Communication system, receiving node, transmitting node and methods to predict channel state information

Abstract

A communication system configured to predict Channel State Information (CSI), based on a transformer model. The communication system comprises a transmitting node and a receiving node, the transmitting node is configured to generate a historical CSI time resource training set that the communication system is expected to handle, generate a future CSI time resource training set and transmit the historical CSI time resource training set and the future CSI time resource training set to the receiving node. In response to, the receiving node is configured to generate a historic CSI sequence and a future CSI sequence and transmit the historic CSI sequence and the future CSI sequence to the transmitting node. In response to, the transmitting node is configured to generate positional encoding sequences and then, train a transformer model configured to predict CSI based on the historic CSI sequence, the future CSI sequence, and the positional encoding sequences.

Description

[0001]COMMUNICATION SYSTEM, RECEIVING NODE, TRANSMITTING NODE AND METHODS TO PREDICTCHANNEL STATE INFORMATION TECHNICAL FIELD The present disclosure relates generally to the field of wireless communication systems; and more specifically, to acommunication system, a receiving node, a transmitting node and methods to predict channel state information (CSI) based ona transformer model.BACKGROUNDIn wireless communication systems, a radio channel serves as a conduit through which signals travel between transmitters andreceivers. The effective transmission of data relies heavily on the ability to comprehend and adapt to the conditions of the radio channel. A pivotal component in achieving this adaptation is Channel State Information (CSI), which provides a comprehensive assessment of the current state of communication links. By leveraging CSI, communication systems can dynamically adjusttransmission parameters, such as signal power, beamforming direction, and modulation scheme to optimize data throughputand reliability.The complexity arises from dynamic nature of radio environments, where channel conditions are constantly in flux due tofactors, like multipath propagation and Doppler shifts caused by movement of transmitters or receivers. To accurately gaugethese conditions, conventional wireless systems deploy pilot signals to probe the channel and estimate CSI at the receiver end.Once determined, the CSI is often fed back to the transmitter, enabling real-time adjustments to transmission strategies.However, the process of obtaining and utilizing CSI is resource-intensive, consuming significant power and bandwidth. The frequent pilot transmissions, CSI estimation, and feedback mechanisms add complexity and latency to the conventional wireless communication systems, particularly in dynamic or densely populated environments. To mitigate these challenges, predictive models based on historical CSI data, such as transformers-based CSI prediction models, have been developed. These modelsforecast future channel conditions, reducing reliance on continuous pilot transmissions and feedback loops. Still, there exists atechnical challenge of ensuring predictive accuracy across diverse operational scenarios, where historical data may not alignwith real-time conditions due to variations in pilot signal frequency, grid configuration, and other operational parameters. Thisdiscrepancy poses a significant obstacle to optimizing communication system performance and reliability. Therefore, thereremains a technical challenge in developing a transformer-based CSI predictor capable of effectively utilizing any subset of historical data, rather than being constrained to the most recent data points. Addressing this challenge would enhance therobustness and efficiency of wireless communication systems by providing more accurate and adaptable CSI predictions,ultimately improving the overall quality and reliability of wireless communications.Therefore, in light of the foregoing discussion, there exists a need to overcome the aforementioned drawbacks associated withthe conventional methods of predicting CSI in wireless communication systems.SUMMARYThe present disclosure provides a communication system, a receiving node, a transmitting node and methods to predict channelstate information (CSI) based on a transformer model. The present disclosure provides a solution to the existing problem ofdeveloping a transformer-based CSI predictor capable of working with any historical data subset, rather than being restrictedto the most recent data instances. An aim of the present disclosure is to provide a solution that overcomes at least partially theproblems encountered in the prior art, and provides an improved communication system, an improved receiving node, animproved transmitting node and improved methods to predict channel state information (CSI) based on a transformer model. The object of the present disclosure is 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 predict Channel State Information (CSI), based on a transformer model, the communication system comprises a transmitting node and a receiving node, the transmitting node comprising a transmitting controller, and the receiving node comprising a receiving controller, the transmitting controller is configured to generate a historical CSI time resource training set (Ω^) comprising historic CSI time resources (^^) that the communication system is expected to handle, generate a future CSI time resource training set (Ω^) comprising estimated future CSI time resources (^F), and transmit the historical CSI time resource training set (Ω^) and the future CSI time resource training set (Ω^) to the receiving node. In response to the receiving controller is configured to receive the historical CSI time resourcetraining set (Ω^) and the future CSI time resource training set (Ω^) from the transmitting node, and in response there to generatea historic CSI sequence (ℎ^^(^^^^^^^^)) comprising historic CSI vectors acquired at the corresponding historical CSI time resource training set (Ω^), generate a future CSI sequence (ℎ^^(^^^^^^^^)) comprising estimated CSI vectors acquired at the corresponding future CSI time resource training set (Ω^), and transmit the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence(ℎ^^(^^^^^^^^)) to the transmitting node. In response to, the transmitting controller is further configured to receive the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence from the receiving node, generate a historical positionalencoding sequence (^(^^^^^^^^)), generate a future positional encoding sequence (^(^^^^^^^^) ^^ ) and then, train a transformermodel configured to predict CSI based on the historic CSI sequence (ℎ^^(^^^^^^^^)), the future CSI sequence (ℎ^^(^^^^^^^^)), the historical positional encoding sequence (^(^^^^^^^^) ^ ), and the future positional encoding sequence (^(^^^^^^^^) ^ ). The disclosed communication system employs the transformer model capable of effectively utilizing any subset of historical data, rather than being constrained to the most recent data points in contrast to a conventional transformer-based CSI predictor. Alternatively, may be stated as, the communication system employs a unified transformer framework that is configured to support different CSI or pilot configurations. The transformer model does not require pre-training or fine tuning to supportdifferent CSI configurations. The transformer model is configured to predict the future channel conditions with enhancedaccuracy by effectively handling sequential data and complex patterns in the training data, leading to more reliable data communication. The comprehensive approach used in the communication system includes generating both historical and future CSI time resource training sets, which ensures robust training of the transformer model on a wide range of data and improves its ability to adapt to different CSI configurations. Additionally, the transformer model employs dynamic positional encoding techniques (i.e., different positional encodings between training and inference phase) that enables the transformer model to adapt to changes happening over the communication channel, and to provide reliable data communication in varying channelconditions, whereas conventional methods use static encoding that may not effectively address the continuously changingchannel conditions. Consequently, the transformer model outperforms the conventional transformer-based CSI predictortrained over a specific CSI and deployed on different CSI configurations. In an implementation form, the transmitting controller is further configured to generate the historical time resources set (Ω^)of the training phase to include any potential historic CSI time resources set of the inference phase ⊆ Ω^Incorporating the potential historic CSI time resources set (^^) into the historical time resources set (Ω^) during the training phase offers several benefits. By including data from past CSI measurements, the transmitting controller may ensure that thetransformer model is trained with a wide range of CSI scenarios, covering all possible historical conditions that thecommunication system has experienced. Such training sequences improves the ability of the transformer model to predict the CSI more accurately, even for previously unseen or rare application scenarios. As a result, the communication system becomesmore robust and adaptable, leading to enhanced performance and data communication reliability in real-time applications. In a further implementation form, the transmitting controller is further configured to generate the historical time resources set(Ω^) based on a slot resolution and where the receiving controller is further configured to acquire a CSI vector at eachsubsequent slot.By aligning the historical time resources training set (Ω^) with the slot resolution, the communication system ensures that CSIdata is captured with precise timing intervals. This approach allows more accurate and detailed modelling of channel conditions over time. In a further implementation form, the transmitting controller is further configured to generate the historical time resources set(Ω^) based on a symbol resolution and the receiving controller is further configured to acquire a CSI vector at each subsequentsymbol. With the symbol resolution, the communication system measures the historic CSI data more precisely, resulting in channel state modelling with enhanced channel information. Such detailed information allows the communication system to detect changes in the channel states for each symbol, resulting in more accurate predictions and adjustments. As a result, the communication system can more effectively manage and optimize communication parameters (i.e., bandwidth, CSI, signal-to- noise ratio (SNR), and frequency), enhancing overall performance and reliability. This approach ensures that thecommunication system is capable of responding to rapid changes and subtle fluctuations in the channel, ultimately leading toa more robust and adaptive communication. In a further implementation form, the receiving controller is further configured to acquire a CSI sequence as a result of pilot transmission and channel estimation processes. By transmitting pilot signals, the communication system is configured to obtain real-time information about the channel conditions. This leads to more precise channel estimates, making the CSI data more reliable. As a result, the communication system can make more informed decisions about resource allocation and signal adjustments. This leads to improved communication reliability and performance, as the system can effectively adapt to different CSI configurations in real time applications scenarios.In a further implementation form, the receiving controller is further configured to acquire a CSI sequence as a result of pilottransmission and channel estimation processes by the transmitting controller being further configured to transmit pilot signals(^^) during an ^^ℎ time