Method and apparatus for semantic text transmission based on semantic relay assistance

Abstract

The application discloses a kind of based on semantic relay auxiliary semantic text transmission method and device, the device includes m semantic sending end, n semantic output end and an intelligent semantic relay;Semantic sending end is used to carry out semantic coding to text information, then channel coding is generated first sending signal;Intelligent semantic relay receives first sending signal, and first sending signal is carried out power compensation, generates second sending signal;Semantic receiving end receives second sending signal, and second sending signal is carried out channel decoding, then semantic decoding is carried out, completes the semantic communication of three-party cooperation.First, the device uses intelligent semantic relay, improves the applicability of communication network;Second, intelligent semantic relay compensates signal loss, reduces the influence of signal attenuation;The device can realize the switching of amplify-and-forward mode and decode-and-forward mode, to avoid the knowledge base of semantic sending end and semantic receiving end is different, leading to the problem of inconvenient signal transmission.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology, and in particular relates to a semantic text transmission method and apparatus based on semantic relay assistance. Background Technology

[0002] The world today has entered a new era of economic development led by the information industry. Communication has become fully integrated into social life, serving as a core engine of rapid global development. Based on the theories of Shannon and Weaver, communication issues can be divided into three levels: (1) Technical aspect: How to correctly transmit communication symbols.

[0003] (2) Semantic level: How communication symbols accurately convey meaning.

[0004] (3) Utility level: How the meaning received effectively influences behavior in the form of expectations.

[0005] At the technical level, numerous transmission schemes (e.g., OFDM, MIMO) have been developed for reliable communication to ensure bit-level transmission accuracy. Most current communication technologies are developed with the goal of precise data transmission, but they neglect the meaning of the information carried within the data. As communication networks evolve towards intelligence, the pursuit of efficiency and sustainability deepens, making it no longer reasonable to ignore the semantic level.

[0006] The emergence of semantic communication is inseparable from the following aspects: (1) The data volume is growing explosively, the bandwidth demand is increasing sharply, and the spectrum resources are scarce. Current cellular networks are struggling to handle the data traffic that is growing at an exponential rate, and existing wireless communication technologies cannot support many intelligent application scenarios in 6G mobile communication.

[0007] (2) In the context of Artificial Intelligence of Things (AIOT), deep neural network models are complex and have large output parameters, making it difficult to meet the high real-time requirements of tasks. Therefore, how to use semantic relationships to further reduce the parameters and complexity of communication devices will alleviate communication pressure while ensuring the performance of intelligent tasks and promote the synergy between communication and intelligent computing processes in AIoT.

[0008] (3) The development of modern communication technology, especially mobile communication technology, has gradually approached the limits of communication theory. For example, source coding technology has approached source entropy, and advanced channel coding technologies such as LDPC codes and polar codes have approached channel capacity. Communication devices based on probabilistic information urgently need technological breakthroughs and transformations.

[0009] (4) The emergence of applications such as the Internet of Things, the Internet of Vehicles, and the Industrial Internet has enabled mobile communication networks to rapidly shift from a rate-driven mode to a scenario-driven mode. Perception of everything, interconnection of everything, and intelligence of everything have become the main characteristics of future communication networks.

[0010] In real semantic communication environments, wireless channels are prone to signal fading due to various factors (such as path loss and obstacle obstruction); most importantly, communication difficulties arise when the knowledge base backgrounds of the sender and receiver do not match. Summary of the Invention

[0011] The purpose of this invention is to provide a semantic text transmission method and apparatus based on semantic relay assistance to solve the problems existing in the above-mentioned semantic communication.

[0012] This invention adopts the following technical solution: a semantic text transmission device based on semantic relay assistance, such as... Figure 1 As shown, the device includes m semantic transmitters, n semantic receivers, and an intelligent semantic relay, where m and n are both positive integers greater than or equal to 1. The semantic transmitter is used to perform semantic encoding on the text information, and then perform channel encoding to generate the first transmission signal; The intelligent semantic relay receives the first transmitted signal, performs power compensation on the first transmitted signal, and generates the second transmitted signal; The semantic receiver receives the second transmitted signal, performs channel decoding on the second transmitted signal, and then performs semantic decoding to complete the semantic communication of the three parties.

