Ad hoc network voice transmission test platform and method

Abstract

The invention discloses an ad hoc network voice transmission test platform and method, and the platform comprises a time synchronization module which achieves the 100-microsecond time synchronization between a PC and an ad hoc network device; the voice program deployment module is used for deploying a special voice program, and the performance monitoring unit monitors various equipment parameters per second; the timestamp recording module is used for recording data packet sending and receiving moments in related equipment; the data analysis module is used for calculating performance indexes such as packet loss rate; and the protocol support and analysis module identifies and decodes different protocol data packets. The test method comprises the steps of time synchronization, program deployment, timestamp recording, performance index calculation, protocol support and analysis and the like. Wherein a deviation prediction model construction unit of the time synchronization module periodically samples and constructs a model, a clock frequency adjustment unit adjusts the frequency in a grading manner, a timestamp recording module records and performs weighted fusion at a relay node, and a data analysis module enables the accuracy to be high, so that the performance in multiple aspects such as the accuracy of an ad hoc network voice transmission test is improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technology applications, and in particular to a self-organizing network voice transmission test platform and method. Background Art

[0002] Currently, in ad hoc network communication environments, accurate evaluation of end-to-end voice data transmission performance is crucial, especially for key indicators such as packet loss rate, disorder rate, delay and jitter. Traditional test tools are often limited by low-precision time synchronization mechanisms, resulting in low reliability of measurement results and unable to meet the growing demand for high-performance communications.

[0003] Common self-organizing network voice transmission tests usually rely on standard network monitoring tools such as ping and traceroute. However, these tools perform poorly in dealing with large-scale, dynamically changing self-organizing network environments, especially in terms of time synchronization, which lacks sufficient accuracy. Due to the low accuracy of time synchronization and the inability to accurately capture the characteristics of high-speed changing data streams, the accuracy of the measurement results will be directly affected, which in turn affects the overall evaluation of network quality and cannot meet the working requirements of communication technology. Therefore, a self-organizing network voice transmission test platform and method are proposed. Summary of the invention

[0004] The present invention provides the following technical solution: a self-organizing network voice transmission test platform, comprising: A time synchronization module, which is used to achieve time synchronization between the test software PC and the sending and receiving devices of the ad hoc network, and ensure that the time synchronization accuracy is controlled within the level of hundreds of microseconds. The time synchronization module is internally integrated with a machine learning algorithm module, a quantum clock interface and a multi-source time synchronization unit. The machine learning algorithm module is internally equipped with a time deviation prediction model construction unit and a clock frequency adjustment unit; A voice program deployment module, which is used to deploy a special voice sending program to a voice sending terminal device of an ad hoc network on the one hand, and to deploy a special voice receiving program to a voice receiving terminal device of an ad hoc network on the other hand. A performance monitoring unit is added to the voice program deployment module, and a parameter set that can be adjusted is built into the voice program inside the voice program deployment module; A timestamp recording module, which is installed on the voice sending end device and the receiving end device of the ad hoc network, and uses the high-precision timestamp on the device to record the sending time and receiving time of the data packet; A data analysis module, which is used to analyze the data packet sequence number to calculate the packet loss rate and disorder rate, and obtain the delay and jitter results based on the time difference between packet sending and receiving. The data analysis module is internally equipped with a deep learning model training unit and an additional prediction unit; The protocol support and analysis module has a protocol analysis unit installed inside the module, and the protocol analysis unit is used to quickly identify and decode data packet structures under different protocols to ensure that the accuracy of the test results is not affected by differences in network protocols.

[0005] The present invention provides a method for testing voice transmission in an ad hoc network, comprising the following steps: S1 time synchronization; The time synchronization module is used to achieve time synchronization between the test software PC and the sending and receiving devices of the ad hoc network. The quantum clock interface and multi-source time synchronization unit are used to ensure that the time synchronization accuracy is controlled within the hundreds of microseconds level. S2 program deployment; The special voice sending program is deployed to the voice sending end device of the self-organizing network through the voice program deployment module, and the special voice receiving program is deployed to the voice receiving end device of the self-organizing network; S3 timestamp records; The timestamp recording module on the voice sending device, receiving device and key relay node of the ad hoc network uses the high-precision timestamp on the device to record the sending time and receiving time of the data packet respectively; S4 performance index calculation; The data analysis module is used to analyze the data packet sequence number to calculate the packet loss rate and disorder rate, and the delay and jitter results are calculated based on the time difference between packet sending and receiving; S5 protocol support and analysis; The protocol parsing unit uses a hierarchical parsing method based on protocol features to identify and decode the data packet structure under different protocols. During the parsing process, the protocol parsing unit regularly obtains updated protocol information from the network and automatically updates its parsing rule base to cope with the continuous updating and expansion of the protocol.

