Communication system, network device, terminal device, and method to estimate channel state information

Abstract

The communication system configured to estimate Channel State Information (CSI), based on channel estimator transformer model, wherein communication system comprises network device comprising network controller and network radio link, and terminal device comprising terminal controller and terminal radio link, wherein network controller is configured to generate training pilot pattern that includes all potential inference pilot patterns, transmit training pilot pattern to terminal device and transmit pilot signals according to training pilot pattern. Moreover, the terminal controller is configured to receive training pilot pattern, map training pilot pattern set into a positional encoding indices set utilising one-to-one mapping function, receive transmitted pilot signals, and train channel estimator transformer model by inputting the superposition of received pilot sequence of training pilot pattern and positional encoding sequence generated over positional encoding indices set.

Description

[0001]COMMUNICATION SYSTEM, NETWORK DEVICE, TERMINAL DEVICE, AND METHOD TO ESTIMATECHANNEL STATE INFORMATION TECHNICAL FIELDThe present disclosure relates generally to the field of wireless communication systems and more specifically, to acommunication system and a method for the communication system configured to estimate channel state information (CSI)based on the channel estimator transformer model. Furthermore, the present disclosure specifically relates to a network deviceand a terminal device configured to operate in the communication system for estimating the CSI based on a channel estimatortransformer model. BACKGROUND Channel estimation based on artificial intelligence represents a promising advancement in communication systems. In channelestimation, a transmitting node transmits predefined signals or pilot signals, which are known by both the transmitter andreceiver sides and propagate through the surrounding wireless channel environment. Moreover, a receiving node receives thepilot signals by using prior knowledge of pre-defined transmitted predefined signals in order to estimate the channel fromcorresponding received signals. Channel estimators are trained over a specific pilot pattern and are further deployed on thesame pilot pattern, which is used during the inference phase. Since time and frequency selectivity between two communicatingnodes (e.g., the transmitting node and the receiving node) could change with time, the pilot pattern has to be adapted, and thereceiving node is further configured to re-design, re-train or fine-tune the channel estimator in accordance with a new transmitted pilot pattern. Currently, certain attempts have been made to estimate the channel state information, for example, by using conventional attention-based channel estimation techniques, such as Reverse Positional Encoding (RPE) is used for Channel StateInformation (CSI) prediction that allocates first positional encoding indices in reverse order, starting from the most recent CSIinstance and going backwards to the oldest instance within the sequence. However, such conventional attention-based channelestimation techniques fail due to many reasons, such as limited length of sequence pattern, dynamic input sequence patterns,and the like. Thus, there exists a technical problem of how to provide a channel estimation that can adapt to different transmittedpilot patterns without retraining or fine-tuning the channel estimator model. Therefore, in light of the foregoing discussion, there exists a need to overcome the aforementioned drawbacks associated with the conventional communication systems and the conventional method for estimating Channel State Information (CSI) based on a channel estimator transformer model. SUMMARYThe present disclosure provides a communication system and a method for the communication system configured to estimatechannel state information (CSI) based on the channel estimator transformer model. Furthermore, the present disclosure providesa network device, and a terminal device configured to operate in the communication system for estimating the CSI based onthe channel estimator transformer model. The present disclosure provides a solution to the existing problem of how to providea channel estimation that can adapt to different transmitted pilot patterns without retraining or fine-tuning the channel estimatormodel. An objective of the present disclosure is to provide a solution that overcomes at least partially the problems encountered in theprior art and provides an improved communication system and method for estimating the CSI based on a channel estimatortransformer model. One or more objectives of the present disclosure are achieved by the solutions provided in the enclosed independent claims. Advantageous implementations of the present disclosure are further defined in the dependent claims.In one aspect, the present disclosure provides a communication system configured to estimate the CSI based on a channelestimator transformer model. The communication system includes a network device and a terminal device. The network deviceincludes a network controller and a network radio link. The terminal device includes a terminal controller and a terminal radiolink. Furthermore, the network controller is configured to generate a training pilot pattern that includes all potential inferencepilot patterns, transmit the training pilot pattern to the terminal device, and transmit pilot signals according to the training pilotpattern. Moreover, the terminal controller is configured to receive the training pilot pattern, map the training pilot pattern setinto a positional encoding index set utilising a one-to-one mapping function. The terminal controller is further configured toreceive the transmitted pilot signals and train the channel estimator transformer model by inputting the superposition of thereceived pilot sequence of the training pilot pattern and a positional encoding sequence generated over the positional encodingindices set.Advantageously, the communication system is configured to enhance the accuracy, reliability, and efficiency of channel stateinformation (CSI) estimation by using the channel estimator transformer model. By generating a training pilot pattern thatincludes all potential inference pilot patterns, the network controller is configured to ensure that the terminal device receives diverse and comprehensive training data in order to allow the channel estimator model to be trained effectively, even in complex and varying channel conditions, leading to accurate CSI estimation. The utilization of positional encoding indices, derived from the training pilot pattern is used to train the channel estimator transformer model to capture both short-term and long-term dependencies in the communication channel, which is used for maintaining high-quality data transmission. Additionally, by dynamically adapting the pilot pattern based on real-time channel conditions, the communication system is configured torespond to environmental changes for ensuring consistent and reliable performance of the communication system. As a result,the communication system is configured to handle complex signal processing operations with enhanced accuracy, robustness, and reduced computational overhead, ultimately improving the overall communication quality of the communication system.In another aspect, the present disclosure provides the network device configured to operate in the communication systemconfigured to estimate the CSI, based on the channel estimator transformer model. The network device includes the networkcontroller and the network radio link. Further, the network controller is configured to generate a training pilot pattern thatincludes all potential inference pilot patterns, transmit the training pilot pattern to the terminal device and transmit pilot signalsaccording to the training pilot pattern.Advantageously, by generating the training pilot pattern that includes all potential inference pilot patterns, the network deviceis configured to allow the channel estimator transformer model to handle a variety of pilot patterns during inference withoutrequiring retraining or fine-tuning. The network device is further configured to train the channel estimator transformer modelwith the superposition of the received pilot sequence and a positional encoding sequence, which allows the channel estimatortransformer model to improve the accuracy of CSI estimation. The training pilot pattern that includes all potential inferencepilot patterns, optimises the utilization of computational and training resources, such as by using a single channel estimatortransformer model for different pilot patterns. As a result, instead of training separate channel estimator transformer modelsfor different pilot patterns, the network device efficiently handles multiple scenarios by using single channel estimatortransformer model. In another aspect, the present disclosure provides the terminal device configured to operate in the communication systemconfigured to estimate the CSI based on the channel estimator transformer model. The terminal device includes a terminalcontroller and a terminal radio link. Further, the terminal controller is configured to receive the training pilot pattern thatincludes all potential inference pilot patterns from the network device, map the training pilot pattern set into a positional encoding index set utilising a one-to-one mapping function. The terminal controller is further configured to receive transmittedpilot signals from the network device and train the channel estimator transformer model by inputting the superposition of thereceived pilot sequence of the training pilot pattern and a positional encoding sequence generated over the positional encodingindices set.Advantageously, by receiving the training pilot pattern that includes all potential inference pilot patterns, the terminal deviceis configured to ensure that the channel estimator transformer model is trained for different pilot patterns encountered duringthe inference phase, thereby eliminating the need for retraining, fine-tuning or separate channel estimator