD) or a radio network node.

In some embodiments, the non-limiting terms wireless device (WD) or a user equipment (UE) are used interchangeably. The WD herein can be any type of wireless device capable of communicating with a network node or another WD over radio signals, such as wireless device (WD). The WD may also be a radio communication device, target device, device to device (D2D) WD, machine type WD or WD capable of machine to machine communication (M2M), low-cost and/or low-complexity WD, a sensor equipped with WD, Tablet, mobile terminals, smart phone, laptop embedded equipped (LEE), laptop mounted equipment (LME), USB dongles, Customer Premises Equipment (CPE), an Internet of Things (loT) device, or a Narrowband loT (NB-IOT) device etc.

Also, in some embodiments the generic term “radio network node” is used. It can be any kind of a radio network node which may comprise any of base station, radio base station, base transceiver station, base station controller, network controller, RNC, evolved Node B (eNB), Node B, gNB, Multi-cell/multicast Coordination Entity (MCE), relay node, access point, radio access point, Remote Radio Unit (RRU) Remote Radio Head (RRH).

Note that although terminology from one particular wireless system, such as, for example, 3GPP LTE and/or New Radio (NR), may be used in this disclosure, this should not be seen as limiting the scope of the disclosure to only the aforementioned system. Other wireless systems, including without limitation Wide Band Code Division Multiple Access (WCDMA), Worldwide Interoperability for Microwave Access (WiMax), Ultra Mobile Broadband (UMB) and Global System for Mobile Communications (GSM), may also benefit from exploiting the ideas covered within this disclosure.

Note further, that functions described herein as being performed by a wireless device or a network node may be distributed over a plurality of wireless devices and/or network nodes. In other words, it is contemplated that the functions of the network node and wireless device described herein are not limited to performance by a single physical device and, in fact, can be distributed among several physical devices.

Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. It will be further understood that terms used herein should be interpreted as having a meaning that is consistent with their meaning in the context of this specification and the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

In some embodiments, methods are provided for detection of passive intermodulation (PIM) by a machine learning system for radio networks, e.g., Fifth Generation (5G) and Fourth Generation (4G) radio networks. Returning now to the drawing figures, in which like elements are referred to by like reference numerals, there is shown in FIG. 7 a block diagram of a radio interface 80 of a network node 78 configured to use a neural network to detect PIM. In the transmit path of the radio interface 80, baseband signals may be received from the processing circuitry 102 and digitally upconverted in frequency by the digital upconverter (DUC) 84 and input to cross factor reduction (CFR) unit 86. CFR unit 86 is employed to reduce the peak to average power ratio (PAPR) of the signal to be transmitted. The output of the CFR unit 86 is input to the PIMC 58, which may apply PIM detection and cancellation techniques to the transmit signal. The output of the CFR unit 86 is also input to the digital pre-distorter (DPD) 88, which pre-distorts the signal to be transmitted to compensate for distortion caused by the RF power amplifier of the RF transmitter front end 90 (which also includes a digital to analog converter (DAC)). The output of the RF transmitter front end 90 is coupled to the antennas 43 which radiate the transmit signal over the air. In the receive path, an RF receiver front end 92 (which includes an RF amplifier and analog to digital converter (ADC)), amplifies a signal received by the antennas 43 and outputs a digital received signal. The digital received signal is input to the PIMC 58 and to a neural network 94. The neural network 94 is trained to detect PIM in the digital received signal. The neural network 94 may be a convolutional neural network (CNN), although some embodiments may employ other classes of neural networks. If PIM is detected by the neural network 94, an indication of PIM presence is sent to the PIMC 58 and a switch 96. In some embodiments, in response to receiving the indication of PIM presence from the neural network 94, the PIMC 58 may produce a cancellation signal to cancel PIM from the digital received signal. The cancellation signal is input to the switch 96. The output of the switch 96 is input to a subtractor 98 which determines the difference between the cancellation signal from the switch 96 and the digital received signal from the RF receiver front end 92. The output of the subtractor 98 is input to the digital downconverter (DDC) 100, which down converts the signal from the subtractor 98. The down converted signal from the DDC 100 is input to the processing circuitry 102, which may be configured to perform baseband signal processing. The state of the switch 96 may be controlled by a PIM detection decision process.