resource, whereby the receiving controller is further configured to receive the pilot signals of the ^^ℎtime resource and estimate a corresponding CSI vector ℎ^^ = ^(^^ , ^^) where ^() is the channel estimation function.The transmitting controller is configured to transmit pilot signals at certain times, and the receiving controller is configured toutilize the pilot signals to compute the historic and future CSI data. By virtue of communication of the pilot signals, the communication system can accurately track and adapt to different channel conditions. This leads to more reliable communication, as the CSI vectors are closely aligned with actual channel conditions. In a further implementation form, the receiving controller is further configured to acquire a CSI sequence utilizing a simulation process at a communicating node using channel modelling techniques. The utilization of simulation process for acquiring the CSI sequence enables the communication system to become more flexible and outperform and ensure efficient and reliable data communication.In a further implementation form, the transmitting controller is further configured to generate the historical positional encodingsequence as ^(^^^^^^^^) ^= [^^ , ^^ , … , ^^] where ^^ = ^^(^, : ) and ^ ∈ Ω^ and a future positional encoding sequence as The utilization of dynamic positional encoding techniques for training of the transformer model enables the transformer model to support different CSI configurations. In further implementation form, the transmitting controller is further configured to, during an inference phase, generate a CSI time resource set , configure position encoding over the CSI time resource set (^^), andtransmit the CSI time resource set (^^) to the receiving node, the receiving controller is further configured to receive the CSI time resource set (^^) from the transmitting node, generate an estimated CSI sequence ) by acquiring CSI vectors(ℎ) corresponding to the CSI time resource set (^^ ) and transmit the estimated CSI sequence (ℎ^(^^^^^^^^^)^ ) to thetransmitting node, the transmitting controller is further configured to receive the estimated CSI sequence (ℎ^^(^^^^^^^^^)) and then perform CSI prediction based on the transformer model applied to the estimated CSI sequence (ℎ^^(^^^^^^^^^)). The utilization of dynamic positional encoding techniques for training of the transformer model enables the communication system to provide more efficient and reliable data communication.In a further implementation form, the transmitting controller is further configured to generate the historical time resources set(Ω^) of the inference phase to include any potential historic CSI time resources set of the inference phase By considering all possible historical CSI scenarios that may be encountered during the inference phase, the communication system manifests an enhanced robustness and adaptability to varying channel conditions. In a further implementation form, the positional encoding function during the deployment phase is adaptive to the available time resources set (^^). In a further implementation form, the transmitting controller is further configured to configure position encoding over the CSI time resource set (^^) after transmitting the CSI time resource set (^^) to the receiving node and before receiving the estimated CSI sequence By configuring positional encoding after the CSI time resource set (^^) is sent to the receiving node, the communication system ensures that the encoding is specifically tailored to the exact time resources used in the communication. This precise alignment improves the accuracy and relevance of the positional information, enhancing the ability of transformer model to more accurately interpret the estimated CSI sequence. In a further implementation form, the transmitting controller is further configured to generate a CSI time resource set (^^)having CSI observations at a first frequency for low-mobility scenarios and at a second frequency for high-mobility scenarios,where the second frequency is higher than the first frequency. Such approach leads to optimization of communication resources and efficient channel estimation.In another aspect, the present disclosure provides a method for predicting CSI, based on a transformer model in acommunication system. The communication system comprises a transmitting node and a receiving node. The method comprisesat the transmitting node, generating a historical CSI time resource training set comprising historic CSI time resources(^^) that the communication system is expected to handle, generating a future CSI time resource training set (Ω^) comprisingestimated future CSI time resources (^^), and transmitting the historical CSI time resource training set and the futureCSI time resource training set (Ω^) to the receiving node. The method further comprises at the receiving node, receiving thehistorical CSI time resource training set and the future CSI time resource training set (Ω^) from the transmitting node,and in response there to generating a historic CSI sequence (ℎ^^(^^^^^^^^)) comprising historic CSI vectors acquired at the corresponding historical CSI time resource training set , generating a future CSI sequence (ℎ^^(^^^^^^^^)) comprising estimated CSI vectors acquired at the corresponding future CSI time resource training set (Ω ), and transmitting the historic^CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^^(^^^^^^^^)) to the transmitting node. The method further comprisesat the transmitting node receiving the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^^(^^^^^^^^)) from the receiving node, generating a historical positional encoding sequence (^(^^^^^^^^) ^), generating a future positional encodingsequence and then training a transformer model configured to predict CSI based on the historic CSI sequence(ℎ^(^^^^^^^^)^(^^^^^^^^) (^^^^^^^^) ^ ), the future CSI sequence (ℎ^ ), the historical positional encoding sequence (^^), and the future positional encoding sequence (^(^^^^^^^^) ^ ). The disclosed method achieves all the advantages and technical effects of the communication system of the present disclosure.In a yet another aspect, the present disclosure provides a computer program product comprising program instructions forperforming the method, when executed by one or more processors in a communication system node.In a yet another aspect, the present disclosure provides a method for use in a receiving node in a communication, the methodbeing for predicting Channel State Information (CSI), based on a transformer model. The method comprises, receivinghistorical CSI time resource training set (Ω ) comprising historic CSI time resources (^ ) that the communication system is^ ^expected to handle from the transmitting node, receiving future CSI time resource training set (Ω ) comprising estimated^future CSI time resources (^ ) from the transmitting node, and in response there to generating a historic CSI sequenceF( )^^^^^^^^ (ℎ^) comprising historic CSI vectors acquired at the corresponding historical CSI time resource training set (Ω ),^^()^^^^^^^^^generating a future CSI sequence (ℎ ) comprising estimated CSI vectors acquired at the corresponding future CSI time^() ( )^^^^^^^^ ^^^^^^^^^ ^resource training set (Ω ), and transmitting the historic CSI sequence (ℎ ) and the future CSI sequence (ℎ )^ ^ ^to the transmitting node.The method for use in the receiving node achieves all the advantages and technical effects of the communication system of thepresent disclosure.In a yet another aspect, the present disclosure provides a receiving node configured to predict CSI, based on a transformermodel. The receiving node is configured to receive historical CSI time resource training set comprising historic CSI resources )from a transmitting node, receive future CSI time resource training set (Ω ) comprising estimated future CSI^ ^.The receiving node achieves all the advantages and technical effects of the method for use in the receiving node. The method for use in the transmitting node significantly improves the accuracy of CSI predictions, as the transformer modelis trained on both past and estimated future data, allowing the transformer model to better anticipate changes in thecommunication channel. Additionally, by leveraging both historical and future information, the communication system can adapt more effectively to varying channel conditions, making the communication link more reliable. 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 the various entities described in the present application as well as the functionalities described to be performed by the various entities are intended to mean that the respective entity is adapted to or configured to perform the respective steps and functionalities. Even if, in the following description of specific embodiments, a specific functionality or step to be performed by external entities is not reflected in the description of a specific detailed element of that entity which performs that specific step or functionality, it should be clear for a skilled person that these methods and functionalities can be implemented in respective software 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 readin conjunction with the appended drawings. For the purpose of illustrating the present disclosure, exemplary constructions ofthe disclosure are shown in the drawings. However, the present disclosure is not limited to specific methods andinstrumentalities disclosed herein. Moreover, those skilled 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 network environment diagram of a communication system comprising a transmitting node and a receiving node, inaccordance with another embodiment of the present disclosure;FIG. 2A is a block diagram that illustrates various exemplary components of a transmitting node, in accordance with an embodiment of the present disclosure; FIG. 2B is a block diagram that illustrates various exemplary components of a receiving node, in accordance with an embodiment of the present disclosure; FIGs.3A and 3B collectively, is a flowchart of a method for predicting Channel State Information (CSI), based on a transformer model in a communication system, in accordance with an embodiment of the present disclosure; FIG.4 is a flowchart of a method for use in a receiving node in a communication system, in accordance with an embodiment of the present disclosure;FIGs. 5A and 5B collectively, is a flowchart of a method for use in a transmitting node, in accordance with an embodiment ofthe present disclosure; FIG. 6 is an exemplary implementation scenario of transmitted pilot grids with different structures under different channel conditions, in accordance with an embodiment of the present disclosure;FIG. 7 is an exemplary implementation scenario of transmitted pilot grids with same structures but to different type of users,in accordance with an embodiment of the present disclosure;FIG. 8 is a flowchart that depicts a series of operations required in Dynamic Positional Encoding (DPE) technique, inaccordance with an embodiment of the present disclosure; and FIG.9 is a flowchart that depicts a series of operations performed in training and inference phase of a transformer model of a communication system, 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 beimplemented. Although some modes of carrying out the present disclosure have been disclosed, those skilled in the art wouldrecognize that other embodiments for carrying out or practicing the present disclosure are also possible. FIG.1 is a network environment diagram of a communication system comprising a transmitting node and a receiving node, in accordance with an embodiment of the present disclosure. With reference to FIG.1, there is shown a communication system100 comprising a transmitting node 102 and a receiving node 104 communicating through a communication network 106. Thetransmitting node 102 comprises a transmitting controller 108, and a transformer model 110. The receiving node 104 comprisesa receiving controller 112. The communication system 100 comprising merely one transmitting node (i.e., the transmitting node102) and one receiving node (i.e., the receiving node 104) is shown in FIG. 1, for sake of brevity. However, in anotherimplementation scenario, the communication system 100 may comprise more than one transmitting nodes and receiving nodes.The transmitting node 102 may include suitable logic circuitry, interfaces, or code that is configured to communicate with thereceiving node 104 via the communication network 106 (e.g., a propagation channel). Examples of the transmitting node 102may include, but are not limited to, an Internet-of-Things (IoT) device, a smart phone, a machine type communication (MTC)device, a computing device, an evolved universal mobile telecommunications system (UMTS) terrestrial radio access (E-UTRAN) NR-dual connectivity (EN-DC) device, a server, an IoT controller, a drone, a customized hardware for wirelesstelecommunication, a transmitter, or any other portable or non-portable electronic device. In the communication system 100,the transmitting node 102 has a single antenna for communication with the receiving node 104. However, in anotherimplementation of the communication system 100, the transmitting node 102 may have more than one antenna forcommunication with the receiving node 104. The receiving node 104 may include suitable circuitry, interfaces, or code that is configured to receive one or more frequency signals from the transmitting node 102, via the communication network 106. Examples of the receiving node 104 may include, but are not limited to, an Internet-of-Things (IoT) controller, a base station, a server, a smart phone, a customized hardware for wireless telecommunication, a receiver, or any other portable or non-portable electronic device. In the communication system 100, the receiving node 104 has a single antenna for communication with the transmitting node 102. However, in another implementation scenario of the communication system 100, the receiving node 104 may have more than one antenna for communication with the transmitting node 102.The communication network 106 includes a medium, such as a communication channel, through which the transmitting node102 potentially communicates with the receiving node 104. Examples of the communication network 106 may include, but arenot limited to, a cellular network (e.g., a 5G, or 5G NR network, such as sub 6 GHz, cmWave, or mmWave communicationnetwork), a cloud network, a Local Area Network (LAN), a vehicle-to-network (V2N) network, a Metropolitan Area Network (MAN), and / or the Internet. The transmitting node 102 is configured to connect to the receiving node 104, in accordance with various wireless communication protocols. Examples of such wireless communication protocols, communication standards, and technologies may include, but are not limited to, IEEE 802.11, 802.11p, 802.15, 802.16, 1609, Worldwide Interoperability for Microwave Access (Wi-MAX), Long-term Evolution (LTE), Voice over Internet Protocol (VoIP), a protocol for email, instant messaging, and / or Short Message Service (SMS), and / or other cellular or IoT communication protocols. The transmitting controller 108 may include suitable circuitry, interfaces, or code that is configured to generate a historical CSItime resources training set (Ω^) comprising historic CSI time resource (^^) sets that the communication system 100 isexpected to handle. Examples of the transmitting controller 108 may include, but are not limited to, an encoder, a modulator,an integrated circuit, a co-processor, 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 longinstruction word (VLIW) processor, a central processing unit (CPU), a data processing unit, and other processors or circuits.Moreover, the transmitting controller 108 may refer to one or more individual processors, processing devices, a processing unit that is a part of the transmitting node 102. The transformer model 110 may correspond to a neural network architecture primarily designed for resolving the tasksinvolving sequential data, such as natural language processing (NLP), but nowadays, the transformer model 110 has also beenapplied to other fields, like wireless communication, including CSI prediction. The transformer model 110 may be configuredto use an attention mechanism to weigh the significance of different parts of an input data. In real-time, the communicationsystem 100 often deal with shifts in network quality, interference, or other factors that can significantly affect the quality ofdata transmission. The transformer model 110 may be configured to process incoming data quickly and efficiently, allowingthe communication system 100 to understand these changes and adapt to these changes, accordingly. The transformer model110 may be configured to use deep learning techniques to understand complex patterns in the input data. Examples of the transformer model 110 may include, but are not limited to, Bidirectional Encoder Representations from Transformers (BERT), Generative Pre-trained Transformer (GPT), Text-to-Text Transfer Transformer (T5), and the like. In the transmitting node 102,the transformer model 110 leverages its ability to process and analyze complex patterns to enhance signal quality, adapt todynamic channel conditions, and efficiently managing data communication. The transformer model 110 is shown as a part ofthe transmitting node 102 in the FIG.1. However, the transformer model 110 may be an independent unit and may not be a part of the transmitting node 102. The receiving controller 112 may include suitable circuitry, interfaces, or code that is configured to receive the historical CSI time resource training set (Ω^) and the future CSI time resource training set (Ω^) from the transmitting node 102. Examples of the receiving controller 112 may include, but are not limited to, a decoder, a demodulator, an integrated circuit, a co- processor, 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 central processing unit (CPU), a data processing unit, and other processors or circuits. Moreover, the receiving controller 112 may refer to one or more individual processors, processing devices, a processing unit that is a part of the receiving node 104. In operation, the communication system 100 configured to predict CSI, based on the transformer model 110. The communication system 100 comprises the transmitting node 102 and the receiving node 104. The transmitting node 102comprising the transmitting controller 108, and the receiving node 104 comprising the receiving controller 112. Thetransmitting controller 108 is configured to, during a training phase, generate a historical CSI time resources training set (Ω^) comprising historic CSI time resources (^^) sets that the communication system 100 is expected to handle. During thetraining phase of the transformer model 110, the transmitting controller 108 is configured to generate a set of historical data,may be referred to as the historical CSI time resources training set (Ω^) which may include various examples of past channelconditions that the communication system 100 may encounter during wireless communication. During the training phase, thetransmitting controller 108 is configured to gather and use past channel condition data (i.e., the historical CSI time resourcestraining set (Ω^)) to train the transformer model 110. Consequently, the transformer model 110 learns how to predict futurechannel conditions based on the patterns observed in the set of historical data. In an example, ℎ^^ ∈ ℂ^×^may be a historicalpart of CSI sequence and may be represented as in Equation (1) comprising ^ past CSI vectors where, Ω^ = {1,2, … , ^} is the set of time resources of the CSI vectors that belongs to the historical CSI part.Further, the transmitting controller 108 is configured to generate a future CSI time resource training set (Ω^) comprising futureCSI time resources (^^) sets. The future CSI time resource training set (Ω^) may be configured to represent estimated futurechannel conditions. The transmitting controller 108 is configured to generate the future CSI time resource training set (Ω^) byIn said example, ℎ^^ is the future part of the CSI sequence that contains ^ future CSI vectors as given in Equation (2) that arepredicted based on the past CSI part, where, Ω^ = {^ + 1, ^ + 2, … , ^ + ^} represent the set of time resources of the CSI vectors that belong to the future part. 5 10 15 20 25 30 In case, if the condition provided in Equation (7) is not satisfied that may lead to performance deterioration. Moreover, there35 is no such a specific design condition on the future CSI time resource training set (Ω^). In accordance with an embodiment, the transmitting controller 108 is further configured to generate the historical CSI timeresources training set (Ω^) based on a slot resolution and wherein the receiving controller 112 is further configured to acquirea CSI vector at each subsequent slot. In an example, the historical CSI time resources training set (Ω^) may include ^historical CSI instances as given in Equation (8) where each CSI instance might happen, at max, once each time slot.Ω^= {1,2, … , ^} (8)In said example, ^ ∈ ^^ may represent the ^^^ time slot resource within the CSI sequence and ^ denotes the number of past(historical) time slots. When the historical CSI time resources training set (Ω^) is designed with the slot resolution then, acorresponding CSI vector is acquired at each subsequent slot. Similarly, the future CSI time resources training set (Ω^) mayinclude ^ future CSI instances based on the slot resolution. At each time slot, the receiving controller 112 may be configuredto acquire the CSI vector, which provides information about the state of the communication channel (or the communicationnetwork 106). The acquired CSI vector represents various characteristics of the communication channel, such as signal strength, interference, multipath fading, and the like. Consequently, a comprehensive analysis of the communication channel conditions can be obtained and used for reliable data communication.In accordance with an embodiment, the transmitting controller 108 is further configured to generate the historical CSI timeresources training set (Ω^) based on a symbol resolution and wherein the receiving controller 112 is further configured toacquire a CSI vector at each subsequent symbol. The Symbol resolution refers to the precision or granularity with which datais sampled or represented in terms of symbols. By generating the historical CSI time resources training set (Ω^) based onsymbol resolution ensures that the training set aligns with the actual communication protocol and symbol timing, providingrelevant and accurate data. The transformer model 110 may benefit the generated historical CSI time resources training set(Ω^) based on the symbol resolution in terms of enhanced ability to predict and adapt to communication channel conditions.Similar to the slot resolution, the historical CSI time resources training set (Ω^) may include ^ historical CSI instances in anexample, as given in Equation (9) where each CSI instance might happen, at max, once each time symbol. Ω^ = {1,2, … , ^} (9)In said example, ^ ∈ ^^ may represent the ^^^ time symbol resource within the CSI sequence and ^ denotes the number ofpast (historical) time symbols. When the historical CSI time resources training set (Ω^) is designed with the symbol resolution then, a corresponding CSI vector is acquired at each subsequent symbol. Similarly, the future CSI time resources training set(Ω^) may include ^ future CSI instances based on the symbol resolution. At each time symbol, the receiving controller 112may be configured to acquire the CSI vector,which provides the information about the current state of the communication channel (i.e., the communication network 106).Moreover, the future CSI time resources training set (Ω^) represented by Equation (10) has no design constraints and differentconfigurations of (Ω^) are possible.