transformer model.The terminal device is further configured to utilise a high-resolution training pilot pattern that encompasses all potentialinference pilot patterns for optimizing the resource utilisation and enhancing the accuracy of the communication system.Additionally, the ability of the terminal device to adapt to different pilot patterns without retraining the channel estimatortransformer model allows the commuciation system to maintain accuracy and reliably while performing CSI estimation.In another aspect, the present disclosure provides a method for the communication system configured to estimate the CSI basedon the channel estimator transformer model. The method comprises: the generating, by the network device, a training pilotpattern that includes all potential inference pilot patterns, transmitting, by the network device, the training pilot pattern to theterminal device and transmitting pilot signals according to the training pilot pattern. The method further includes receiving, bythe terminal device, the training pilot pattern, mapping, by the terminal device, the training pilot pattern set into a positionalencoding indices set, utilising, by the terminal device, a one-to-one mapping function, receiving, by the terminal device, thetransmitted pilot signals, and training, by the terminal device, the channel estimator transformer model by inputting thesuperposition of the received pilot sequence of the training pilot pattern and a positional encoding sequence generated over thepositional encoding indices set. The disclosed method achieves all the advantages and technical effects of the communication system. It is to be appreciated that all the aforementioned implementation forms can be combined. It has to be noted that all devices, elements, circuitry, units, and means described in the present application could be implemented in the software or hardware elements or any kind of combination thereof. All steps which are performed by thevarious entities described in the present application, as well as the functionalities described to be performed by the variousentities, are intended to mean that the respective entity is adapted to or configured to perform the respective steps andfunctionalities. Even if, in the following description of specific embodiments, a specific functionality or step to be performedby external entities is not reflected in the description of a specific detailed element of that entity that performs that specific stepor functionality, it should be clear for a skilled person that these methods and functionalities can be implemented in respectivesoftware or hardware elements or any kind of combination thereof. It will be appreciated that features of the present disclosure are susceptible to being combined in various combinations without departing from the scope of the present disclosure as defined by the appended claims. Additional aspects, advantages, features, and objects of the present disclosure would be made apparent from the drawings and the detailed description of the illustrative implementations construed in conjunction with the appended claims that follow. BRIEF DESCRIPTION OF THE DRAWINGS The summary above, as well as the following detailed description of illustrative embodiments, is better understood when read in conjunction with the appended drawings. For the purpose of illustrating the present disclosure, exemplary constructions of the disclosure are shown in the drawings. However, the present disclosure is not limited to specific methods and instrumentalities disclosed herein. Moreover, those in the art will understand that the drawings are not to scale. Wherever possible, like elements have been indicated by identical numbers. Embodiments of the present disclosure will now be described, by way of example only, with reference to the following diagrams wherein: FIG. 1 is a block diagram that depicts a communication system configured to estimate Channel State Information (CSI) based on a channel estimator transformer model, in accordance with an embodiment of the present disclosure; FIG. 2 is a flow chart that depicts a method for a communication system configured to estimate the CSI based on achannel estimator transformer model, in accordance with an embodiment of the present disclosure; FIG. 3 is a block diagram that depicts a network device configured to operate in a communication system configuredto estimate Channel State Information (CSI) based on a channel estimator transformer model, in accordance with anembodiment of the present disclosure;FIG.4 is a block diagram that depicts a terminal device configured to operate in a communication system configuredto estimate Channel State Information (CSI) based on a channel estimator transformer model, in accordance with anembodiment of the present disclosure; FIG. 5 is a diagram that depicts a dynamic positional encoding-based attention model for channel estimation, inaccordance with an embodiment of the present disclosure; and FIG. 6 is a diagram that depicts an exemplary scenario of training pilot pattern design, in accordance with an embodiment of the present disclosure. In the accompanying drawings, an underlined number is employed to represent an item over which the underlined number is positioned or an item to which the underlined number is adjacent. A non-underlined number relates to an item identified by a line linking the non-underlined number to the item. When a number is non-underlined and accompanied by an associated arrow, the non-underlined number is used to identify a general item at which the arrow is pointing. DETAILED DESCRIPTION OF EMBODIMENTS The following detailed description illustrates embodiments of the present disclosure and ways in which they can be implemented. Although some modes of carrying out the present disclosure have been disclosed, those skilled in the art wouldrecognise that other embodiments for carrying out or practising the present disclosure are also possible.FIG. 1 is a block diagram that depicts a communication system configured to estimate Channel State Information (CSI) basedon a channel estimator transformer model, in accordance with an embodiment of the present disclosure. With the reference toFIG. 1, there is shown a diagram that includes the communication system 100. The communication system 100 further includesa network device 102, a terminal device 104, and a communication network 114 configured to estimate Channel StateInformation (CSI) based on a channel estimator transformer model.The network device 102 is configured to operate in the communication system 100 that is configured to estimate the CSI basedon the channel estimator transformer model. In an implementation, the network device 102 includes a network controller 106and a network radio link 108. Examples of the network device 102 may include but are not limited to user equipment, such asa computer, a personal digital assistant, a portable computing device, or an electronic device.The terminal device 104 is configured to operate in the communication system 100 that is configured to estimate the CSI basedon the channel estimator transformer model. Examples of the terminal device 104 may include but are not limited to userequipment, such as a computer, a personal digital assistant, a portable computing device, or an electronic device.The network controller 106 is configured to generate a training pilot pattern that includes all potential inference pilot patterns,transmits the training pilot pattern to the terminal device, and pilot signals according to the training pilot pattern. Examples ofthe network controller 106 may include but are not limited to a central data processing device, a microprocessor, amicrocontroller, a complex instruction set computing (CISC) processor, an application-specific integrated circuit (ASIC) processor, a reduced instruction set (RISC) processor, a very long instruction word (VLIW) processor, a state machine, and other processors or control circuitry.The terminal controller 110 is configured to receive the training pilot pattern that includes all potential inference pilot patternsfrom the network device 102, map the training pilot pattern set into a positional encoding indices set utilizing a one-to-onemapping function receive transmitted pilot signals from the network device 102, and train the channel estimator transformermodel by inputting the superposition of the received pilot sequence of the training pilot pattern and the positional encodingsequence generated over the positional encoding indices set. Examples of the terminal controller 110 may include but are notlimited to a central data processing device, a microprocessor, a microcontroller, a complex instruction set computing (CISC) processor, an application-specific integrated circuit (ASIC) processor, a reduced instruction set (RISC) processor, a very long instruction word (VLIW) processor, a state machine, and other processors or control circuitry. There is provided the communication system 100 configured to estimate the CSI based on the channel estimator transformermodel. The CSI may include but not limited to a signal strength, interference, noise, and the like. The estimation of the CSIbased on the channel estimator transformer model is used to ensure an efficient and reliable data transmission between thenetwork device 102 and the terminal device 104.Furthermore, the communication system 100 includes the network device 102, which includes the network controller 106 andthe network radio link 108. The network controller 106 is configured to generate and transmit the training pilot pattern thatincludes all the potential inference pilot patterns. Moreover, the network radio link 108 is configured to facilitate thetransmission of the training pilot pattern to the terminal device 104. As a result, by generating and transmitting the trainingpilot pattern, which includes all potential inference pilot patterns, the communication system 100 ensures that the terminaldevice 104 receives diverse training data for accurate and robust training of the channel estimator transformer model, leadingto an improved CSI estimation.Furthermore, the communication system 100 includes the terminal device 104 that further includes the terminal controller 110and the