[0013] Optionally, when m=1, n=1 and the semantic sender and receiver share the same knowledge base, the intelligent semantic relay adopts an amplified forwarding mode to complete the transmission of text information.

[0014] Optionally, when the intelligent semantic relay uses the amplification and forwarding mode to complete the text information transmission, the intelligent semantic relay receives the first transmission signal, performs power compensation on the first transmission signal, and generates the second transmission signal.

[0015] Optionally, when and / or Furthermore, when the semantic sender and the semantic receiver use different knowledge bases, the intelligent semantic relay adopts a decoding and forwarding mode to complete the transmission of text information.

[0016] Optionally, when the intelligent semantic relay uses the decoding and forwarding mode to complete the transmission of text information, each semantic sender performs first semantic encoding on the corresponding text information and then performs first channel encoding to generate the corresponding third transmission signal. Optionally, the intelligent semantic relay is based on a first knowledge base shared with the semantic transmitter. First, each third transmitted signal is amplified in power. Then, the corresponding neural network parameters are called and the amplified signal is decoded in the first channel through the neural network. Finally, the signal decoded in the first channel is semantically decoded to obtain the first decoded signal.

[0017] Optionally, after obtaining the first decoded signal, the intelligent semantic relay, based on the second knowledge base shared with the semantic receiver, first performs second semantic encoding and then performs second channel encoding to generate the corresponding fourth transmission signal; Secondly, the intelligent semantic relay transmits the fourth signal to the corresponding semantic receiver via a wireless channel.

[0018] Optionally, after each semantic receiver receives the corresponding fourth transmission signal, each semantic receiver performs second channel decoding on the fourth transmission signal and then performs second semantic decoding to complete the semantic communication of the three parties.

[0019] The present invention also provides a semantic text transmission method based on semantic relay assistance, applied to the aforementioned semantic text transmission device based on semantic relay assistance, the method comprising: The intelligent semantic relay receives the first transmission signal, performs power compensation on the first transmission signal, generates a second transmission signal, and sends it to the semantic transmitter. The first transmission signal is generated by the semantic transmitter after semantically encoding the text information and then performing channel coding.

[0020] Optionally, when there are m semantic senders and n semantic receivers, where m=1 and n=1, and the semantic senders and semantic receivers share the same knowledge base, the intelligent semantic relay adopts an amplified forwarding mode to complete the transmission of text information.

[0021] Optionally, when the intelligent semantic relay uses the amplification and forwarding mode to complete the text information transmission, the intelligent semantic relay receives the first transmission signal, performs power compensation on the first transmission signal, and generates the second transmission signal.

[0022] Optionally, when there are m semantic senders and n semantic receivers, and / or Furthermore, when the semantic sender and the semantic receiver use different knowledge bases, the intelligent semantic relay adopts a decoding and forwarding mode to complete the transmission of text information.

[0023] Optionally, the intelligent semantic relay is based on a first knowledge base shared with the semantic transmitter. First, each third transmission signal is amplified in power. Then, the corresponding neural network parameters are called and the amplified signal is decoded by the neural network through the first channel. Then, the signal after the first channel decoding is decoded by the first semantic decoding to obtain the first decoded signal. The third transmission signal information is generated by each semantic transmitter performing first semantic encoding on the corresponding text information and then performing first channel encoding.

[0024] Optionally, after obtaining the first decoded signal, the intelligent semantic relay, based on the second knowledge base shared with the semantic receiver, first performs second semantic encoding and then performs second channel encoding to generate the corresponding fourth transmission signal; Secondly, the intelligent semantic relay transmits the fourth signal to the corresponding semantic receiver via a wireless channel.