[0006] Preferably, the deviation prediction model construction unit adopts the historical time synchronization data collection method as periodic sampling collection, that is, the time synchronization data is collected every 5 minutes according to the preset time period, and the collected data needs to cover multiple levels, including but not limited to the frequency deviation of the clock, phase deviation and the impact of environmental factors on time synchronization. These data are used to construct the time deviation prediction model. At the same time, through the analysis of a large amount of periodically collected data, the validity and integrity of the data are ensured. In the data collection process, a data verification mechanism is also provided to perform real-time verification on the collected data and remove abnormal data to improve the accuracy of model construction.

[0007] Preferably, when the clock frequency adjustment unit dynamically adjusts the clock frequency based on the prediction results, the adjustment strategy adopted is a graded adjustment. When the predicted time drift trend is small, a fine-tuning strategy is adopted, and the amplitude of each adjustment of the clock frequency is 0.01%-0.05% of the original frequency. When the time deviation is large, the adjustment amplitude is increased to 0.05%-0.1% of the original frequency.

[0008] Preferably, the monitoring parameters of the performance monitoring unit include CPU utilization, memory usage, network bandwidth, device temperature and battery power. The performance monitoring unit monitors these parameters once per second. When it is detected that the device temperature exceeds a preset threshold, a signal is sent to the voice program deployment module. The voice program deployment module adjusts the adjustable parameter set in the voice program according to the received signal and the built-in strategy.

[0009] Preferably, the timestamp recording module is also installed on key relay nodes in the ad hoc network, and when data packets pass through these relay nodes, timestamps will also be recorded, and during data analysis, a weighted algorithm is used to fuse the timestamps recorded by each node.

[0010] Preferably, the deep learning model training unit adopts a hybrid model combining a convolutional neural network and a recurrent neural network. During the model training process, the deep learning model training unit adopts a cross-validation method to evaluate the performance of the model, and adjusts the hyperparameters of the model according to the verification results. These parameters include the size of the convolution kernel and the number of layers of the neural network. After multiple iterative training, the model can accurately learn the complex relationship between performance indicators such as packet loss rate, disorder rate, delay and jitter from the input data, thereby improving the accuracy of performance indicator calculation.

[0011] Preferably, the protocol parsing unit adopts a layered parsing method based on protocol features when identifying and decoding data packet structures under different protocols. During the parsing process, in order to cope with the continuous updating and expansion of the protocol, the protocol parsing unit will regularly obtain protocol update information from the network and automatically update its parsing rule base.

[0012] Preferably, the multi-source time synchronization unit in step S1 adopts a priority sorting mechanism when acquiring multi-source time signals, and sorts the time signals from different sources according to pre-set priorities, and the priority is set based on factors such as the historical stability, accuracy and adaptability of each time source to the self-organizing network environment.

[0013] Preferably, the selection of key relay nodes in step S3 is based on a dynamic evaluation mechanism, that is, the nodes in the network perform a self-evaluation at regular intervals, and the evaluation parameters include the connection stability, data traffic load, and data forwarding delay of the nodes.

[0014] In summary, compared with the prior art, the present invention provides a self-organizing network voice transmission test platform and method, which has the following beneficial effects: The invention can achieve time synchronization with an accuracy of hundreds of microseconds between the test software PC and the self-organizing network device through the added time synchronization module, and the internal multi-source time synchronization unit adopts a priority sorting mechanism to obtain multi-source time signals, and its priority is set based on multiple factors. There is also a quantum clock interface to improve accuracy, and the deviation prediction model construction unit periodically samples and collects multi-level data to build a model, and the data verification mechanism ensures data validity. The clock frequency adjustment unit adjusts the clock frequency in stages to adapt to different time drift conditions, and the voice program deployment module can deploy special voice transceiver programs. Its performance monitoring unit monitors multiple device parameters every second. When the device is over-temperature, it can notify the voice program deployment module to adjust the voice program parameter set. The added timestamp recording module records timestamps at the sending end, receiving end and key relay nodes, and adopts weighted algorithm fusion during data analysis, which helps to accurately obtain the sending and receiving time information of the data packet, and cooperates with the deep learning model training unit in the data analysis module to adopt a hybrid model and adjust the hyperparameters through cross-validation. After multiple iterative training, the complex relationship between performance indicators is accurately learned, thereby improving the accuracy of performance indicator calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the test platform module of the present invention.