terminal radio link 112. The terminal controller 110 is configured to receive the training pilot pattern and the transmittedpilot signals that are transmitted by the network device 102. Additionally, the terminal controller 110 of the terminal device104 is also configured to map the training pilot pattern and train the channel estimator transformer model. Moreover, theterminal radio link 112 is configured to facilitate the reception of the training pilot pattern from the network device 102. Thus,by receiving the training pilot pattern and the transmitted pilot signals from the network device 102, the terminal device 110 isconfigured to provide a comprehensive training of the channel estimator transformer model in order to provide an accurateestimation of the CSI, such as by aligning the training data with the actual pilot signals used in the communication system 100. Furthermore, the network controller 106 is configured to generate a training pilot pattern that includes all potential inferencepilot patterns. In an implementation, the network controller 106 is configured to identify all the potential inference pilot patternsthat can be used during the inference phase. Thereafter, the identified inference potential pilot patterns are used to generate thetraining pilot pattern, which is further transmitted to the terminal device 104. As a result, the network controller 106 isconfigured to ensure that the channel estimator model is trained with all the potential inference pilot patterns in order to handledifferent pilot patterns during real-world applications, thereby enhancing the overall performance and reliability of thecommunication system 100. In accordance with an embodiment, the network controller 106 is further configured to generate the selected pilot pattern based on channel conditions. In an implementation, the network controller 106 is configured to access current channel conditions, such as signal quality, interference levels, and environmental changes. After that, the network controller 106 is configured togenerate the selected pilot pattern based on the channel conditions. As a result, by dynamically adapting the pilot pattern basedon real-time channel information, the communication system 100 is configured to accurately estimate the CSI, leading to animproved generation of the selected pilot pattern based on the chanel conditions based on the real-time conditions.Furthermore, the network controller 106 is configured to transmit the training pilot pattern to the terminal device 104. Thetraining pilot pattern refers to a predefined set of pilot signals that are used during the training of the channel estimatortransformer model of the communication system 100 duirng the training phase. Moreover, the transmission of the training pilotpattern to the terminal device 104 is utilized to train the channel estimator model for an accurate and reliable CSI estimation.Furthermore, the network controller 106 is configured to transmit pilot signals according to the training pilot pattern. In otherwords, the network controller 106 is configured to generate the training pilot pattern, ensuring that the generated training pilot patterns include all the possible pilot patterns that can be used for actual data transmission (i.e., inference phase). Once thetraining pilot pattern is generated, then, after that, the network controller 106 is configured to transmit the pilot signals to theterminal device 104 based on the training pilot pattern. Therefore, by transmitting the pilot signals according to the trainingpilot pattern, the terminal device 104 is configured to provide an accurate channel estimation for reliable data transmission,especially in varying channel conditions that reduces the overall computational overhead of the communication system 100.Furthermore, the terminal controller 110 is configured to receive the training pilot pattern. In an implementation, the terminalcontroller 110 is configured to receive the training pilot pattern, which is transmitted by the network device 102. In animplementation, the received ,mtraining pilot pattern may include a set of known pilot signals that include all possible pilotpattern. Thus, the received training pilot pattern can be used to train the channel estimator transformer model for estimating the channel state information with enhanced accuracy and reduced errors within the communication system 100.Furthermore, the terminal controller 110 is configured to map the training pilot pattern set into a positional encoding indicesset utilising a one-to-one mapping function. In other words, upon receiving the training pilot pattern from the network device102, the terminal controller 110 is configured to use the one-to-one mapping function to map the training pilot pattern set intothe positional encoding indices set. The utilization of the one-to-one mapping function ensures that each element of the trainingpilot pattern is mapped uniquely with the corresponding positional encoding index. In addition, the mapped training pilot pattern set is further used as an input to the channel estimator transformer model during the training phase to ensure accurate, reliable, and efficient training of the channel estimator transformer model. Furthermore, the terminal controller 110 is configured to receive the transmitted pilot signals and train the channel estimatortransformer model by inputting the superposition of the received pilot sequence of the training pilot pattern and a positionalencoding sequence generated over the positional encoding indices set. In an implementation, the positional encoding indicesset refers to a collection of indices that maps the positions of various pilot signals within the training pilot pattern. In anotherimplementation, the terminal controller 110 is configured to superimpose the received pilot sequence with the generated positional encoding sequence. The superposition of the received pilot sequence and the positional encoding sequence allows the terminal controller 110 to train the channel estimator transformer model to adapt to different real-time transmission scenarios in order to provide an accurate channel estimation, even in varying and complex environments within the communication system 100. In accordance with an embodiment, the terminal controller 110 is further configured to generate the positional encodingsequence using the positional encoding indices set η, ^(^^^^^^^^) = ^^^^ , ^^^ , … , ^^^^ where ^^^ = ^^(^^ , : ) and ^^ ∈ η, wherePE is a positional encoding function. In an implementation, in order to generate the positional encoding sequence, the terminal controller 110 is configured to map the training pilot pattern with the positional encoding indices set using the one-to-one mapping function. Each index (^^) in the positional encoding indices set (η) is used to calculate a corresponding positionalencoding value (^^^) by using the positional encoding function. In an implementation, the positional encoding function refersto a trigonometric function that encodes the position information into a format, which is suitable for training the channel estimator transformer model. The positional encoding sequence (^(^^^^^^^^)) provides a comprehensive and detailed spatial representation of the pilot signals, enabling the channel estimator transformer model to process the positional encoding sequence of each of the pilot signal. Thus, by incorporating positional encoding, the channel estimator transformer model is trained effectively, reliably, and accurately, leading to an accurate and robust CSI estimation.In accordance with an embodiment, the PE() is the positional encoding function designed as follows ^^(^^ , 2^) =^^ ∈ η and ^^^^^^ corresponds to the dimension of the channel estimator input vector, and ^ is the index of the dimension. The positional encoding is calculated for each index (^^) inthe positional encoding indices set (η). In an implementation, the sine and cosine functions are used to ensure that the positionalencodings cover a range of frequencies, enabling the channel estimator transformer model to capture both short-term and long- term dependencies. In accordance with an embodiment, the ith channel estimator input vector corresponds to the received signal at one pilot positionand multiple spatial resources. The ith channel estimator input vector is used to capture the spatial diversity of the receivedsignal that allows the channel estimator to process signals from different spatial resources (e.g., antennas)at a particular pilotposition, leading to an accurate channel state information (CSI) estimation. In another embodiment, the ith channel estimatorinput vector corresponds to the received signal at one pilot position and multiple frequency resources (e.g., subcarriers)In accordance with an embodiment, the received pilot sequence of the training pilot pattern is superimposed with the positionalencoding sequence, and the channel estimator transformer model is trained using the superimposed pilot signals and inferencepositional encoding sequences. The superimposition of the received pilot sequence with the generated positional encodingsequence is used by the channel estimator transformer model to integrate both the actual received signal data and the positional context, which enhances the ability of the channel estimator transformer model to estimate channel conditions accurately. By virtue of using the superimposed sequence that includes both the received pilot signals and the positional encoding, the channel estimator transformer model is trained to handle the complex patterns and variations of the communication system 100. In accordance with an embodiment, the network controller 106 is further configured to generate a selected pilot pattern thatwill be used for transmission and further transmits the selected pilot pattern to the terminal device 104 and pilot signalsaccording to the selected pilot pattern (^^). The network controller 106 is configured to select and transmit pilot patterns basedon the current communication conditions dynamically in order to ensure that the terminal device 104 is configured to accurately estimate the communication channel information that is further used for maintaining an efficient and reliable data transmission. In