[0025] The beneficial effects of this invention are as follows: This invention provides a semantic text transmission method and apparatus based on semantic relay assistance. The intelligent semantic relay in this apparatus includes two operating modes: amplification-forwarding mode (AF) and decoding-forwarding mode (DF). First, the text information is semantically encoded by the semantic transmitter. Then, based on the intelligent semantic relay, only the information is amplified and forwarded in the amplification-forwarding mode to compensate for signal loss and reduce the impact of signal attenuation. Finally, by using intelligent semantic relay, the switching between amplification-forwarding mode and decoding-forwarding mode can be realized, thus avoiding the problem of signal transmission inconvenience caused by differences in the knowledge bases of the semantic transmitter and the semantic receiver. Attached Figure Description

[0026] Figure 1 This invention provides a basic architecture diagram based on intelligent semantic relay; Figure 2 A network diagram based on semantic communication is provided for this invention; Figure 3 This invention provides a diagram of a one-to-one intelligent semantic relay DF mode architecture. Figure 4 This invention provides a diagram of a many-to-many intelligent semantic relay DF mode architecture. Figure 5 A network diagram based on semantic communication is provided for this invention; Figure 6 This is a simulation diagram of the AF mode of the present invention; Figure 7 This is a simulation diagram of the DF model of the present invention under a background of low knowledge base difference; Figure 8 This is a simulation diagram of the DF model of the present invention under the background of high knowledge base difference. Detailed Implementation

[0027] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0028] A semantic text transmission device based on semantic relay assistance, such as Figure 1 As shown, the device includes m semantic transmitters, n semantic receivers and an intelligent semantic relay, where m and n are both positive integers greater than or equal to 1; The semantic transmitter is used to perform semantic encoding on the text information, and then perform channel encoding to generate the first transmission signal.

[0029] The intelligent semantic relay receives the first transmitted signal, performs power compensation on the first transmitted signal, and generates a second transmitted signal.

[0030] The semantic receiver receives the second transmitted signal, performs channel decoding on the second transmitted signal, and then performs semantic decoding to complete the semantic communication of the three parties.

[0031] In one embodiment, it should be noted that the transposition proposed in this invention is actually a system, referred to as a device for ease of definition. The semantic text transmission device based on semantic relay assistance basically includes three parts: a semantic transmitter, a semantic receiver, and an intelligent semantic relay. The semantic transmitter includes an encoder, and the semantic receiver includes a semantic decoder, such as... Figure 2 As shown, at the semantic sending end, the text information S is semantically encoded through N layers of Transformer coding blocks, and then channel encoded through a fully connected layer to generate a first transmission signal, which is transmitted through a wireless channel. The intelligent semantic relay receives the first transmission signal, performs power compensation on it, generates a second transmission signal, and forwards it to the semantic receiving end to counteract signal attenuation caused by path loss, obstacle interference, etc., and improve the signal-to-noise ratio. The semantic receiving end receives the second transmission signal, first decodes the second transmission information through the corresponding fully connected layer and N layers of Transformer decoding blocks, and finally restores the original text information semantics through a softmax operation, thus completing the three-way collaborative semantic communication.

[0032] It should be noted that, firstly, this device employs intelligent semantic relay, enabling one-to-one or many-to-many communication between the transmitter and receiver, greatly improving the applicability of the communication network; secondly, intelligent semantic relay performs power compensation on the first transmitted signal to combat signal attenuation caused by path loss, obstacle interference, etc., thereby improving the signal-to-noise ratio; finally, the use of intelligent semantic relay allows for switching between amplification-forwarding mode and decoding-forwarding mode. When the semantic transmitter and receiver share the same knowledge base, the intelligent semantic relay uses amplification-forwarding mode (AF); when the semantic transmitter and receiver use different knowledge bases, the intelligent semantic relay uses decoding-forwarding mode (DF) to avoid signal transmission inconvenience caused by different knowledge bases between the semantic transmitter and receiver.

[0033] In one embodiment, when m=1, n=1, and the semantic sender and receiver share the same knowledge base, the intelligent semantic relay uses an amplified forwarding mode to complete the transmission of text information.

[0034] When the intelligent semantic relay uses the amplification and forwarding mode to complete the transmission of text information, the intelligent semantic relay receives the first transmission signal, performs power compensation on the first transmission signal, and generates the second transmission signal.