FIG. 8 is a flowchart of one example of such a process for making a PIM detection decision. The process may be performed at least in part by processing circuitry 102 and/or processing circuitry 104. The process includes performing applying an alternative PIM detection algorithm to the receive signal (Block S20), and simultaneously performing PIM detection on the received signal using the neural network 94 (Block S22). The alternative PIM detection algorithm is not neural network based, but rather, may be an adaptive filtering, least squares or least mean square algorithms. The alternative PIM detection technique indicates whether PIM is detected in the receive signal (Block S24). If PIM is not detected by the alternative PIM detection, then the process of transmission signal-based PIM detection continues (Block S20). If PIM is detected by the alternative PIM detection algorithm, then a positive indication is sent to the PIM decision Block S26.

Similarly, if PIM is not detected by the neural network 94 (Block S28), the neural network 94 continues to process input to determine whether PIM is detected in the received signal (Block S22). If PIM is detected in the received signal by the neural network 94 (Block S28), then a positive indication is sent to the PIM decision Block S26. The PIM decision Block S26, may be implemented by the processing circuitry 102 and/or processing circuitry 104 to output an indication of PIM presence to the PIMC 58 when one or both of the following occurs: (1) PIM is detected by the neural network 94 and/or (2) PIM is detected by the alternative PIM detection techniques applied by PIMC 58 to the receive signal. In some embodiments, the PIM decision Block S26 compares a result of the PIM detection applied to the transmit signal to a result of the PIM detection by the neural network 94. In some embodiments, the PIM decision Block S26 switches between PIM cancellation based on a PIM presence detected by the neural network 94 and PIM cancellation based on a PIM presence detected in the received signal by the PIMC 58.

Note that some or all of the processing related to PIM detection and cancellation may be performed by the processing circuitry 102 residing external to the radio interface 80 or by processing circuitry 104 residing within the radio interface 80 closer to the RF transmitter and receiver front ends 90 and 92. Note also that the processor 50 of network node 16 can be programmed to perform some or all of the functions of processing circuitry 102 and/or 104. In other words, some embodiments may include modifying network node 16 to perform the functions described herein as being performed by the processing circuitry 102 and/or 104 and/or the neural network 94.

FIG. 9 is a flowchart of one example process for training the neural network 94. The process may be performed at least in part by processing circuitry 102 and/or processing circuitry 104. The process includes taking the Fourier transform of test signals on defined frequency bands (Block S30). The test signals may include carrier frequencies plus noise plus PIM. An example of the Fourier transform of one example test signal is shown in FIG. 10. The Fourier transform of a test signal may be performed using a Fast Fourier Transform (FFT) algorithm. Then the mean, variance (or standard deviation), maximum and minimum of the frequency domain test signal may be computed over each of one or more frequency bands (Block S32). Using the frequency domain representation of an example test signal shown in FIG. 10, the frequency axis may be divided into three regions: [-1.9, -0.7], [-0.5, 0.5] and [0.7, 1.9], (using normalized frequency), and the mean, variance (or standard deviation), maximum and minimum may be computed separately over each region. As used herein, the determined mean, variance (or standard deviation), maximum and minimum in a region, are referred to as the signal statistics of the region. Note that regions other than the three regions given above can be specified for computing the signal statistics in each region. For each test signal, a data set is created (Block S34) that has the determined mean, variance (or standard deviation), maximum and minimum, and a label indicating whether PIM is present in the test signal or not, e.g.: the data set, S={FFT_mean, FFT_SD, FFT_max, FFT_min, PIM_label}. Note that some test signals may have no PIM. The collection of data sets for signals with and without PIM are used to train the neural network 94 (Block S36).

As an alternative to training the neural network 94 by inputting the mean, variance (or standard deviation), maximum and minimum of the Fourier transform of a test signal, the neural network 94 can be trained by inputting a spectrogram of a short time Fourier transform of a test signal. A short time Fourier transform involves taking the Fourier transform of the input signal over a time window. The magnitude squared of the short time Fourier transform of the test signal is the spectrogram of the short time Fourier transform as a function of position of the time window. The neural network 94 may be trained by inputting the spectrogram of different test signals to the neural network 94.