Ω^ = {^ + 1, ^ + 2, … , ^ + ^} (10)For example, ^ ∈ Ω^ may represent the ^^^symbol and time slot to predict channel conditions depending on other communication factors. In accordance with an embodiment, the receiving controller 112 is further configured to acquire a CSI sequence as a result ofpilot transmission and channel estimation processes. The acquired CSI sequence may have the following representation asshown in Equation (11) The generation of the CSI sequences during the training phase can follow different strategies which already exist in the literature. For example, the CSI sequences may be acquired or designed as a result of pilot transmission and channel estimation processes. This may be achieved via a simulation process or via a real-world measurement campaign. The receiving controller 112 may be configured to obtain a series of CSI measurements over time. The pilot signal is predefined to the transmitting node 102 and the receiving node 104. The receiving node 104 may be configured to use the received pilot signal to estimate the properties of the channel. By comparing the received pilot signals with the known transmitted pilotsignals, the receiving controller 112 can infer how the communication channel (i.e., the communication network 106) willaffect the transmitting signal. In accordance with an embodiment, the receiving controller 112 is further configured to acquire a CSI sequence utilizing asimulation process at the receiving node 104 using channel modelling techniques. The simulation process is a virtual processwhere a communication scenario is modelled without actually transmitting real signals over a physical channel. In the simulation process, mathematical or computational models are employed to mimic the behavior of the communication channelunder various conditions. The channel modelling techniques replicate real-world factors, such as interference, noise, and signaldegradation, allowing the receiving controller 112 to simulate the process of sending and receiving signals. In an exemplaryscenario of acquiring the CSI sequence, the CSI sequences may be directly generated via the simulation processes at thecommunicating node (i.e., the receiving node 104) of interest using channel modelling techniques at that node. Furthermore,the receiving controller 112 may be configured to add an additional noise component to the simulated channels to compensatefor the estimation errors according to Equation (13) ℎ^^ = ℎ^ + ^^ (13)where ℎ^^ denotes the simulated channel vector at the ^^ℎ time resource within the CSI sequence and ^^ denotes the additivenoise vector. In another exemplary scenario of acquiring the CSI sequence, the CSI sequences may be directly generated viathe simulation processes at the transmitting node 102 using channel modelling techniques at the transmitting node 102. 5 10 1520 25 30controller 108 configures how to encode this timing information. The receiving node 104 then sends back the estimated CSIsequence to the transmitting node 102, which is further used as an input to the transformer model 110 for improved future CSI prediction. In accordance with an embodiment, the transmitting controller 108 is further configured to generate a CSI time resource set(^^) having CSI observations at a first frequency for low-mobility scenarios and at a second frequency for high-mobilityscenarios, where the second frequency is higher than the first frequency. The transmitting controller 108 is configured to generate the CSI time resource set (^^) having CSI observations at different frequencies, described in detail, for example, in FIG.6. there remains a technical challenge in developing a transformer-based CSI predictor capable of effectively utilizing any subsetof historical data, rather than being constrained to the most recent data pointsThe communication system 100 employs the transformer model 110 capable of effectively utilizing any subset of historicaldata, rather than being constrained to the most recent data points in contrast to a conventional transformer-based CSI predictor. Alternatively, may be stated as, the communication system 100 employs a unified transformer framework that is configured to support different CSI or pilot configurations. The transformer model 110 does not require pre-training or fine tuning to support different CSI configurations. The transformer model 110 is configured to predict the future channel conditions with enhancedaccuracy by effectively handling sequential data and complex patterns in the training data, leading to more reliable datacommunication. The comprehensive approach used in the communication system 100 includes generating both historical and future CSI time resource training sets, which ensures robust training of the transformer model 110 on a wide range of data andimproves its ability to adapt to different CSI configurations. Additionally, the transformer model 110 employs dynamicpositional encoding techniques (i.e., different positional encodings between training and inference phase) that enables the transformer model 110 to adapt to changes happening over the communication channel, and to provide reliable datacommunication in varying channel conditions, whereas conventional methods use static encoding that may not effectivelyaddress the continuously changing channel conditions. Consequently, the transformer model 110 outperforms the conventionaltransformer-based CSI predictor trained over a specific CSI and deployed on different CSI configurations. FIG. 2A is a block diagram that illustrates various exemplary components of a transmitting node, in accordance with an embodiment of the present disclosure. FIG.2A is described in conjunction with elements from FIG.1. With reference to FIG.2A, there is shown a block diagram 200A of the transmitting node 102 that includes a communication interface 202 and amemory 204 in addition to the transmitting controller 108 and the transformer model 110.The communication interface 202 may include suitable logic, circuitry, and / or interfaces that is configured to transmit the historical CSI time resource training set (Ω^) and the future CSI time resource training set (Ω^) to the receiving node 104. Examples of the communication interface 202 may include, but are not limited to, a radio frequency transceiver, a network interface, a telematics unit, or any antenna suitable for use in an IoT device, an IoT controller, a user equipment, a repeater, a base station or other portable or non-portable communication devices. The communication interface 202 may wirelessly communicate by use of various wireless communication protocols. The memory 204 may include suitable logic, circuitry, and / or interfaces that is configured to store machine code and / orinstructions executable by the transmitting controller 108. Examples of implementation of the memory 204 may include, butare not limited to, an 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), acomputer readable storage medium, and / or CPU cache memory. The memory 204 may store an operating system and / or acomputer program product to operate the transmitting node 102. A computer readable storage medium for providing a non-transient memory may include, but is not limited to, an electronic storage device, a magnetic storage device, an optical storagedevice, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. FIG. 2B is a block diagram that illustrates various exemplary components of a receiving node, in accordance with an embodiment of the present disclosure. FIG. 2B is described in conjunction with elements from FIGs. 1 and 2A. With referenceto FIG. 2B, there is shown a block diagram 200B of the receiving node 104 that includes a communication interface 206 and amemory 208 in addition to the receiving controller 112. The memory 208 may include suitable logic, circuitry, and / or interfaces that is configured to store machine code and / orinstructions executable by the receiving controller 112. Example of implementation of the memory 208 of the receiving node104 are similar to that of the memory 204 of the transmitting node 102. predicting CSI based on the transformer model 110 in the communication system 100. The method 300 includes steps 302 to322 (the steps 302 to 310 of the method 300 are shown in FIG. 3A and the steps 312 to 322 are shown in FIG. 3B). Thetransmitting node 102 and the receiving node 104 of the communication system 100 are configured to execute the method 300.There is provided the method 300 for predicting CSI, based on the transformer model 110 in the communication system 100that includes the transmitting node 102 and the receiving node 104. The method 300 includes training the transformer model110 over historical CSI sequences with high resolution time resource sets and corresponding positional encodingsequence (^(^^^^^^^^) ^) with same high resolution time resource set.Referring to FIG. 3A, at step 302, the method 300 comprises at the transmitting node 102 generating a historical CSI timeresource training set comprising historic CSI time resources (^^) that the communication system 100 is expected tohandle. The historical CSI time resource training set is composed of historic CSI time resources (^^), which may bereferred to as data points that reflect the past channel conditions the communication system 100 has experienced previously.The historic CSI time resources (^^) capture information about the state of the channels over time, such as signal strength,interference levels, fading, cross-interference levels, and other relevant metrics that affect data transmission over thecommunication network 106. By using the historic CSI time resources (^^) for training of the transformer model 110, thecommunication system 100 may benefit of more precise channel estimations leading to efficient utilization of communication resources and reliable data transmission. At step 304, the method 300 further comprises at the transmitting node 102 generating a future CSI time resource training set(Ω^) comprising estimated future CSI time resources (^^). The future CSI time resource training set (Ω^) is composed ofestimated future CSI time resources (^^), which may represent predictions or estimations of the channel conditions that thecommunication system 100 is expected to handle in the future.At step 306, the method 300 further comprises at the transmitting node 102 transmitting the historical CSI time resource trainingset and the future CSI time resource training set (Ω^) to the receiving node 104. The transmission of both that is thehistorical CSI time resource training set and the future CSI time resource training set (Ω^) to the receiving node 104enables the receiving node 104 to access and process both the past and anticipated future CSI data. At step 308, the method 300 further comprises at the receiving node 104 receiving the historical CSI time resource training set(Ω^) and the future CSI time resource training set (Ω^) from the transmitting node 102. The historical CSI time resourcetraining set includes past channel conditions that the communication system 100 has encountered, while the future CSItime resource training set (Ω^) includes estimated or predicted channel conditions that the communication system 100 isexpected to handle. By receiving aforementioned training sets, the receiving node 104 may get equipped with the required data to analyze both past and anticipated future channel states.At step 310, the method 300 further comprises at the receiving node 104 generating a historic CSI sequence (ℎ^^(^^^^^^^^))comprising historic CSI vectors acquired at the corresponding historical CSI time resource training set . The receivingnode 104 processes the historical CSI time resource training set to generate the historic CSI sequence which comprises the historic CSI vectors. The historic CSI vectors represent the order and timing of the past channel states.Now referring to FIG. 3B, at step 312, the method 300 further comprises at the receiving node 104, generating a future CSIsequence (ℎ^(^^^^^^^^)^ ) comprising estimated CSI vectors acquired at the corresponding future CSI time resource training set(Ω^). The future CSI time resource training set (Ω^) may be used for generating the future CSI sequence (ℎ^^(^^^^^^^^)) comprising the future CSI vectors. The