[0035] It should be noted that the semantic transmitter uses the Transformer architecture and source channel to jointly encode and generate the first transmission signal, and transmits it to the intelligent semantic relay through the wireless channel transmitter. The wireless channel can be an AWGN channel or other adapted channels, which is determined according to the actual situation and is not restricted.

[0036] In one embodiment, when and / or Furthermore, when the semantic sender and receiver use different knowledge bases, the intelligent semantic relay adopts a decoding and forwarding mode to complete the transmission of text information. First, it takes the dataset of the European Conference Proceedings and divides the dataset so that the knowledge bases of the sender and receiver have certain differences.

[0037] Based on the decoding and forwarding mode, there are two cases. The first case is when m=1 and n=1, such as... Figure 3 As shown; When the intelligent semantic relay uses the decoding and forwarding mode to complete the transmission of text information, the semantic sending end performs the first semantic encoding on the corresponding text information, and then performs the first channel encoding to generate the corresponding third transmission signal. The intelligent semantic relay is based on a first knowledge base shared with the semantic transmitter. First, the third transmitted signal is amplified. Then, the corresponding neural network parameters are called and the amplified signal is decoded through the neural network. Finally, the signal decoded through the first channel is semantically decoded to obtain the first decoded signal.

[0038] After obtaining the first decoded signal, the intelligent semantic relay, based on the second knowledge base shared with the semantic receiver, first performs second semantic encoding, that is, re-encoding, and then performs second channel encoding to generate the corresponding fourth transmission signal. Secondly, the intelligent semantic relay transmits the fourth signal to the corresponding semantic receiver via a wireless channel.

[0039] In the second scenario, when For example, semantic sender 1, semantic sender 2... and semantic receiver 1. Or a third case, when... For example, semantic sender 1, semantic sender 2... and semantic receiver 1, semantic receiver 2, such as... Figure 4 As shown: When the intelligent semantic relay uses the decoding and forwarding mode to complete the transmission of text information, each semantic sender calls the corresponding neural network parameters to perform the first semantic encoding on the corresponding text information, and then performs the first channel encoding to generate the corresponding third transmission signal. Here, the third transmission signal is a plurality of encoded signals. The third is only used to distinguish the transmission signals generated under different circumstances and does not represent the meaning of the number or sequence number. For example, semantic transmitter 1 calls network parameters to perform corresponding first semantic encoding on the corresponding text information, and then performs first channel encoding to generate the corresponding third transmission signal; semantic transmitter 2 calls network parameters to perform corresponding first semantic encoding on the corresponding text information, and then performs first channel encoding to generate the corresponding third transmission signal.

[0040] The intelligent semantic relay is based on a first knowledge base shared with the semantic transmitter. First, each third transmitted signal is amplified in power. Then, the corresponding neural network parameters are called and the amplified signal is decoded in the first channel through the neural network. Finally, the signal decoded in the first channel is semantically decoded to obtain the first decoded signal.

[0041] Specifically, the intelligent semantic relay is based on a first knowledge base shared with the semantic transmitter. First, each third transmission signal is amplified in power. Here, the third transmission signal can be the third transmission signal sent by semantic transmitter 1 or the third transmission signal sent by semantic transmitter 2. Second, the corresponding neural network parameters are called and the neural network is used to perform first channel decoding on each amplified third transmission signal. Then, each signal after first channel decoding is used to perform first semantic decoding to obtain the first decoded signal.

[0042] After obtaining the first decoded signal, the intelligent semantic relay, based on the second knowledge base shared with the semantic receiver, first performs second semantic encoding, and then performs second channel encoding to generate the corresponding fourth transmission signal. Secondly, the intelligent semantic relay transmits the fourth signal to the corresponding semantic receiver via a wireless channel.

[0043] After each semantic receiver receives the corresponding fourth transmission signal, each semantic receiver performs second channel decoding on the fourth transmission signal and then performs second semantic decoding to complete the semantic communication of the three parties.

[0044] This embodiment enables semantic communication between multiple semantic senders and multiple semantic receivers even when the knowledge bases do not match. This is a feasible solution that does not exist in existing semantic communication technologies.