Once the neural network 94 is trained, the neural network 94 may be used to determine whether a received signal has PIM. Such a process is shown in FIG. 11. A Fourier transform is determined for the digital received signal (Block S38). The Fourier transform may be performed by the processing circuitry 102 and/or the processing circuitry 104. The signal statistics for the Fourier transform of the digital received signal are computed via the processing circuitry 102 and/or the processing circuitry 104 (Block S40). The signal statistics may include the mean, standard deviation, maximum and minimum, for example. The computed signal statistics are input to the neural network 94 (Block S42), which outputs an indication of PIM presence (Block S44). The indication of PIM presence may be binary (yes or no) or may be a vector of PIM presence indications for PIM of different orders such as {PIM3, PIM5, PIM7...}. Note that the number of layers of the neural network 94 may be determined by experimentation or selected based on experience. For example, a convolutional neural network having four hidden layers may be able to detect PIM based on the four signal statistics of mean, standard deviation, maximum and minimum, or based on the spectrogram of the received signal, depending on how the neural network 94 is trained.

FIG. 12 is a flowchart of one example process for detecting PIM in a signal received by a network node. The process may be performed by the processing circuitry 102 and/or the processing circuitry 104. The process includes detecting a presence of PIM in the received signal by inputting signal statistics of a Fourier transform of the received signal to a neural network trained to determine a presence of PIM (Block S46).

It is noted that, although the method described above refers to the network node 78 as detecting the PIM, it is understood that such description is provided as a non-limiting example, and that other network elements can determine the PIM in the signal received by a network node 78. In other words, it is contemplated that the processing to detecting the PIM by the neural network 94 can occur in a node in other than the network node 78 receiving the signal. As an example, a distributed processing system can be implemented to incorporate the neural network 94 and perform the functions described herein.

In the embodiment shown in FIG. 7, the processing circuitry 102 and/or 104 may comprise integrated circuitry for processing and/or control, e.g., one or more processors and/or processor cores and/or FPGAs (Field Programmable Gate Array) and/or ASICs (Application Specific Integrated Circuitry) adapted to execute instructions. A processor of the processing circuitry 102, 104 may be configured to access (e.g., write to and/or read from) a memory local to the processing circuitry 102, 104, which may comprise any kind of volatile and/or nonvolatile memory, e.g., cache and/or buffer memory and/or RAM (Random Access Memory) and/or ROM (Read- Only Memory) and/or optical memory and/or EPROM (Erasable Programmable Read-Only Memory).

The processing circuitry 102 may be configured to control any of the methods and/or processes described herein and/or to cause such methods, and/or processes to be performed, e.g., by network node 78. The processing circuitry 102 is in communication with the transceiver 82 of the radio interface 80. The transceiver 82 includes a transmitter for transmitting signals to WDs 22 and a receiver for receiving signals from WDs 22, via the antennas 43. In some embodiments, neural network 94 can be used by network node 78 to provide PIM cancellation data to a network node, such as network node 16 for use by PIMC 58.

According to one aspect, a method for detecting passive intermodulation, PIM, in a signal received at a network node 78 is provided. The method includes detecting a presence of PIM in the received signal by inputting signal statistics of a Fourier transform of the received signal to a neural network 94 trained to determine a presence of PIM.

According to this aspect, in some embodiments, the method further includes providing a determination of PIM presence to a PIM cancellation (PIMC) unit 58 configured to cancel PIM from the received signal. In some embodiments, the method also includes switching, via the switch 96, between (a) PIM cancellation based on detecting a PIM presence by applying the neural network 94 to the received signal and (b) PIM cancellation based on detecting a PIM presence by applying a non- neural network based detection algorithm to the received signal. In some embodiments, the method also includes simultaneously performing (a) PIM detection by applying the neural network 94 to the received signal and (b) PIM detection by applying a non-neural network based detection algorithm to the received signal. In some embodiments, the Fourier transform is a short time Fourier transform, STFT, and the signal statistics input to the neural network are based at least in part on a magnitude squared of the STFT. In some embodiments, the signal statistics include at least one of a mean value, a standard deviation, a signal maximum and a signal minimum, of a sub band of the Fourier transform of the received signal. In some embodiments, the neural network 94 is trained by a plurality of test signals, at least some of the test signals having PIM in a frequency band encompassing a frequency of the received signal. In some embodiments, the neural network 94 is trained by a plurality of test signals, at least some of the test signals having PIM of a highest order sought to be cancelled from the received signal. In some embodiments, the neural network 94 is trained by a plurality of test signals, at least some of the test signals having PIM of a highest order sought to be detected in the received signal. In some embodiments, the neural network 94 is configured to apply a first set of predetermined weights and biases to the inputted signal statistics, the predetermined weights and biases being determined by training the neural network 94.