other words, the network controller 106 analyzes the current communication environment and selects an appropriate pilot pattern (ψk) that suits the existing conditions of the channel, such as channel variability, interference, and other data rate requirements. Moreover, once the pilot pattern is selected, the network controller 106 is configured to transmit the channel information to the terminal device 104 to ensure that both the network controller 106 and the terminal device 104 are synchronized and have a common understanding of the pilot pattern that will be used. Furthermore, the terminal controller 110is configured to receive the selected pilot pattern, pilot signals and generate positional encoding indices set (γk) using the one-to-one mapping function (f) and the selected pilot pattern, generate an inference positional encoding indices set by adapting the positional encoding scheme based on the positional encoding indices set, superimpose the pilot signals with the inference positional encoding sequences, and estimate the CSI by applying the channel estimator transformer model to the superimposed pilot signals and inference positional encoding sequences. By receiving the selected pilot pattern and generating the corresponding positional encoding scheme, the terminal controller 110 is configured to adapt to varying channel conditionsand maintain high-quality data transmission for handling complex signal processing tasks with enhanced accuracy androbustness. The terminal controller 110 is configured to receive the selected pilot pattern from the network controller 106,ensuring that both the terminal device 104 and the network device 102 are aligned in terms of the pilot pattern to be used.Moreover, the terminal device 104 is configured to receive the pilot signals that are transmitted to the selected pilot pattern forchannel estimation. The terminal controller 110 is configured to use one-to-one mapping function (f) to create a set of positional encoding indices (γk) based on the selected pilot pattern for adapting the positional encoding to the specific pilot pattern usedduring transmission. Furthermore, the terminal controller 110 is configured to adapt the positional encoding set according tothe generated indices set for creating an inference positional encoding set that matches the received pilot pattern. After that, the received pilot signals are superimposed with the inference positional encoding sequences, creating a combined input that reflects both the received signals and the adapted positional encoding. Finally, the terminal controller 110 is configured to use the channel estimator transformer model to the combined input (i.e., superimposed pilot signals and inference positional encoding sequences) to estimate the CSI. As a result, by generating and adapting the positional encoding sequence to match the received pilot pattern, the terminal controller 110 ensures an accurate CSI estimation. Advantageously, the communication system 100 is configured to enhance the accuracy, reliability, and efficiency of channel state information (CSI) estimation using a channel estimator transformer model. By generating a training pilot pattern thatincludes all potential inference pilot patterns, the network controller 106 is configured to ensure that the terminal device 104receives diverse and comprehensive training data in order to allow the channel estimator model to be trained effectively, evenin complex and varying channel conditions, leading to accurate CSI estimation. The utilization of positional encoding indices,derived from the training pilot pattern, is used to train the channel estimator transformer model to capture both short-term andlong-term dependencies in the channel, which is used for maintaining high-quality data transmission. Additionally, bydynamically adapting the pilot pattern based on real-time channel conditions, the communication system 100 is configured to respond to environmental changes, ensuring consistent and reliable performance. Overall, the communication system 100 is configured to handle complex signal processing tasks with enhanced accuracy, robustness, and reduced computational overhead, ultimately improving the overall communication quality and efficiency. FIG. 2 is a flow chart that depicts a method for a communication system configured to estimate the CSI based on a channelestimator transformer model, in accordance with an embodiment of the present disclosure. With reference to FIG. 2, there isshown a flowchart of a method 200 for the communication system 100. The method 200 includes steps 202 to 214.There is provided the method 200 for the communication system 100 configured to estimate Channel State Information (CSI)based on a channel estimator transformer model. The communication system 100 includes the network device 102 and theterminal device 104. Moreover, the network device 102 includes the network controller 106 and the network radio link 108,and the terminal device 104 includes the terminal controller 110 and the terminal radio link 112. The estimation of the CSIbased on the channel estimator transformer model is used to ensure an efficient and reliable data transmission between the network device 102 and the terminal device 104.At step 202, the method 200 includes generating the training pilot pattern Ω that includes all potential inference pilot patterns.In an implementation, the network controller 106 is configured to generate the training pilot pattern that includes all potentialinference pilot patterns. At step 204, the method 200 includes transmitting the training pilot pattern Ω to the terminal device104. By virtue of sending the training pilot pattern, the method 200 is used to provide the terminal device 104 with the dataneeded to train the channel estimator. At step 206, the method 200 includes transmitting the pilot signals according to thetraining pilot pattern Ω. The training pilot pattern is used to provide all potential inference pilot patterns, ensuringcomprehensive coverage for various channel conditions and configurations. The transmission of the pilot signals according to the training pilot pattern facilitates the ability of the terminal device 104 to accurately train the channel estimator transformermodel. At step 208, the method 200 includes receiving the training pilot pattern. At step 210, the method 200 includes mappingthe training pilot pattern set into the positional encoding indices set ^ utilising a one-to-one mapping function, ^(. ), ^ = ^(Ω).By creating a set of positional encoding indices that correspond to the training pilot pattern, the terminal controller 110 ensures that the channel estimator transformer model can properly interpret and process the pilot signals. At step 212, the method 200 includes receiving the transmitted pilot signals. The transmitted pilot signals serve as the primary input for the channel estimation process. By receiving the pilot signals, the terminal device 104 can proceed with further steps, such as applying the channel estimator transformer model to the transmitted pilot signals, which is essential for deriving precise channel stateestimates. At step 214, the method 200 includes training the channel estimator transformer model by inputting the superpositionof the received pilot sequence of the training pilot pattern and the positional encoding sequence, ^(^^^^^^^^), generated over the positional encoding indices set. By training the channel estimator transformer model with this combined data, the terminal controller 110 enhances the ability of the channel estimator transformer model to estimate the CSI effectively. In accordance with an embodiment, the method 200 further includes the network controller 106 for generating a selected pilotpattern that will be used for transmission. The choice of pilot pattern is specifically based on the current channel conditionsor other criteria relevant to the communication system 100. The selected pilot pattern is used for transmission during the operational phase to gather real-time channel data. In accordance with an embodiment, the method 200 includes the networkcontroller 106 for generating a selected pilot pattern (^^) that will be used for transmission. By generating and choosing asuitable pilot pattern, the network controller 106 ensures that the terminal device 104 receives pilot signals that are mosteffective for an accurate CSI estimation. In accordance with an embodiment, the method 200 includes the network controller106 for transmitting pilot signals according to the selected pilot pattern, ^^. By using the selected pilot pattern, the networkcontroller 106 ensures that the transmitted pilot signals are tailored to the current communication scenario, thereby enhancingthe accuracy and relevance of the CSI estimation. In accordance with an embodiment, the method 200 includes the terminalcontroller for receiving the selected pilot pattern. By receiving the training pilot patterns, the terminal controller 110 is configured to map the training pilot patterns into positional encoding indices, which allows the terminal device 104 to align the channel estimator transformer model with the various patterns that might be encountered during the inference phase. In accordance with an embodiment, the method 200 includes the terminal controller 110 for receiving pilot signals and generating positional encoding indices set ^^using the one-to-one mapping function f and the selected pilot pattern. By generating the positional encoding indices set, the terminal controller 110 prepares the system for accurate CSI estimation, thereby ensuring that the channel estimator transformer model can apply the correct positional encodings to the received pilot signals. In accordance with an embodiment, the method 200 includes the terminal controller 110 for generating an inference positionalencoding scheme by adapting the positional encoding scheme based on the positional encoding indices set, superimposing thepilot signals with the inference positional encoding sequences, and estimating the CSI by applying the channel estimatortransformer model to the superimposed pilot signals and inference positional encoding sequence. Advantageously, the method 200 is configured to enhance the accuracy, reliability, and