[0045] The present invention also provides a semantic text transmission method based on semantic relay assistance, applicable to the semantic text transmission device based on semantic relay assistance involved in any of the above embodiments, the method comprising: The intelligent semantic relay receives the first transmission signal, performs power compensation on the first transmission signal, generates a second transmission signal, and sends it to the semantic transmitter. The first transmission signal is generated by the semantic transmitter after semantically encoding the text information and then performing channel coding.

[0046] In one embodiment, the method is based on a communication model such as Figure 5 As shown, the basic framework of semantic communication consists of three parts: a transmitter, a wireless channel, and a receiver. The encoder includes semantic encoding and channel encoding components, and the decoder includes channel decoding and semantic decoding components. The inventive point of this invention is to use an intelligent semantic relay to intelligently process the semantic text information generated by the semantic transmitter, such as by amplifying the power or re-encoding it, and then forward it to the semantic receiver, thus completing the three-way collaborative semantic communication.

[0047] Specifically, the semantic receiver typically performs semantic encoding on the text information based on a knowledge base, followed by channel encoding to generate a first transmission signal, which is then sent to the intelligent semantic relay via a wireless channel. If the semantic transmitter and receiver use the same knowledge base, the intelligent semantic relay amplifies the first transmission signal and forwards it directly to the semantic receiver. The semantic receiver then decodes the received signal to complete the three-way collaborative semantic communication. If the semantic transmitter and receiver use different knowledge bases, the intelligent semantic relay uses a decode-forward mode. It first amplifies the first transmission signal, then decodes the amplified signal based on the same knowledge base shared with the semantic transmitter, re-encodes the decoded signal based on the same knowledge base shared with the receiver, and then forwards it to the semantic receiver to complete the three-way collaborative semantic communication.

[0048] In one embodiment, when there are m semantic senders and n semantic receivers, where m=1 and n=1, and the semantic senders and semantic receivers share the same knowledge base, the intelligent semantic relay uses an amplified forwarding mode to complete the transmission of text information.

[0049] Suppose the input to the semantic communication network is a sentence, denoted by s = [w1, w2, ..., wl], where wl represents the l-th word in the sentence. The semantic sender is used to extract semantic information from the text information s and ensure the successful transmission of the semantic information over the physical channel.

[0050] The encoded symbol stream, which is the first transmitted information, can be represented as: , in, It is a semantic encoder with parameter set β. It is a channel encoder with parameter set α.

[0051] To simplify the analysis, assume the coherence time is M, and the second transmitted signal generated after being received and amplified by the intelligent semantic relay is: , in, denoted by , where h is the channel gain and n is the channel noise.

[0052] The semantic receiver includes a channel decoder and a semantic decoder, which respectively recover the transmitted second transmitted signal. The decoded signal can be represented as: , in It is the signal recovered after the semantic receiver decodes it. It is a channel decoder with parameter set δ. It is a semantic decoder with parameter set χ.

[0053] In one embodiment, when there are m semantic senders and n semantic receivers, and / or Furthermore, when the semantic sender and receiver use different knowledge bases, the intelligent semantic relay uses a decoding-forwarding mode to complete the text information transmission. In this case, the intelligent semantic relay contains... Network parameters.

[0054] First, from the semantic sender to the intelligent semantic relay, based on the first knowledge base shared with the semantic sender, the semantic sender sends the first and third transmission signals: , in, For the first semantic encoder, For the first channel encoder, This is the third transmission signal sent by the semantic sender.

[0055] The intelligent semantic relay first amplifies the power of the third transmitted signal, then performs channel decoding and semantic decoding to obtain the first decoded signal.

[0056] , in, This is the signal after the third transmitted signal has been amplified by the intelligent semantic relay.

[0057] , in, For the first channel decoder, As the first semantic decoder, This is the first decoded signal.

[0058] It should be noted that when there are multiple semantic transmitters, the intelligent semantic relay will call the neural network parameters that match the corresponding semantic transmitters to perform channel decoding and semantic decoding on the power-amplified third signal to obtain multiple first decoded signals.