[0016] Figure 2 It is a flow chart of the steps of the self-organizing network voice transmission test method of the present invention.

[0017] Figure 3 It is a schematic diagram of the time synchronization module of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] See also Figure 1 The present invention provides a technical solution, a self-organizing network voice transmission test platform, comprising: Time synchronization module: The time synchronization module is used to achieve time synchronization between the test software PC and the sending and receiving devices of the self-organizing network, and ensure that the time synchronization accuracy is controlled within the level of hundreds of microseconds. The time synchronization module establishes the time connection between the test software PC and the sending and receiving devices of the self-organizing network through the quantum clock interface and the multi-source time synchronization unit. The quantum clock interface provides an accurate time reference with its high-precision clock signal, and the multi-source time synchronization unit integrates multiple time sources to ensure stable time synchronization in various complex environments. Please refer to Figure 3 The time synchronization module integrates a machine learning algorithm module, a quantum clock interface and a multi-source time synchronization unit. The machine learning algorithm module is equipped with a time deviation prediction model construction unit and a clock frequency adjustment unit. The deviation prediction model construction unit adopts the historical time synchronization data collection method as periodic sampling collection, that is, according to the preset time period, the time synchronization data is collected every 5 minutes, and the collected data needs to cover multiple levels, including but not limited to the frequency deviation and phase deviation of the clock and the impact of environmental factors on time synchronization. These data are used to build the time deviation prediction model. At the same time, the validity and integrity of the data are ensured by analyzing a large amount of periodically collected data. This periodicity Sampling can obtain data that is sufficient to reflect the state of time synchronization without excessive consumption of resources. Since the collected data covers multiple levels, the state of time synchronization can be fully understood. In the process of data collection, a data verification mechanism is also provided to verify the collected data in real time and remove abnormal data to improve the accuracy of model construction. When the clock frequency adjustment unit dynamically adjusts the clock frequency based on the prediction results, the adjustment strategy adopted is hierarchical adjustment. When the predicted time drift trend is small, a fine-tuning strategy is adopted. The amplitude of each adjustment of the clock frequency is 0.01%-0.05% of the original frequency. When the time deviation is large, the adjustment amplitude is increased to 0.05%-0.1% of the original frequency. The hardware implementation of the time synchronization module involves the hardware connection of the quantum clock interface and the hardware implementation of the multi-source time synchronization unit. The quantum clock interface includes a specific clock chip and a high-precision clock source device. These hardware devices are connected to the relevant hardware interfaces of the ad hoc network device and the test software PC through physical lines to provide accurate clock signals. The multi-source time synchronization unit has multiple clock input interfaces for receiving time signals from different sources, and processes, compares and selects signals through hardware circuits, and determines the final synchronization time signal according to the priority sorting mechanism. The quantum clock interface can use a programmable quantum system controller model PQSC, which has a ZSync proprietary interface, 18 ports, and a cable length of 3 meters. It can selectively lock to an external 10 MHz or 100 MHz reference clock and provide an internal reference clock (i.e., 10 The quantum clock interface uses a cesium atomic clock chip as the external interface for MHz clock signals. The chip uses the specific energy level transition of cesium atoms to determine the precise time interval. Inside the chip, there are special atomic excitation and detection circuits that precisely control the state of cesium atoms to generate a stable clock signal. Therefore, the clock signal accuracy generated by this chip can reach nanoseconds or even picoseconds, which is very suitable as the high-precision clock source part in the quantum clock interface. See also Figure 1 , Voice program deployment module, the voice program deployment module is used to deploy a special voice sending program to the voice sending end device of the self-organizing network on the one hand, and to deploy a special voice receiving program to the voice receiving end device of the self-organizing network on the other hand. A performance monitoring unit is added inside the voice program deployment module. The voice program inside the voice program deployment module has a built-in parameter set that can be adjusted. The monitoring parameters of the performance monitoring unit include CPU utilization, memory usage, network bandwidth, device temperature and battery power. The performance monitoring unit monitors these parameters once per second. When it is detected that the device temperature exceeds the preset threshold, a signal will be sent to the voice program deployment module. This High-frequency monitoring can timely capture changes in the device's operating status. For example, monitoring of CPU utilization can reflect the device's computing resource usage, memory usage reflects the device's storage resource utilization, network bandwidth monitoring helps understand network transmission capabilities, device temperature monitoring can prevent device overheating, and battery power monitoring can ensure that the device works normally when it has sufficient power. The voice program deployment module adjusts the adjustable parameter set in the voice program based on the received signal and the built-in strategy. This adjustment includes adjusting the voice encoding rate and transmission power parameters to adapt to the impact