future CSI vectors represent predictions or estimations of channel conditions that the communication system 100 is expected to encounter in the future.At step 314, the method 300 further comprises at the receiving node 104, transmitting the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^^(^^^^^^^^)) to the transmitting node 102. The transmission of the historic CSI sequence(ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^^(^^^^^^^^)) to the transmitting node 102 enables the transmitting node 102 to moreaccurately compute the positional encoding sequences using CSI sequences.At step 316, the method 300 further comprises at the transmitting node 102, receiving the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^^(^^^^^^^^)) from the receiving node 104. At step 318, the method 300 further comprises at the transmitting node 102, generating a historical positional encoding sequence (^(^^^^^^^^) ^ ). The historical positional encoding sequence (^(^^^^^^^^) ^) is designed to capture the positional information related to the historical CSI data, such as order andtiming of the CSI vectors within the historical CSI sequence.At step 320, the method 300 further comprises at the transmitting node 102 generating a future positional encoding sequence(^(^^^^^^^^)). The future positional encoding sequence (^(^^^^^^^^) ^^) encodes positional information related to the future CSI data, such as the sequence's order and timing of the estimated future channel conditions.At step 322, the method 300 further comprises at the transmitting node 102, training the transformer model 110 configured topredict CSI based on the historic CSI sequence , the future CSI sequence (ℎ^^(^^^^^^^^)), the historical positional encoding sequence (p(^^^^^^^^) ^ ), and the future positional encoding sequence (^(^^^^^^^^) ^ ). By using the historic and futureCSI sequences along with the corresponding positional encoding sequences, the transformer model 110 may be configured topredict future channel conditions with enhanced accuracy. The steps 302 to 322 are related to training of the transformer model110 for estimating future channel conditions utilizing the historic CSI sequence (ℎ^^(^^^^^^^^)), the future CSI sequence the historical positional encoding sequence (p(^^^^^^^^) ^ ), and the future positional encoding sequence (^(^^^^^^^^) ^ ).In accordance with an embodiment, the method 300 further comprises at the transmitting node 102 generating a CSI timeresource set (^^), configuring position encoding over the CSI time resource set (^^), and transmitting the CSI time resourceset (^^) to the receiving node 104. The method further comprises at the receiving node 104, receiving the CSI time resourceset (^^) from the transmitting node 102, generating an estimated CSI sequence by acquiring CSI vectors (ℎ)corresponding to the CSI time resource set (^^) and transmitting the estimated CSI sequence to the transmittingnode 102. The method 300 further comprises at the transmitting node 102, receiving the estimated CSI sequence (ℎ^^(^^^^^^^^^))and then performing CSI prediction based on the transformer model 110 applied to the estimated CSI sequence This step is related to the deployment phase of the transformer model 110 for estimating future channel conditions in real time application scenarios. In the deployment phase, when the historical CSI sequences are available with lower resolution time resource set (^^), the positional encoding sequence is sampled according to Equation (18). The method 300 is about trainingthe transformer model 110 over high resolution sets (Ω^) and dynamically sampling the positional encoding sequence duringthe deployment phase in correspondence with the available time resource subset according to Equation (7). Thus, the method 300 results in a unified predictor model (i.e., the transformer model 110) that can be deployed over different time resource sets without the requirement for the re-training or fine-tuning procedures. The steps 302 to 322 are only illustrative, and other alternatives can also be provided where one or more steps are added, or one or more steps are provided in a different sequence without departing from the scope of the claims herein.There is provided a computer program product comprising program instructions for performing the method 300, when executedby one or more processors in the communication system node (e.g., the transmitting node 102 and the receiving node 104). Thecomputer program is executed on a computer system. The computer program is implemented as an algorithm, embedded in a software stored in the non-transitory computer-readable storage medium having program instructions stored thereon, the program instructions being executable by the one or more processors in the computer system to execute the method 300. 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, an Electrically Erasable Programmable Read-Only Memory (EEPROM), a Random Access Memory (RAM), a Read Only Memory (ROM), a Hard Disk Drive (HDD), a Flash memory, a Secure Digital (SD) card, a Solid-State Drive (SSD), a computer-readable storage medium, and / or a CPU cache memory. FIG.4 is a flowchart of a method for use in a receiving node in a communication system, in accordance with an embodimentof the present disclosure. FIG. 4 is described in conjunction with elements from FIGs. 1, 2A, 2B, and 3A-3B. With referenceto FIG. 4, there is shown a method 400 for use in the receiving node 104 of the communication system 100 (of FIG. 1). Themethod 400 includes steps 402 to 416. The receiving controller 112 of the receiving node 104 is configured to execute the method 400. There is provided the method 400 for use in the receiving node 104 in the communication system 100 (of FIG.1) comprising the transmitting node 102 and the receiving node 104. The method 400 is used for predicting CSI based on the transformer model 110 in the communication system 100. The method 400 comprises following steps to be performed at the receiving node 104. At step 402, the method 400 comprises at the receiving node 104, receiving historical CSI time resource training set (Ω^)comprising historic CSI time resources (^^) that the communication system 100 is expected to handle from the transmittingnode 102.At step 404, the method 400 further comprises at the receiving node 104, receiving future CSI time resource training set (Ω^)comprising estimated future CSI time resources (^^) from the transmitting node 102. The future CSI time resource training setincludes estimated future CSI data (^^) which may include the information about how the channel is expected to behave in thefuture. At step 406, the method 400 comprises at the receiving node 104, generating a historic CSI sequence (ℎ^^(^^^^^^^^)) comprisinghistoric CSI vectors acquired at the corresponding historical CSI time resource training set (Ω^). Each historic CSI vectorrepresents the information about how the communication channel (i.e., the communication network 106) has behaved at specifictimes in the past. At step 408, the method 400 comprises at the receiving node 104, generating a future CSI sequence (ℎ^^(^^^^^^^^)) comprisingestimated CSI vectors acquired at the corresponding future CSI time resource training set (Ω^). Each of the estimated CSIvectors may represent the information about future channel conditions that the communication system 100 is expected to handle at specific times. At step 410, the method 400 further comprises at receiving node 104, transmitting the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^(^^^^^^^^)^ ) to the transmitting node 102. By transmitting the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^(^^^^^^^^)^ ) sequences to the transmitting node 102, the receiving node 104 enables the transmittingnode 102 to generate more accurate positional encoding sequences. At step 412, the method 400 further comprises generating a historical positional encoding sequence . The historical positional encoding sequence (^(^^^^^^^^) ^) provides information about timing or position of the historical CSI datawithin the sequence.At step 414, the method 400 further comprises generating a future positional encoding sequence . The futurepositional encoding sequence (^(^^^^^^^^) ^) provides information about the timing or position of the future CSI data withinthe sequence.At step 416, the method 400 further comprises train a transformer model 110 configured to predict CSI based on the historic CSI sequence (ℎ^(^^^^^^^^) ^(^^^^^^^^)), the future CSI sequence (ℎ^^(^^^^^^^^)), the historical positional encoding sequence (^^), and the future positional encoding sequence (^(^^^^^^^^) ^) . The utilization of the historic CSI sequence (ℎ^^(^^^^^^^^)), the future CSI sequence (ℎ^^(^^^^^^^^)), the historical positional encoding sequence (^(^^^^^^^^) ^ ), and the future positional encoding sequence (^(^^^^^^^^) ^) for training of the transformer model 110 enables the transformer model 110 to support different CSIor pilot configurations.In accordance with an embodiment, the method 400 further comprises receiving a CSI time resource set (^^) from thetransmitting node 102, generating an estimated CSI sequence by acquiring CSI vectors (ℎ) corresponding tothe CSI time resource set (^^) and transmitting the estimated CSI sequence transmitting node 102,thereby enabling the transmitting node 102 performing CSI prediction based on the transformer model 110 applied to the estimated CSI sequence the deployment phase, the transformer model 110 is configured to use the estimated as an input and provide the future CSI sequenceCSI sequence as an output. The steps 402 to 416 are only illustrative, and other alternatives can also be provided where one or more steps are added, or one or more steps are provided in a different sequence without departing from the scope of the claims herein.FIGs. 5A and 5B collectively, is a flowchart of a method for use in a transmitting node, in accordance with an embodiment ofthe present disclosure. FIGs.5A and 5B are described in conjunction with elements from FIGs.1, 2A, 2B, 3A-3B and 4. Withreference to FIGs. 5A and 5B, there is shown a method 500 for use in the transmitting node 102 of the communication system100. The method 500 includes steps 502 to 518 (steps 502 to 510 are shown in FIG. 5A and steps 512 to 518 are shown in FIG.5B). The transmitting controller 108 of the transmitting node 102 is configured to execute the method 500.There is provided the method 500 for use in the transmitting node 102 of the communication system 100 comprising thetransmitting node 102 and the receiving node 104. The method 500 is for predicting CSI based on the transformer model 110 in the communication system 100.Referring to FIG. 5A, at step 502, the method 500 comprises at the transmitting node 102, generating a historical CSI timeresource training set comprising historic CSI time resources (^ ) that the communication system 100 is expected to^ handle. At step 504, the method 500 further comprises at the transmitting node 102, generating a future CSI time resourcetraining set (Ω ) comprising estimated future CSI time resources (^ ). At step 506, the method 500 further comprises at the^ ^ transmitting node 102, transmitting the historical CSI time resource training set and the future CSI time resource trainingset (Ω^) to the receiving node 104. At step 508, the method 500 further comprises enabling the receiving node 104, generatinga historic CSI sequence (ℎ^(^^^^^^^^)^ ) comprising historic CSI vectors acquired at the corresponding historical CSI time resourcetraining set . At step 510, the method 500 further comprises enabling the receiving node 104, generating a future CSIsequence (ℎ^^(^^^^^^^^)) comprising estimated CSI vectors acquired at the corresponding future CSI time resource training set(Ω^).Now referring to FIG. 5B, at step 512, the method 500 further comprises at the transmitting node 102, receiving the historicCSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^(^^^^^^^^)^ ) from the receiving node 104. At step 514, the method500 further comprises at the transmitting node 102, generating a historical