[0059] Secondly, the intelligent semantic relay is transmitted to each semantic receiver, and the first decoded information is re-encoded based on the second knowledge base shared with the semantic receiver: , in, For the second semantic encoder, For the second semantic encoder, This is the fourth signal sent.

[0060] The intelligent semantic relay transmits the fourth transmission signal to the semantic receiver via a wireless channel: , The semantic receiver first performs second channel decoding on the received signal, and then performs second semantic decoding to recover the text information.

[0061] , in, For the second channel decoder, For the second semantic decoder, The text information recovered by the receiving end.

[0062] It should be noted that in this embodiment, by using the dataset from the European Conference proceedings and dividing the dataset, the knowledge bases of the sending and receiving ends are somewhat different. The semantic sending end and the semantic receiving end are not limited by the background of the knowledge base and can differ significantly or even be completely different. Furthermore, the number of Transformer layers in the semantic sending end and the semantic receiving end is compressed, so that the final network performance is less correlated with the network depth.

[0063] This semantic text transmission method based on semantic relay introduces the concept of traditional communication relay into semantic communication. It designs an intelligent semantic relay-assisted smart semantic text transmission technology, which includes two operating modes: amplify-forward mode (AF) and decode-forward mode (DF). The amplify-forward mode only amplifies and forwards the information to compensate for signal loss and reduce the impact of signal attenuation. The decode-forward mode decodes and re-encodes the information and then amplifies it. This solves the problem of how to achieve semantic communication when the semantic sender and receiver have different knowledge base backgrounds, and provides a new method for research on one-to-many and many-to-many semantic communication.

[0064] Regarding the semantic text transmission method based on semantic relay assistance in this invention, this invention also provides experimental simulation results, the specific simulation of which is as follows: 1. Simulation conditions: 1) The number of Transformer layers in AF mode and the traditional text semantic communication mode deepSC is 4; 2) The number of Transformer layers in DF mode is 2; 3) The selected channel is an AWGN channel; 4) The comparison schemes are deepSC network, Huffman coding + RS coding, and Huffman coding + Turbo coding; 5) The signal-to-noise ratio is improved by 3dB through intelligent semantic relay signal compensation.

[0065] 6) The evaluation index is the BLEU score, which ranges from 0 to 1. The higher the score, the stronger the semantic communication device's ability to reproduce information content.

[0066] 2. Simulation content: 1) For the AF mode, simulations were conducted to compare the relationship between the performance of semantic communication and the signal-to-noise ratio under different coding schemes. The results are as follows: Figure 6 As shown. Figure 6 In the graph, the vertical axis represents the signal-to-noise ratio (SNR), and the horizontal axis represents the BLEU score.

[0067] Depend on Figure 6Simulation results show that, in the AF mode, the solution provided by this invention, under low signal-to-noise ratio conditions, using Huffman coding + RS coding or Huffman coding + Turbo coding results in a very small proportion of valid messages and a large amount of noise in the traditional communication transmission, leading to a high bit error rate and poor BLEU score. Meanwhile, due to the impact of signal fading, semantic communication devices assisted by intelligent semantic relays perform better than deepSC networks without intelligent semantic relay assistance. This is because intelligent semantic relays compensate for fading signals, improving the signal-to-noise ratio.

[0068] 2) For the DF mode, we first simulated a scenario where the background differences between the knowledge bases of the sending and receiving ends are not significant. The results are as follows: Figure 7 As shown in the figure. Next, a simulation was conducted where the knowledge base backgrounds of the sending and receiving ends differed significantly, and the results are as follows. Figure 8 ,exist Figure 7 and 8 In the graph, the horizontal axis represents the signal-to-noise ratio (SNR), and the vertical axis represents the BLEU score.

[0069] Depend on Figure 7 Simulation results show that when the knowledge background is mismatched, deepSC is more susceptible to signal fading and noise interference. At low signal-to-noise ratios (SNR), AF and DF modes are more stable due to signal compensation. With increasing SNR, both deepSC and AF modes reach their performance limits due to a consistent training environment. Although this invention features re-encoding / decoding capabilities, and DF mode outperforms AF mode overall, the performance difference is small because the difference in knowledge background is not significant.