efficiency of channel state information (CSI) estimation using a channel estimator transformer model. By generating a training pilot pattern that includes all potential inference pilot patterns, the method 200 ensures that the terminal device 104 receives diverse and comprehensive training data in order to allow the channel estimator model to be trained effectively, even in complex and varying channel conditions, leading to an accurate CSI estimation. The utilization of positional encoding indices, derived from the training pilot pattern is used to train the channel estimator transformer model to capture both short-term and long-term dependencies in the channel, which is used for maintaining high-quality data transmission. The steps 202 to 214 are only illustrative, and other alternatives can also be provided where one or more steps are added, one or more steps are removed, or one or more steps are provided in a different sequence without departing from the scope of the claims herein. There is further provided a computer program product comprising program instructions for performing the method 200 whenexecuted by one or more processors in the communication system 100. The computer program product is implemented as analgorithm, embedded in software stored in a non-transitory computer-readable storage medium. The non-transitory computer-readable storage means may include but are not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. Examples of implementation of computer-readable storage medium, but are not limited to, Electrically Erasable Programmable Read-Only Memory (EEPROM), Random Access Memory (RAM), Read Only Memory (ROM), Hard Disk Drive (HDD), Flash memory, a Secure Digital (SD) card, Solid-State Drive (SSD), a computer-readable storage medium, and / or CPU cache memory.FIG. 3 is a block diagram that depicts a network device configured to operate in a communication system configured to estimateChannel State Information (CSI) based on a channel estimator transformer model, in accordance with an embodiment of thepresent disclosure. FIG. 3 is described in conjunction with elements from FIG. 1. With reference to FIG. 3, there is provided adiagram 300 that includes the network device 102. The network device 102 includes the network controller 106, the networkradio link 108, a first memory 302, and a first network interface 304.The first memory is configured to store the generated pilot patterns. Examples of implementation of the first memory 302 mayinclude but are not limited to, an Electrically Erasable Programmable Read-Only Memory (EEPROM), Dynamic Random- Access Memory (DRAM), Random Access Memory (RAM), Read-Only Memory (ROM), Hard Disk Drive (HDD), Flashmemory, a Secure Digital (SD) card, Solid-State Drive (SSD), and / or CPU cache memory.The first network interface 304 refers is configured to allow the communication between the network controller 106 and thenetwork radio link 108. Examples of the first network interface 304 may include but are not limited to a computer port, a network socket, a network interface controller (NIC), and any other network interface device.There is provided the network device 102, that is configured to operate in the communication system 100, which is configuredto estimate the CSI, based on the channel estimator transformer model. In an implementation, the channel estimator transformermodel is configured to process the pilot pattern and use the pilot patterns to estimate the current state of the channel. Byincorporating the channel estimator transformer model, the network device 102 is configured to estimate the CSI within the communication system 100.Furthermore, the network controller 106 is configured to generate a training pilot pattern that includes all potential inferencepilot patterns. In an implementation, the network controller 106 is configured to identify all potential inference pilot patterns that can be used during the inference phase. Thereafter, the identified inference potential pilot patterns are used to generate thetraining pilot pattern, which is further transmitted to the terminal device 104. As a result, the network controller 106 isconfigured to ensure that the channel estimator model is trained with all the potential inference pilot patterns in order to handle different pilot patterns during real-world applications, thereby enhancing the performance and reliability of the communicationsystem 100. Furthermore, the network controller 106 is configured to transmit the training pilot pattern to the terminal device104. In an implementation, the training pilot pattern refers to a predefined set of pilot signals that are used during the trainingphase of the channel estimator transformer model of the communication system 100. Moreover, the transmission of the trainingpilot pattern to the terminal device 104 is further utilized to train the channel estimator model in order to allow the terminal device 104 to train the channel estimator model accurately and reliably. Furthermore, the network controller 106 is configured to transmit pilot signals according to the training pilot pattern. The transmission of pilot signals according to the training pilotpattern is used by the terminal device 104 to estimate the CSI accurately. In other words, the network controller 106 isconfigured to generate the training pilot pattern, ensuring that the generated training pilot patterns includes all the possible pilotpatterns that can be used for actual data transmission (i.e., inference phase). Once the training pilot pattern is generated, then,after that, the network controller 106 is configured to transmit the pilot signals to the terminal device 104 based on the trainingpilot pattern. Therefore, by transmitting the pilot signals according to the training pilot pattern, the terminal device 104 is configured to provide an accurate channel estimation for reliable data transmission, especially in varying channel conditions that reduce the overall computational overhead of the communication system 100. In accordance with an embodiment, the network controller 106 is further configured to generate a selected pilot pattern that will be used for transmission and further transmits the selected pilot pattern to the terminal device 104 and pilot signals according to the selected pilot pattern (^^). The network controller 106 is configured to select and transmit pilot patterns based on the current communication conditions dynamically in order to ensure that the terminal device 104 is configured to accurately estimate the communication channel information that is further used for maintaining an efficient and reliable data transmission. In other words, the network controller 106 analyzes the current communication environment and selects an appropriate pilot pattern (ψk) that suits the existing conditions of the channel, such as channel variability, interference, and other data rate requirements. Moreover, once the pilot pattern is selected, the network controller 106 is configured to transmit the channel information to the terminal device 104 to ensure that both the network controller 106 and the terminal device 104 are synchronized and have a common understanding of the pilot pattern that will be used. Advantageously, by generating the training pilot pattern that includes all potential inference pilot patterns, the network device102 is configured to allow the channel estimator transformer model to handle a variety of pilot patterns during inference withoutrequiring retraining or fine-tuning. The network device 102 is further configured to train the channel estimator transformermodel with the superposition of the received pilot sequence and a positional encoding sequence, which allows the channelestimator transformer model to improve the accuracy of CSI estimation. The training pilot pattern that includes all potentialinference pilot patterns, optimises the utilization of computational and training resources, such as by using a single channelestimator transformer model for different pilot patterns. As a result, instead of training separate channel estimator transformer models for different pilot patterns, the network device efficiently handles multiple scenarios by using single channel estimator transformer model. FIG.4 is a block diagram that depicts a terminal device configured to operate in a communication system configured to estimate Channel State Information (CSI) based on a channel estimator transformer model, in accordance with an embodiment of the present disclosure. FIG.4 is described in conjunction with elements from FIG.1. With the reference to figure FIG.4, there isprovided a diagram 400, that includes the terminal device 104. The terminal device 104 includes the terminal controller 110and the terminal radio link 112. The terminal device 104 further includes a second memory 402 and a second network interface404.The second memory 402 is configured to store the generated pilot patterns.Examples of implementation of the second memory402 may include, but are not limited to, an Electrically Erasable Programmable Read-Only Memory (EEPROM), Dynamic Random-Access Memory (DRAM), Random Access Memory (RAM), Read-Only Memory (ROM), Hard Disk Drive (HDD), Flash memory, a Secure Digital (SD) card, Solid-State Drive (SSD), and / or CPU cache memory.The second network interface 404 is configured to allow the communication between the network controller 106 and thenetwork radio link 108. Examples of the second network interface 404 may include but are not limited to a computer port, anetwork socket, a network interface controller (NIC), and any other network interface device.There is provided the terminal device 104 configured to operate in the communication system 100 and configured to estimateChannel State Information, CSI, based on a channel estimator transformer model. The terminal device 104 is configured to utilise the channel estimator transformer model to analyse received pilot signals and generate accurate estimates of the CSI.Initially, the terminal device 104 is configured to receive the pilot signals transmitted by the network device 102. Further, theterminal device 104 is configured to apply the channel estimator transformer model to the input pilot signals. Furthermore, thechannel estimator