positional encoding sequence (^(^^^^^^^^) ^). At step516, the method 500 further comprises at the transmitting node 102, generating a future positional encoding sequence (^(^^^^^^^^) ^ ). At step 518, the method 500 further comprises at the transmitting node 102, training the transformer model 110 configured to predict CSI based on the historic CSI sequence , the future CSI sequence (ℎ^^(^^^^^^^^)), the historical positional encoding sequence (^(^^^^^^^^) ^ ), and the future positional encoding sequence (^(^^^^^^^^) ^ ). The steps 502 to 518 are related to the training phase of the transformer model 110.In accordance with an embodiment, the method 500 further comprises at the transmitting node 102, generating a CSI timeresource set (^^), configuring position encoding over the CSI time resource set (^^), and transmitting the CSI time resourceset (^^ ) to the receiving node 104, thereby enabling the receiving node 104 generating an estimated CSI sequence by acquiring CSI vectors (ℎ) corresponding to the CSI time resource set (^^) and transmitting the estimatedCSI sequence (ℎ^(^^^^^^^^^)^ ) transmitting node 102, whereby the method 500 further comprises at the transmitting node102 receiving the estimated CSI sequence (ℎ^(^^^^^^^^^)^ ) and then performing CSI prediction based on the transformer model110 applied to the estimated CSI sequence (ℎ^^(^^^^^^^^^)). This step is related to the deployment phase of the transformer model 110. The steps 502 to 518 are only illustrative, and other alternatives can also be provided where one or more steps are added, or one or more steps are provided in a different sequence without departing from the scope of the claims herein.FIG. 6 is an exemplary implementation scenario of transmitted pilot grids with different structures under different channelconditions, in accordance with an embodiment of the present disclosure. FIG.6 is described in conjunction with elements fromFIGs. 1, 2A, 2B, 3A-3B, 4 and 5A-5B. With reference to FIG. 6, there is shown an exemplary implementation scenario 600 oftransmitted pilot grids with different structures under different channel conditions. With reference to FIG.6, there is shown a first User Equipment (UE) 602 and a second UE 604 and a first historic CSI time resource 602A and a second historic CSI timeresource 604B associated with the first UE 602 and the second UE 604, respectively. There is further shown a first pilot grid606 sent to the first UE 602 and a second pilot grid 608 sent to the second UE 604 by a Base Station (BS) 610.In the exemplary implementation scenario 600, the historical CSI time resources training set (Ω^) is generated based on thesymbol resolution, for example, Ω^ = {1,2, … , ^ = 19}. Therefore, the choice of the historic CSI time resource (^^) setsmay be affected by different conditions, such as for low mobility scenarios and high mobility scenarios. For low mobility scenarios, less frequent CSI estimates are required to capture channel variations. Thus, having a CSI observation each at ^ = 6 OFDM symbols may be a reasonable choice hence, the historic CSI time resource (^^) set maycorrespond to ^^ = {1,7,13,19} where the design condition provided in Equation (7) is satisfied. The first historic CSI timeresource 602A corresponds to the CSI time resource set obtained at ^ = 6 OFDM symbols. Alternatively, may be stated asthe first historic CSI time resource 602A may correspond to the CSI time resource set obtained at a pilot spacing (^ = 6). Thefirst UE 602 may be referred to as a slow time varying user. For high mobility scenarios, more frequent CSI estimates are required to capture the channel variations. Thus, having a CSIobservation each at ^ = 3 OFDM symbols may be a reasonable choice hence, the historic CSI time resource (^^) set maycorrespond to ^^= {1,3,7,10,13,16,19} where the design condition provided in Equation (7) is still valid. The secondhistoric CSI time resource 604A corresponds to the CSI time resource set obtained at ^ = 3 OFDM symbols. Alternatively,may be stated as the second historic CSI time resource 604A may correspond to the CSI time resource set obtained at a pilotspacing (^ = 3). The second UE 604 may be referred to as a fast time varying user. Thus, depending on different channelconditions, the base station 610 is configured to send different pilot grids, such as the first pilot grid 606 to the first UE 602and the second pilot grid 608 to the second UE 604.FIG. 7 is an exemplary implementation scenario of transmitted pilot grids with same structures but to different type of users,in accordance with an embodiment of the present disclosure. FIG.7 is described in conjunction with elements from FIGs.1, 2A, 2B, 3A-3B, 4, 5A-5B and 6. With reference to FIG. 7, there is shown an exemplary implementation scenario 700 of transmitted pilot grids with same structures but to different user equipments. With reference to FIG. 7, there is shown a first UE 702 and a second UE 704 and a second historic CSI time resource 704A associated with the second UE 704, respectively. There is further shown a first pilot grid 706 sent to the first UE 702 and a second pilot grid 708 sent to the second UE 704 by a Base Station (BS) 710.In the exemplary implementation scenario 700, the first UE 702 is always awake and the second UE 704 gets awaken on certainconditions, for example, as in case of Internet-of-Things (IoT) node. Therefore, the first UE 702 and the second UE 704 have different discontinuous reception (DRX) cycles, thus, have access to different CSI measurement instances. Therefore, the firstpilot grid 706 sent to the first UE 702 and the second pilot grid 708 sent to the second UE 704 have same structures but sent todifferent kind of UEs.FIG. 8 is a flowchart that depicts a series of operations required in Dynamic Positional Encoding (DPE) technique, inaccordance with an embodiment of the present disclosure. FIG.8 is described in conjunction with elements from FIGs.1, 2A, 2B, 3A-3B, 4, 5A-5B, 6, and 7. With reference to FIG.8, there is shown a flowchart 800 that includes a series of operations 802 to 818. At operation 802, it is determined that whether the training of the transformer model 110 is performed or not. In case, if the training of the transformer model 110 is performed then, an inference phase of the transformer model 110 is executed atoperation 804, otherwise a training phase of the transformer model 110 is executed at operation 806.In the inference phase, at operation 804, the transmitting node 102 is configured to generate the historic CSI time resource set(^^) and the future CSI time resource set (^^).In the inference phase, at operation 808, the transmitting node 102 is configured to configure position encoding over the historic CSI time resource set (^^) and the future CSI time resource set (^^).In the inference phase, at operation 810, the transmitting node 102 is configured to transmit the historic CSI time resource set(^^) and the future CSI time resource set (^^) to the receiving node 104. In response to, the receiving node 104 is configured to generate an estimated CSI sequence (ℎ^^^^^^^^^^^ ) by acquiring CSI vectors (ℎ) corresponding to the CSI time resourceset (^^) and to transmit the estimated CSI sequence (ℎ^^^^^^^^^^^ ) to the transmitting node 102. In the training phase, at operation 806, the transmitting node 102 is configured to generate the historical CSI time resourcestraining set (Ω^) and the future CSI time resource training set (Ω^) and transmit the generated historical CSI time resourcestraining set (Ω^) and the future CSI time resource training set (Ω^) to the receiving node 104.In the training phase, at operation 812, the transmitting node 102 is configured to generate position encoding sequences. In the training phase, at operation 814, the receiving node 104 is configured to generate the historic CSI sequence (ℎ^^(^^^^^^^^))comprising historic CSI vectors acquired at the corresponding historical CSI time resource training set (Ω^) and the futureCSI sequence (ℎ^^(^^^^^^^^)) comprising future CSI vectors acquired at the corresponding future CSI time resource training set (Ω^). At operation 816, the transmitting node 102 is configured to train the transformer model 110 using the historic CSI sequence ( ) the future CSI sequence (ℎ^^^^^^^^^ ^ ) and the position encoding sequences. If the training phase is alreadyperformed, then, at operation 818, the transmitting node 102 is configured to use the estimated CSI sequence (ℎ^ ^^^^^^^^^^ ) asan input to the transformer model 110 for predicting future CSI sequence (ℎ^^(^^^^^^^^^)).FIG. 9 is a flowchart that depicts a series of operations performed in training and inference phase of a transformer model of acommunication system, in accordance with an embodiment of the present disclosure. FIG. 9 is described in conjunction withelements from FIGs. 1, 2A, 2B, 3A-3B, 4, 5A-5B, 6, 7 and 8. With reference to FIG. 9, there is shown a flowchart 900 thatincludes a series of operations 902 to 922. There is further shown the transmitting node 102 and the receiving node 104 of thecommunication system 100 (of FIG.1). At operation 902, the transmitting node 102 is configured to send training pilots over the historical CSI time resources trainingset (Ω^) and the future CSI time resource training set (Ω^) to the receiving node 104.At operation 904, the receiving node 104 is configured to estimate the CSI sequences, such as the historic CSI sequence(ℎ^(^^^^^^^^)^ ) comprising historic CSI vectors (ℎ^^ , ℎ^^ , … , and the future CSI sequence (ℎ^^(^^^^^^^^)) comprising future CSIvectors (ℎ^^^^, ℎ^^^^, … , ℎ^^^^).At operation 906, the receiving node 104 is configured to send the estimated CSI sequences that is the historic CSI sequence(ℎ^(^^^^^^^^) ^ ) and the future CSI sequence (ℎ^(^^^^^^^^) ^ ) to the transmitting node 102. At operation 908, the transmitting node 102 is configured to generate the positional encoding sequences using the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^(^^^^^^^^)^ ) and train the transformer model 110 using the generatedpositional encoding sequences, the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^^(^^^^^^^^)). The operations 902 to 908 are related to the training phase of the transformer model 110. At operation 910, the receiving node 104 is configured to inform the transmitting node 102 about its positional encoding capabilities.At operation 912, the transmitting node 102 is configured to generate a CSI time resource set (^^) based on channel conditionsand UEs category (described in detail, for example, in FIGs.6 and 7). At operation 914, the transmitting node 102 is configured to transmit the CSI time resource set (^^) to the receiving node 104. At operation 916, the transmitting node 102 is configured to configure position encoding over the CSI time resource set (^^). At operation 918, the receiving node 104 is configured to generate an estimated CSI sequence (ℎ^^^^^^^^^^ ^) by acquiring CSIvectors (ℎ) corresponding to the CSI time resource set (^^).At operation 920, the receiving node 104 is configured to transmit the estimated CSI sequence (ℎ^ ^^^^^^^^^^ ) to the transmittingnode 102.At operation 922, the transmitting node 102 is configured to provide the estimated CSI sequence (ℎ^^^^^^^^^^^ ) as an input tothe transformer model 110 for predicting future CSI sequence (ℎ^^^^^^^^^^^). The operations 910 to 922 are related to the deployment phase of the transformer model 110. 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 the singular 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 or advantageous over other embodiments and / 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 is appreciatedthat certain features of the present disclosure, which are, for clarity, described in the context of separate embodiments, mayalso be provided in combination in a single embodiment. Conversely, various features of the present disclosure, which are, for brevity, described in the context of a single embodiment, may also be provided separately or in any suitable combination or as suitable in any other described embodiment of the disclosure.