[0070] Depend on Figure 8 Simulation results show that the overall BLEU score of AF mode and deepSC is poor and does not increase with the increase of SNR. This is due to the significant difference in knowledge background between the transmitter and receiver, and is not significantly related to the wireless channel environment. Therefore, compensating for signal power cannot effectively improve communication quality. The DF mode provided by this invention solves the problem of knowledge background mismatch due to the re-encoding and decoding function, thus achieving better overall performance.

[0071] In summary, this invention proposes an intelligent semantic relay-assisted smart semantic text transmission technology. Simulation results show that the semantic communication device supported by the intelligent semantic relay can reduce the impact of channel fading (e.g., path loss, obstacle interference) and has better performance in low SNR environments. Furthermore, in cases of knowledge background mismatch, the DF mode exhibits good performance through recoding and recoding functions. Therefore, the DF mode can realize one-to-many or many-to-many semantic communication.

[0072] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0073] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0074] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A semantic text transmission device based on semantic relay assistance, characterized in that, The device includes m semantic transmitters, n semantic receivers, and an intelligent semantic relay, where m and n are both positive integers greater than or equal to 1; The semantic transmitting end is used to perform semantic encoding on the text information and then perform channel encoding to generate the first transmission signal; The intelligent semantic relay is used to receive the first transmitted signal, perform power compensation on the first transmitted signal, and generate a second transmitted signal; The semantic receiver receives the second transmitted signal, performs channel decoding on the second transmitted signal, and then performs semantic decoding to complete the semantic communication of the three parties. When m=1, n=1, and the semantic sending end and the semantic receiving end share the same knowledge base, the intelligent semantic relay uses the amplification forwarding mode to complete the transmission of the text information; when and / or At the same time, the semantic sending end and the semantic receiving end use different knowledge bases, and the intelligent semantic relay uses a decoding and forwarding mode to complete the transmission of the text information.

2. The semantic text transmission device based on semantic relay assistance as described in claim 1, characterized in that, When the intelligent semantic relay completes the text information transmission using the amplification forwarding mode, the intelligent semantic relay is used to receive the first transmitted signal, perform power compensation on the first transmitted signal, and generate a second transmitted signal.

3. The semantic text transmission device based on semantic relay assistance as described in claim 1, characterized in that, When the intelligent semantic relay completes the transmission of the text information using the decoding and forwarding mode, each semantic sender performs first semantic encoding on the corresponding text information, and then performs first channel encoding to generate the corresponding third transmission signal.

4. The semantic text transmission device based on semantic relay assistance as described in claim 3, characterized in that, The intelligent semantic relay is based on a first knowledge base shared with the semantic transmitter. First, it amplifies the power of each third transmitted signal. Then, it calls the corresponding neural network parameters and performs first channel decoding on the amplified signal through the neural network. Finally, it performs first semantic decoding on the first channel decoded signal to obtain the first decoded signal.

5. A semantic text transmission device based on semantic relay assistance as described in claim 4, characterized in that, After obtaining the first decoded signal, the intelligent semantic relay, based on the second knowledge base shared with the semantic receiver, first performs second semantic encoding, and then performs second channel encoding to generate the corresponding fourth transmission signal; Secondly, the intelligent semantic relay transmits the fourth transmission signal to the corresponding semantic receiver via a wireless channel.

6. A semantic text transmission device based on semantic relay assistance as described in claim 5, characterized in that, After each semantic receiver receives the corresponding fourth transmission signal, each semantic receiver performs second channel decoding on the fourth transmission signal and then performs second semantic decoding to complete the semantic communication of the three parties.

7. A semantic text transmission method based on semantic relay assistance, characterized in that, The method applied to the semantic text transmission apparatus based on semantic relay assistance as described in any one of claims 1-6 includes: The intelligent semantic relay receives the first transmission signal, performs power compensation on the first transmission signal, generates a second transmission signal, and sends it to the semantic transmitter. The first transmission signal is generated by the semantic transmitter after semantically encoding the text information and then performing channel coding.

Images