transformer model to analyse the received pilot signals to capture and understand the underlying patterns and complexities of the communication channel. Moreover, based on the analysis performed by the channel estimator transformer model, the terminal device 104 estimates the channel state. Furthermore, the terminal controller 110 is configured to receive the training pilot pattern. In an implementation, the terminal controller 110 is configured to receive the training pilot pattern, which is transmitted by the network device 102. In animplementation, the received ,mtraining pilot pattern may include a set of known pilot signals that include all possible pilotpattern. Thus, the received training pilot pattern can be used to train the channel estimator transformer model for estimating the channel state information with enhanced accuracy and reduced errors within the communication system 100. Furthermore, the terminal controller 110 is configured to map the training pilot pattern set into a positional encoding indices set utilising a one-to-one mapping function. In other words, upon receiving the training pilot pattern from the network device 102, the terminal controller 110 is configured to use the one-to-one mapping function to map the training pilot pattern set intothe positional encoding indices set. The utilization of the one-to-one mapping function ensures that each element of the trainingpilot pattern is mapped uniquely with the corresponding positional encoding index. In addition, the mapped training pilot pattern set is further used as an input to the channel estimator transformer model during the training phase to ensure accurate,reliable, and efficient training of the channel estimator transformer model.Furthermore, the terminal controller 110 is configured to receive the transmitted pilot signals and train the channel estimatortransformer model by inputting the superposition of the received pilot sequence of the training pilot pattern and a positionalencoding sequence generated over the positional encoding indices set. In an implementation, the positional encoding indicesset refers to a collection of indices that maps the positions of various pilot signals within the training pilot pattern. In another implementation, the terminal controller 110 is configured to superimpose the received pilot sequence with the generated positional encoding sequence. The superposition of the received pilot sequence and the positional encoding sequence allows the terminal controller 110 to train the channel estimator transformer model to adapt to different real-time transmission scenarios in order to provide an accurate channel estimation, even in varying and complex environments within the communication system 100. Furthermore, the terminal controller 110 is configured to receive the training pilot pattern. In an implementation, the terminalcontroller 110 is configured to receive the training pilot pattern, which is transmitted by the network device 102. The trainingpilot pattern provides the terminal controller 110 with a known reference signal that can be used to understand the communication channel and further alters the transmitted signals accordingly. By analysing the received pilot pattern, the terminal controller 110 is configured to train the channel estimation model, such as for decoding the actual data transmissions with enhanced accuracy and reduced errors within the communication system 100. Furthermore, the terminal controller 110 is configured to map the training pilot pattern set into a positional encoding indices set utilising a one-to-one mapping function. In other words, upon receiving the training pilot pattern from the network device102, the terminal controller 110 is configured to use a one-to-one mapping function to map the training pilot pattern set intothe positional encoding indices set. The use of the one-to-one mapping function ensures that each element of the training pilot pattern is mapped uniquely with the corresponding positional encoding index. In addition, the mapped training pilot pattern set is further used as an input to the channel estimator transformer model during the training phase to ensure accurate, reliable, and efficient training of the channel estimator transformer model. Furthermore, the terminal controller 110 is configured to receive the transmitted pilot signals and train the channel estimatortransformer model by inputting the superposition of the received pilot sequence of the training pilot pattern and a positionalencoding sequence generated over the positional encoding indices set. In an implementation, the positional encoding indicesset is a collection of indices that map the positions of various pilot signals within the training pilot pattern. Further, each indexof the positional encoding indices set corresponds to a specific position in the pilot pattern, allowing the communication system 100 to keep track of where each signal is located relative to others. In an implementation, the terminal controller 110 is configured to superimpose the received pilot sequence with the generated positional encoding sequence. The superposition of the received pilot sequence and the positional encoding sequence allows the terminal controller 110 to effectively learn the characteristics of the communication channel and further train the channel estimator model to adapt to different transmission scenarios, ultimately improving the accuracy and reliability of the data transmission. As a result, by training with the superposition of the pilot sequence and the positional encoding, the channel estimator transformer model is used to provide an accurate channel estimations, even in varying and complex environments and adapt to different pilot patterns that may be used during real-time communication within the communication system 100. In accordance with an embodiment, the terminal controller 110 is further configured to receive a selected pilot pattern, thatwill be used for transmission from the network device 102. The selected pilot pattern refers to the specific sequence of pilotsignals that is predetermined and used within the communication system 100 to facilitate accurate channel estimation. In an implementation, upon receiving the selected pilot pattern, the terminal controller 110 is configured to use the selected pilot pattern to guide the reception and processing of the pilot signals. By guiding the reception and processing of the pilot signals, the communication system 100 ensures that the incoming pilot signals from the network device 102 are processed in a mannerthat optimizes the channel estimation. Furthermore, the terminal controller 110 is configured to receive pilot signals accordingto the selected pilot pattern, ψk, from the network device 102. In an implementation, the terminal controller 110 is configuredto receive the pilot signals by monitoring the communication channel at specific intervals or frequencies defined by the selectedpilot pattern. In another implementation, after receiving the pilot signals, the terminal controller 110 is configured to processthe pilot signals to analyse the characteristics of the communication channel. , Moreover, the terminal controller 110 isconfigured to generate the positional encoding indices set, γk, using the one-to-one mapping function, f, and the selected pilot pattern ψk, γk=f(ψk). In another words, after receiving the selected pilot pattern, the terminal controller 110 is configured to generate the positional encoding indices by using the one-to-one mapping function. In an implementation, the terminal controller 110 is configured to map each element of the selected pilot pattern with a unique index, which results in the positional encoding indices set. The mapping of each element of the selected pilot pattern with a unique index ensures that each pilotsignal is associated with a specific position within the communication system 100. In another implementation, when thepositional encoding indices set is generated, the terminal controller 110 is configured to use the positional encoding indices set to correctly interpret the pilot signals, facilitating accurate channel estimation. Furthermore, the terminal controller 110 isconfigured to generate an inference positional encoding scheme, ^(^^^^^^^^^) , by adapting the positional encoding scheme,^(^^^^^^^^) , based on the positional encoding indices set, ^ , ^(^ ^^^^^^^^^) = ^(^^^^^^^^)(^^). In another words, the terminalcontroller 110 is configured to generate the inference positional encoding scheme by utilizing the positional encoding indices set derived from the selected pilot pattern. In an implementation, the terminal controller 110 is configured to apply the positional encoding indices to the training positional encoding scheme, thereby effectively mapping the training data to the current pilot positions. Moreover, the terminal controller 110 is configured to superimpose the pilot signals, ^(^^), with the inferencepositional encoding sequences ^(^^^^^^^^^). By superimposing the pilot signals with the inference positional encodingsequences, the terminal controller 110 ensures that the channel estimator transformer model receives a signal that is not only reflective of the current channel conditions but also incorporates a robust positional structure. In accordance with an embodiment, the terminal controller 110 is configured to estimate the CSI by applying the channel estimator transformer modelto the superimposed pilot signals ^(^^) and inference positional encoding sequences ^(^^^^^^^^^). The CSI is configured toprovide a detailed knowledge about the propagations of between the network device 102 and the terminal device 104. Thechannel estimator transformer model is configured to effectively analyse the relationship between the pilot signals and the positional encoding. Further, by using the channel estimator transformer model, the terminal controller 110 is configured to provide a more precise and detailed CSI.Advantageously, by receiving the training pilot pattern that includes all potential inference pilot patterns, the terminal device104 ensures that the channel estimator transformer model is well-prepared for various pilot patterns encountered during inference phase, thereby eliminating