Claims

CLAIMS1. A communication system (100) configured to predict Channel State Information, CSI, based on a transformer model (110),the communication system (100) comprises a transmitting node (102) and a receiving node (104), the transmitting node (102) comprising a transmitting controller (108), and the receiving node (104) comprising a receiving controller (112), the transmitting controller (108) is configured to, during a training phase, generate a historical CSI time resources training set (Ω^) comprising historic CSI time resource (^^) sets that thecommunication system (100) is expected to handle, generate a future CSI time resource training set (Ω^) comprising future CSI time resource (^^) sets, andtransmit the historical CSI time resource training set (Ω^) and the future CSI time resource training set (Ω^) to the receiving node, whereby the receiving controller (112) is configured to receive the historical CSI time resource training set (Ω^) and the future CSI time resource training set (Ω^) from the transmitting node (102), and in response there to generate a historic CSI sequence (ℎ^(^^^^^^^^)^ ) comprising historic CSI vectors acquired at the correspondinghistorical CSI time resource training set (Ω^), generate a future CSI sequence (ℎ^(^^^^^^^^)^ ) comprising future CSI vectors acquired at the corresponding future CSItime resource training set (Ω^), and transmit the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^^(^^^^^^^^)) to the transmitting node (102),the transmitting controller (108) is further configured toreceive the historic CSI sequence (ℎ^(^^^^^^^^ ^)) and the future CSI sequence (ℎ^(^^^^^^^^ ^)) from the receiving node (104), generate a historical positional encoding sequence (^(^^^^^^^^) ^ ), generate a future positional encoding sequence (^(^^^^^^^^) ^ ) and then train the transformer model (110) configured to predict future CSI based on the historic CSI sequencethe future CSI sequence (ℎ^^(^^^^^^^^)), the historical positional encoding sequence (^(^^^^^^^^) ^ ), and the future positional encoding sequence (^(^^^^^^^^) ^ ).

2. The communication system (100) according to claim 1, wherein the transmitting controller (108) is further configured togenerate the historical CSI time resources training set (Ω^) of the training phase to include any potential historic CSI timeresources set of an inference phase3. The communication system (100) according to claim 1 or 2, wherein the transmitting controller (108) is further configuredto generate the historical CSI time resources training set (Ω^) based on a slot resolution and wherein the receiving controller(112) is further configured to acquire a CSI vector at each subsequent slot.

4. The communication system (100) according to claim 1 or 2, wherein the transmitting controller (108) is further configuredto generate the historical CSI time resources training set (Ω^) based on a symbol resolution and wherein the receivingcontroller (112) is further configured to acquire a CSI vector at each subsequent symbol.

5. The communication system (100) according to any preceding claim, wherein the receiving controller (112) is further configured to acquire a CSI sequence as a result of pilot transmission and channel estimation processes.

6. The communication system (100) according to claim 5, wherein the receiving controller (112) is further configured to acquire a CSI sequence as a result of pilot transmission and channel estimation processes by the transmitting controller (108) being further configured to transmit pilot signals (^^) during an ^^ℎ time resource,whereby the receiving controller (112) is further configured toreceive the pilot signals of the ^^ℎ time resource (^^) andestimate a corresponding CSI vector ℎ^^ = ^(^^ , ^^)where ^() is a channel estimation function.

7. The communication system (100) according to any of claims 1 to 4, wherein the receiving controller (112) is furtherconfigured to acquire a CSI sequence utilizing a simulation process at the receiving node (104) using channel modellingtechniques.

8. The communication system (100) according to any preceding claim, wherein the transmitting controller (108) is further configured to generate the historical positional encoding sequence as= [^^ , ^^ , … , ^^] where ^^ = ^^(^, : ) and^ ∈ Ω^ and a future positional encoding sequence as ^(^^^^^^^^) ^= [^^^^ , ^^^^ , … , ^^^^] where ^^ = ^^(^, : ) and ^ ∈ Ω^,and wherein9. The communication system (100) according to any preceding claim, wherein the transmitting controller (108) is further configured to, during an inference phase, generate a CSI time resource set (^^), configure position encoding over the CSI time resource setandtransmit the CSI time resource set (^^) to the receiving node (104), whereby the receiving controller (112) is further configured to receive the CSI time resource set (^^) from the transmitting node (102), generate an estimated CSI sequence (ℎ^(^^^^^^^^^) ^) by acquiring CSI vectors (ℎ) corresponding to the CSI timeresource set (^^) and to transmit the estimated CSI sequence (ℎ^(^^^^^^^^^) ^ ) to the transmitting node (102), whereby the transmitting controller (108) is further configured to receive the estimated CSI sequenceand thenperform CSI prediction based on the transformer model (110) applied to the estimated CSI sequence10. The communication system (100) according to claim 9, wherein the transmitting controller (108) is further configured to generate the historical time resources set (Ω^) of the inference phase to include any potential historic CSI time resources set of the inference phase^^⊆Ω^.