According to another aspect, a network node 78 is configured to detect passive intermodulation, PIM, in a signal received by the network node 78, the network node 78 comprising processing circuitry 102, 104, configured to detect a presence of PIM in the received signal by inputting signal statistics of a Fourier transform of the received signal to a neural network 94 trained to determine a presence of PIM.

According to this aspect, in some embodiments, the processing circuitry 102, 104, is further configured to provide a determination of PIM presence to a PIM cancellation unit configured to cancel PIM the received signal. In some embodiments, the processing circuitry 102, 104, is further configured to switch between (a) PIM cancellation based on detecting a PIM presence by applying the neural network 94 to the received signal and (b) PIM cancellation based on detecting a PIM presence by applying a non-neural network based detection algorithm to the received signal. In some embodiments, the processing circuitry 102, 104, is further configured to simultaneously perform (a) PIM detection by applying the neural network 94 to the received signal and (b) PIM detection by applying a non-neural network based detection algorithm to the received signal. In some embodiments, the Fourier transform is a short time Fourier transform, STFT, and the signal statistics input to the neural network are based at least in part on a magnitude squared of the STFT. In some embodiments, the signal statistics include at least one of a mean value, a standard deviation, a signal maximum and a signal minimum, of a sub band of the Fourier transform of the received signal. In some embodiments, the neural network 94 is trained by a plurality of test signals, at least some of the test signals having PIM in a frequency band encompassing a frequency of the received signal. In some embodiments, the neural network 94 is trained by a plurality of test signals, at least some of the test signals having PIM of a highest order sought to be cancelled from the received signal. In some embodiments, the neural network 94 is trained by a plurality of test signals, at least some of the test signals having PIM of a highest order sought to be detected in the received signal. In some embodiments, the neural network 94 is configured to apply a first set of predetermined weights and biases to the inputted signal statistics, the predetermined weights and biases being determined by training the neural network 94.

As will be appreciated by one of skill in the art, the concepts described herein may be embodied as a method, data processing system, and/or computer program product. Accordingly, the concepts described herein may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects all generally referred to herein as a “circuit” or “module.” Furthermore, the disclosure may take the form of a computer program product on a tangible computer usable storage medium having computer program code embodied in the medium that can be executed by a computer. Any suitable tangible computer readable medium may be utilized including hard disks, CD-ROMs, electronic storage devices, optical storage devices, or magnetic storage devices.

Some embodiments are described herein with reference to flowchart illustrations and/or block diagrams of methods, systems and computer program products. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer readable memory or storage medium that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instruction means which implement the function/act specified in the flowchart and/or block diagram block or blocks. The computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. It is to be understood that the functions/acts noted in the blocks may occur out of the order noted in the operational illustrations. For example, two blocks shown in succession may in fact be executed substantially concurrently or the blocks may sometimes be executed in the reverse order, depending upon the functionality/acts involved. Although some of the diagrams include arrows on communication paths to show a primary direction of communication, it is to be understood that communication may occur in the opposite direction to the depicted arrows.

Computer program code for carrying out operations of the concepts described herein may be written in an object oriented programming language such as Python, Java® or C++. However, the computer program code for carrying out operations of the disclosure may also be written in conventional procedural programming languages, such as the "C" programming language. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or in the Cloud. In the latter scenario, the remote computer may be connected to the user's computer through a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider).

Many different embodiments have been disclosed herein, in connection with the above description and the drawings. It will be understood that it would be unduly repetitious and obfuscating to literally describe and illustrate every combination and subcombination of these embodiments. Accordingly, all embodiments can be combined in any way and/or combination, and the present specification, including the drawings, shall be construed to constitute a complete written description of all combinations and subcombinations of the embodiments described herein, and of the manner and process of making and using them, and shall support claims to any such combination or subcombination.

Some abbreviation used herein are explained as follows:

CNN Convolutional Neural Network

PIM Passive Inter-modulation

PIMD PIM Detection

PIMC PIM Cancellation

It will be appreciated by persons skilled in the art that the embodiments described herein are not limited to what has been particularly shown and described herein above. In addition, unless mention was made above to the contrary, it should be noted that all of the accompanying drawings are not to scale. A variety of modifications and variations are possible in light of the above teachings without departing from the scope of the following claims.