the need for retraining or fine-tuning. The terminal device 104 is further configured to utilise a high-resolution training pilot pattern that encompasses all potential inference pilot patterns, thereby optimizing theresource utilisation and enhancing the accuracy of the communication system 100. Additionally, the ability of the terminaldevice 104 to adapt to different pilot patterns without retraining the channel estimator transformer model makes the communication system 100 highly practical for deployment in modern wireless networks, where maintaining accurate CSI is critical for optimal performance of the communication system 100. FIG. 5 is a diagram that depicts a dynamic positional encoding-based attention model for channel estimation, in accordancewith an embodiment of the present disclosure. With the reference to FIG. 5, there is provided a diagram of a dynamic positionalencoding-based attention model 500 for channel estimation.In an implementation, a pilot pattern 502 depicts the signalling of the transmitted pilot pattern, serving as a key input for thedynamic positional encoding process. At operation 504, the dynamic positional encoding-based attention model 500 isconfigured to generate adaptive positional encodings based on the transmitted pilot pattern. At operation 506, the dynamicpositional encoding-based attention model 500 is configured to receive information about the current pilot pattern and createsa corresponding encoding sequence, thereby allowing the communication system 100 to adjust to different pilot patterns duringthe inference phase. At operation 508, the dynamic positional encoding-based attention model 500 is configured to perform theinitial processing task. The initial processing tasks may include signal normalisation, noise reduction, or the application of an initial channel estimation algorithm. In an implementation, the output of the dynamic positional encoding is configured to work in conjunction with the received pilot signals, providing crucial positional context to the dynamic positional encoding-based attention model 500. At operation 510, the dynamic positional encoding-based attention model 500 is configured to process the combined input of pre-processed received pilots and the dynamic positional encodings.In an implementation, the channel estimation method using the dynamic positional encoding-based attention model 500 consistsof different steps or procedures during the training and inference phases. Furthermore, the training phase includes designingthe pilot pattern (Ω) that includes all potentially transmitted pilot patterns during the inference phase, designing the one-to-onemapping function (f(.)) that maps the training pilot pattern set into positional encoding set (η=f(Ω)), training the channelestimator by inputting the superposition of the received pilot sequence of the training pilot pattern (y(Ω)) and the positionalencoding sequence ^(^^^^^^^^) generated over the positional encoding indices set (^). Moreover, the interference phaseincludes mapping the transmitted pilot pattern set (^^) into corresponding positional encoding indices set (^^) via the designedone-to-one mapping function (^^ = ^(^^)), adapting the positional encoding sequence in accordance with, (^^ ),^(^^^^^^^^^) ^stimating the channel via the attention-based model by inputting the superposition of thereceived pilots sequence of the transmitted pilot pattern ^(^^) and the adapted positional encoding sequence (^(^^^^^^^^^)). In another implementation, the received pilot signal might go through some pre-processing stages before being superimposed with the positional encoding sequence, such as an embedding layer and / or an initial channel estimator that gives initial channel estimation at pilot positions.FIG. 6 is a diagram that depicts an exemplary scenario of training pilot pattern design, in accordance with an embodiment ofthe present disclosure. With the reference to FIG. 6, there is shown a design of pilot pattern training 600.The plurality of pilot patterns 602 represents various configurations of pilot signals used in the communication system 100.The plurality of pilot patterns 602 includes the frequencies (f1 to f12) on the vertical axis and time slots (t1 to t14) on thehorizontal axis. The first pilot pattern 602A represents the pilot resource elements scattered across the time-frequency grid. Thefirst pilot pattern 602A has pilots placed at different frequencies (i.e., f1, f2, f6) and time slots (i.e., t5, t9, t13). The scatteredarrangement of the pilot resource element allows for channel estimation across various pilot points in the time-frequencydomain, thereby providing a good balance between frequency and time-domain sampling of the channel. Further, a second pilotpattern 602B represents the pilot resource elements concentrated in specific frequency subcarriers (i.e., f3 and f7) acrossmultiple time slots. The second pilot pattern 602B configuration may be particularly useful for scenarios where certainfrequency subcarriers are known to provide more reliable channel information. Furthermore, a third pilot pattern 602Crepresents a different arrangement with pilots placed at frequencies (i.e., f2 and f7), but at different time slots compared to theother patterns. The third pilot pattern 602C configuration provides yet another sampling of the time-frequency space, potentiallycapturing channel variations that the other patterns might miss. The fourth pilot pattern 602D represents a unique arrangementof pilot resource elements with different arrangements with pilots placed at frequencies (i.e., f3, f6, f9, and f12), with the timeslots (i.e., t2 and t10). The training pilot pattern 604 represents a comprehensive design that incorporates all the pilot positions from the individualpilot patterns (602A- 602D) into a single, unified pattern. The composite pilot pattern serves as a superset of all possible pilotconfigurations that might be encountered during the inference phase of channel estimation. The training pilot pattern 604 isused for designing the positional encoding scheme and training the attention-based channel estimator model.In an implementation, the provided downlink scenario for the training of the pilot pattern further includes two phases, that isthe training phase and the interference phase. During the training phase, the training includes training pilot pattern (Ω) isdesigned to include at least all the ^ supported pilot patterns of the inference phase ^^ , ^^, … , ^^. In an example, the designoption will follow ( Ω = ⋃^ ^^^ ^^ .) The network device 102 is configured to provide the designed training pilot pattern Ω tothe receiving node, design and transmit pilot signals within the resource grid according to the designed training pilot patternset Ω. Further, the terminal device 104 is configured to design the positional encoding indices set ( ^ = {^^ , ^^ , … , ^^}) via aone-to-one mapping function (^: Ω → ^) , that maps the training pilot pattern set Ω into positional encoding indices set (^ =^(Ω)). In an example, the one-to-one mapping function can be ^ = Ω. Furthermore, the terminal device is configured to designa positional encoding sequence using the positional encoding indices set (^), (^(^^^^^^^^) = ^^^^ , ^^^ , … , ^^^^ where ^^^ =^^(^^ , : )) and ^^ ∈ η. Moreover, PE () is the positional encoding function that could be designed based on following equations: In another implementation, ^^^^^^ corresponds to the dimension of the channel estimator input vector and ^ is the index of thedimension. Furthermore, the (^^^) channel estimator input vector could be the received signal at one pilot position and thespatial dimension. The initial channel estimate at one pilot position and the (^^^^^^^) spatial dimension, and he receivedsignal of the transmitted pilot pattern positions is superimposed with the designed positional encoding sequences, (^(Ω) +^(^^^^^^^^)). Furthermore, during the interference phase, the network device 102 is configured to inform the terminal device 104 of the selected pilot pattern ^^that will be used for transmission. Further, the network device 102 transmits pilot signals within theresource grid according to the selected pilot pattern (^^), the terminal device 104 is configured to acquire the positionalencoding indices set by using the designed one to one mapping function and the selected pilot pattern (^^, ^^= ^(^^)).Furthermore, the terminal device adapts the positional encoding sequence based on the positional encoding indices set(^^), ^(^^^^^^^^^) = ^(^^^^^^^^)(^^). Moreover, the terminal device 104 uses the received signal of the transmitting pilotpattern ^^and the positional encoding sequence (^(^^^^^^^^^)) with the trained channel estimation model. The received signalof the transmitted pilot pattern positions is superimposed with the designed positional encoding sequences, ^(^^) +^(^^^^^^^^^). Modifications to embodiments of the present disclosure described in the foregoing are possible without departing from thescope of the present disclosure as defined by the accompanying claims. Expressions such as “including,” “comprising,”“incorporating,” “have,” “is” used to describe, and claim the present disclosure are intended to be construed in a non-exclusivemanner, namely allowing for items, components or elements not explicitly described also to be present. Reference to thesingular is also to be construed to relate to the plural. The word “exemplary” is used herein to mean “serving as an example,instance, or illustration.” Any embodiment described as “exemplary” is not necessarily to be construed as preferred oradvantageous over other embodiments or to exclude the incorporation of features from other embodiments. The word“optionally” is used herein to mean “is provided in some embodiments and not provided in other embodiments.” It isappreciated that certain features of the present disclosure, which are, for clarity, described in the context of separate embodiments, may also be provided in combination in a single embodiment. Conversely, various features of the invention, which are, for brevity, described in the context of a single embodiment, may also be provided separately or in any suitablecombination or as suitable in any other described embodiment of the disclosure.