11. The communication system (100) according to claim 9 or 10, wherein the positional encoding function during thedeployment phase is adaptive to the available time resources set ^^.

12. The communication system (100) according to claim 9, wherein the transmitting controller (108) is further configured to configure position encoding over the CSI time resource set (^^) after transmitting the CSI time resource set (^^) to the receiving node (104) and before receiving the estimated CSI sequence13. The communication system (100) according to claim 9 or 12, wherein the transmitting controller (108) is further configured to generate a CSI time resource set (^^) having CSI observations at a first frequency for low-mobility scenarios and at a second frequency for high-mobility scenarios, wherein the second frequency is higher than the first frequency.

14. A method (300) for predicting Channel State Information, CSI, based on a transformer model (110) in a communicationsystem (100), wherein the communication system (100) comprises a transmitting node (102) and a receiving node (104),wherein the method (300) comprises: at the transmitting node (102) generating a historical CSI time resource training set (Ω^) comprising historic CSI time resources (^^) that the communication system (100) is expected to handle, generating a future CSI time resource training set (Ω^) comprising estimated future CSI time resources (^^), and transmitting the historical CSI time resource training set (Ω^) and the future CSI time resource training set (Ω^) tothe receiving node (104), whereby the method (300) further comprises: at the receiving node (104) receiving the historical CSI time resource training set (Ω^) and the future CSI time resource training set (Ω^) from the transmitting node (102), and in response there to generating a historic CSI sequence (ℎ^(^^^^^^^^)^ ) comprising historic CSI vectors acquired at the correspondinghistorical CSI time resource training set (Ω^), generating a future CSI sequence (ℎ^(^^^^^^^^) ^ ) comprising estimated CSI vectors acquired at the corresponding future CSI time resource training set (Ω^), and transmitting the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^^(^^^^^^^^)) to the transmitting node (102), whereby the method (300) further comprises:at the transmitting node (102)receiving the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^^(^^^^^^^^)) from the receiving node (104), generating a historical positional encoding sequence (^(^^^^^^^^) ^ ), generating a future positional encoding sequence (^(^^^^^^^^) ^ ) and thentraining the transformer model (110) configured to predict CSI based on the historic CSI sequence (ℎ^^(^^^^^^^^)), the future CSI sequence (ℎ^^(^^^^^^^^)), the historical positional encoding sequence (^(^^^^^^^^) ^ ), and the future positional encoding sequence (^(^^^^^^^^) ^ ).

15. The method according to claim 14, wherein the method (300) further comprises:at the transmitting node (102)generating a CSI time resource set (^^), configuring position encoding over the CSI time resource set (^^), and transmitting the CSI time resource set (^^) to the receiving node (104), whereby the method (300) further comprises: at the receiving node (104) receiving the CSI time resource set (^^) from the transmitting node (102), generating an estimated CSI sequence (ℎ^(^^^^^^^^^) ^) by acquiring CSI vectors (ℎ) corresponding to the CSI timeresource set (^^) and transmitting the estimated CSI sequence (ℎ^(^^^^^^^^^) ^ ) to the transmitting node (102), whereby the method (300) further comprises:at the transmitting node (102)receiving the estimated CSI sequenceand then performing CSI prediction based on the transformer model (110) applied to the estimated CSI sequence16. A computer program product comprising program instructions for performing the method (300) according to claim 14 or15, when executed by one or more processors in a communication system node.

17. A method (400) for use in a receiving node (104) in a communication system (100), the method (400) being for predictingChannel State Information, CSI, based on a transformer model (110), wherein the method (400) comprises:receiving historical CSI time resource training set (Ω^) comprising historic CSI time resources (^^) from a transmitting node (102), receiving future CSI time resource training set (Ω^) comprising estimated future CSI time resources (^^) from thetransmitting node (102), and in response there to generating a historic CSI sequence (ℎ^(^^^^^^^^) ^ ) comprising historic CSI vectors acquired at the corresponding historical CSI time resource training set (Ω^), generating a future CSI sequence (ℎ^(^^^^^^^^)^ ) comprising estimated CSI vectors acquired at the corresponding futureCSI time resource training set (Ω^), and transmitting the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^^(^^^^^^^^)) to the transmitting node (102).

18. The method (400) according to claim 17, wherein the method (400) further comprises: receiving the CSI time resource set (^^) from the transmitting node (102), generating an estimated CSI sequence (ℎ^(^^^^^^^^^) ^) by acquiring CSI vectors (ℎ) corresponding to the CSI timeresource set (^^), and transmitting the estimated CSI sequencetransmitting node (102).

19. A receiving node (104) configured to predict Channel State Information, CSI, based on a transformer model (110) in acommunication system (100), wherein the receiving node (104) is configured toreceive historical CSI time resource training set (Ω^) comprising historic CSI time resources (^^) from a transmittingnode (102), receive future CSI time resource training set (Ω^) comprising estimated future CSI time resources (^^) from the transmitting node (102), and in response there to generate a historic CSI sequence (ℎ^(^^^^^^^^)^ ) comprising historic CSI vectors acquired at the correspondinghistorical CSI time resource training set (Ω^), generate a future CSI sequence (ℎ^(^^^^^^^^)^ ) comprising estimated CSI vectors acquired at the corresponding futureCSI time resource training set (Ω^), andtransmit the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^^(^^^^^^^^)) to the transmitting node (102).

20. The receiving node (104) according to claim 19, wherein the receiving node (104) is further configured to receive the CSI time resource set (^^) from the transmitting node (102), generate an estimated CSI sequence (ℎ^(^^^^^^^^^) ^) by acquiring CSI vectors (ℎ) corresponding to the CSI timeresource set (^^), andtransmit the estimated CSI sequence (ℎ^(^^^^^^^^^) ^) totransmitting node (102), thereby enabling the transmittingnode (102), performing CSI prediction based on the transformer model (110) applied to the estimated CSI sequence21. A method (500) for use in a transmitting node (102) in a communication system (100), the method (500) being for predictingChannel State Information, CSI, based on a transformer model (110), wherein the method (500) comprises: generating a historical CSI time resource training set (Ω^) comprising historic CSI time resources (^^), generating a future CSI time resource training set (Ω^) comprising estimated future CSI time resources (^^), andtransmitting the historical CSI time resource training set (Ω^) and the future CSI time resource training set (Ω^) to a receivingnode (104), receiving the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^^(^^^^^^^^)) from the receiving node (104), generating a historical positional encoding sequence (^(^^^^^^^^) ^ ), generating a future positional encoding sequence (^(^^^^^^^^) ^ ), and then training the transformer model configured to predict CSI based on the historic CSI sequence (ℎ^^(^^^^^^^^)), the future CSI sequence (ℎ^^(^^^^^^^^)), the historical positional encoding sequence (^(^^^^^^^^) ^ ), and the future positional encoding sequence (^(^^^^^^^^) ^ ).

22. The method (500) according to claim 21, wherein the method (500) further comprises:generating a CSI time resource set (^^), configuring position encoding over the CSI time resource set (^^), andtransmitting the CSI time resource set (^^) to the receiving node (104), thereby enabling the receiving node (104),generating an estimated CSI sequence (ℎ^(^^^^^^^^^) ^) by acquiring CSI vectors (ℎ) corresponding to the CSI timeresource set (^^), transmitting the estimated CSI sequencetransmitting node (102), receiving the estimated CSI sequenceperforming CSI prediction based on the transformer model (110) applied to the estimated CSI sequence 23. A transmitting node (102) configured to predict Channel State Information, CSI, based on a transformer model (110) in acommunication system (100), wherein the transmitting node (102) is configured togenerate a historical CSI time resource training set (Ω^) comprising historic CSI time resources (^^) that the communication system (100) is expected to handle, generate a future CSI time resource training set (Ω^) comprising estimated future CSI time resources (^^), andtransmit the historical CSI time resource training set (Ω^) and the future CSI time resource training set (^F) to a receiving node (104), receive the historic CSI sequence (ℎ^^(^^^^^^^^)) and the future CSI sequence (ℎ^^(^^^^^^^^)) from the receiving node (104), generate a historical positional encoding sequence (^(^^^^^^^^) ^ ), generate a future positional encoding sequenceand then train the transformer model (110) configured to predict CSI based on the historic CSI sequence (ℎ^^(^^^^^^^^)), the future CSI sequence (ℎ^^(^^^^^^^^)), the historical positional encoding sequence (^(^^^^^^^^) ^ ), and the future positional encoding sequence (^(^^^^^^^^) ^ ).

24. The transmitting node (102) according to claim 23, wherein the transmitting node (102) is further configured to generate a CSI time resource set (^^), configure position encoding over the CSI time resource set (^^), andtransmit the CSI time resource set (^^) to the receiving node (104),receive the estimated CSI sequenceperform CSI prediction based on the transformer model (110) applied to the estimated CSI sequence (ℎ^(^^^^^^^^^) ^ )

Images