Claims

CLAIMS 1. A communication system (100) configured to estimate Channel State Information (CSI), based on a channelestimator transformer model, wherein the communication system (100) comprises a network device (102) and aterminal device (104), the network device (102) comprising a network controller (106) and a network radio link(108), and the terminal device (104) comprising a terminal controller (110) and a terminal radio link (112), wherein the network controller (106) is configured to:generate a training pilot pattern that includes all potential inference pilot patterns, transmit the training pilot pattern to the terminal device (104) andtransmit pilot signals according to the training pilot pattern, whereby the terminal controller (110) is configured to: receive the training pilot pattern, map the training pilot pattern set into a positional encoding indices set utilising a one-to-one mapping function, receive the transmitted pilot signals, and train the channel estimator transformer model by inputting the superposition of:the received pilot sequence of the training pilot pattern and a positional encodingsequence generated over the positional encoding indices set.

2. The communication system (100) according to claim 1, wherein the terminal controller (110) is further configured togenerate the positional encoding sequence using the positional encoding indices set, η, wherein ^(^^^^^^^^)= ^^^^, ^^^, … , ^^^^ where ^^^= ^^(^^, : ) and ^^∈ η.

3. The communication system (100) according to claim 2, wherein PE () is the positional encoding function designedas follows:where, ^^∈ η and, ^^^^^^corresponds to the dimension of the channel estimator input vector and, ^ is the index ofthe dimension.

4. The communication system (100) according to claims 2 or 3, wherein the, ith channel estimator input vectorcorresponds to the received signal at one pilot position and multiple spatial resources.

5. The communication system (100) according to any preceding claim, the received pilot sequence of the training pilotpattern, ^(Ω) is superimposed with the positional encoding sequence, ^(^^^^^^^^): ^(Ω) + ^(^^^^^^^^), and the channel estimator transformer model is trained using o the superimposed pilot signals, ^(^^) and inference positionalencoding sequences, ^(^^^^^^^^^).

6. The communication system (100) according to any preceding claim, wherein the network controller (106) is furtherconfigured to:generate a selected pilot pattern that will be used for transmission, transmit the selected pilot pattern to the terminal device (104), andtransmit pilot signals according to the selected pilot pattern ^^, wherein the terminal controller (110) is further configured toreceive the selected pilot pattern,receive pilot signals, generate positional encoding indices set, ^^ , using the one-to-one mapping function f and theselected pilot pattern, generate an inference positional encoding scheme by adapting the positional encoding schemebased on the positional encoding indices set,superimpose the pilot signals with the inference positional encoding sequences, and estimate the CSI by applying the channel estimator transformer model to the superimposed pilot signals and inference positional encoding sequences.

7. The communication system (100) according to claim 5, wherein the network controller (106) is further configured togenerate the selected pilot pattern based on channel conditions.

8. A method (200) for a communication system (100) configured to estimate Channel State Information (CSI), basedon a channel estimator transformer model, wherein the communication system (100) comprises a network device(102) and a terminal device (104), wherein the method (200) comprising the network device (102)generating a training pilot pattern that includes all potential inference pilot patterns, transmitting the training pilot pattern to the terminal device (104) and transmitting pilot signals according to the training pilot pattern, whereby the method (200) comprising the terminal device (104)receiving the training pilot pattern, mapping the training pilot pattern set into a positional encoding indices set, ^ utilising a one-to-onemapping function, receiving the transmitted pilot signals, and training the channel estimator transformer model by inputting the superposition of: the received pilot sequence of the training pilot pattern and a positional encoding sequence generated overthe positional encoding indices set.

9. The method (200) according to claim 7, the method (200) further comprising the network device (102) generating a selected pilot pattern, ^^ , that will be used for transmission,transmitting the selected pilot pattern, ^^, to the terminal device, andtransmitting pilot signals according to the selected pilot pattern, ^^, wherein the method (200) comprisingthe terminal device (104) receiving the selected pilot pattern,receiving pilot signals, generating positional encoding indices set, ^^, using the one-to-one mapping function f and theselected pilot pattern, generating an inference positional encoding scheme by adapting the positional encoding scheme based on the positional encoding indices set superimposing the pilot signals with the inference positional encoding sequences, andestimating the CSI by applying the channel estimator transformer model to the superimposed pilot signals and inference positional encoding sequence.

10. A computer program product comprising program instructions for performing the method (200) according to claim8 or 9, when executed by one or more processors in a communication system (100).

11. A network device (102) configured to operate in the communication system (100) configured to estimate Channel State Information, CSI, based on a channel estimator transformer model, the network device (102) comprising a network controller (106) and a network radio link (108), whereinthe network controller (106) is configured togenerate a training pilot pattern that includes all potential inference pilot patterns, transmit the training pilot pattern to a terminal device (104), and transmit pilot signals according to the training pilot pattern.

12. The network device (102) according to claim 11, wherein the network controller (106) is further configured togenerate a selected pilot pattern that will be used for transmission,transmit the selected pilot pattern to the terminal device (104), andtransmit pilot signals according to the selected pilot pattern .

13. A terminal device (104) configured to operate in a communication system and configured to estimate Channel StateInformation, CSI, based on a channel estimator transformer model, wherein the terminal comprising a terminal controller (110) and a terminal radio link (112), wherein the terminal controller (110) is configured toreceive a training pilot pattern, that includes all potential inference pilot patterns from a network device(102), map the training pilot pattern set into a positional encoding indices set ^, utilising a one-to-one mappingfunction, receive transmitted pilot signals from the network device (102), andtrain the channel estimator transformer model by inputting the superposition of: the received pilot sequence of the training pilot pattern, and apositional encoding sequence generated over the positional encoding indices set.

14. The terminal device (104) according to claim 13, the terminal controller (110) is further configured to:receive a selected pilot pattern that will be used for transmission from the network device (102),receive pilot signals according to the selected pilot pattern from the network device (102),generate positional encoding indices set ^^,using the one-to-one mapping function and the selected pilot pattern, generate an inference positional encoding scheme by adapting the positional encoding scheme based on the positional encoding indices set ^^, superimpose the pilot signals with the inference positional encoding sequences, andestimate the CSI by applying the channel estimator transformer model to the superimposed pilot signalsand inference positional encoding sequences.

Images