of device temperature changes and ensure the stable operation of the voice program. The hardware implementation of the voice program deployment module is related to the storage and processing hardware of the ad hoc network device. In the ad hoc network voice sending and receiving devices, multiple sets of storage cards and hard disks are set to store special voice sending and receiving programs. At the same time, the processors of the ad hoc network voice sending and receiving devices need to have sufficient computing power to run the voice program deployment module to ensure that the program can be smoothly deployed to the device's operating environment; Timestamp recording module: The timestamp recording module is installed on the voice sending and receiving devices of the self-organizing network, and uses the high-precision timestamp on the device to record the sending and receiving times of the data packet. This high-precision timestamp can accurately mark the transmission time point of the data packet in the network, which is crucial for subsequent performance analysis, including the calculation of delay and jitter indicators. The timestamp recording module is also installed on the key relay nodes in the self-organizing network, and when the data packet passes through these relay nodes, the timestamp will also be recorded. In the self-organizing network, the relay nodes play the role of forwarding data packets. Recording the timestamps on these nodes can more comprehensively understand the transmission path and time consumption of the data packet in the network. When analyzing the data, a weighted algorithm is used to fuse the timestamps recorded by each node. The weighted algorithm will assign different weights to the timestamps of each node according to factors such as the importance of the node and the stability of data transmission, and then fuse them, so that more accurate data packet transmission time information can be obtained; The hardware implementation of the timestamp recording module includes high-precision clock hardware and data acquisition hardware. In the ad hoc network devices and relay nodes, the high-precision clock hardware provides an accurate time reference, and the data acquisition hardware is responsible for collecting information related to the timestamp when the data packet is sent, received or passes through the relay node. For example, by integrating the timestamp acquisition circuit in the network interface chip, the accurate recording of the data packet time information can be achieved; Data analysis module: The data analysis module is used to analyze the data packet sequence number to calculate the packet loss rate and disorder rate, and obtain the delay and jitter results based on the time difference between packet sending and receiving. This method can determine whether the data packet is lost or disordered by analyzing the data packet sequence number. For example, if a data packet with a certain number is missing according to the sequence number of the data packet, the packet loss rate can be calculated. If the receiving order of the data packet is inconsistent with the sending order, the disorder rate can be counted. According to the time difference between packet sending and receiving, the transmission delay of the data packet in the network can be calculated. By analyzing the delay changes of multiple data packets, the jitter index can be obtained. The data analysis module is equipped with a deep learning model training unit and an additional prediction unit. The deep learning model training unit adopts a hybrid model combining convolutional neural networks and recurrent neural networks. Convolutional neural networks are good at processing the spatial characteristics of data, while recurrent neural networks have advantages in processing sequence data. The combination of the two can better mine hidden information from the relevant data of the data packet. During the model training process, the deep learning model training unit adopts a cross-validation method to evaluate the performance of the model and adjust the model according to the verification results. The parameters of the deep learning model training unit are as follows: these parameters include the size of the convolution kernel and the number of layers of the neural network, and after multiple iterations of training, the model can accurately learn the complex relationship between performance indicators such as packet loss rate, disorder rate, delay and jitter from the input data, thereby improving the accuracy of performance indicator calculation. In addition, cross-validation divides the data set into multiple subsets, and takes one of the subsets as the validation set and the remaining subsets as the training set in turn, so that the generalization ability of the model can be more comprehensively evaluated. By adjusting the hyperparameters, the structure of the model can be optimized, so that it can more accurately learn the complex relationship between performance indicators such as packet loss rate, disorder rate, delay and jitter from the input data, thereby improving the accuracy of performance indicator calculation. The input data format of the deep learning model training unit adopts mixed structure data, specifically a mixture of numerical and categorical data, that is, a combination of a multidimensional array and sequence data. For data packet related data, the header information, load part, etc. of the data packet are encoded into a multidimensional array according to certain rules to reflect the spatial structural relationship of the data. Secondly, the training data set source of the deep learning model training unit includes network monitoring data and simulated network environment data. The network monitoring data is obtained from the network monitoring device. These devices can monitor the transmission of data packets in the network in real time and record various characteristics of each data packet, such as source address, destination address, packet length, transmission time, etc.Through long-term monitoring, a large amount of data packet related data can be accumulated, and the simulated network environment data is generated by setting different network topologies, traffic patterns, fault scenarios, etc. For example, different bandwidth limits, packet loss probabilities, delay parameters, etc. are set in the simulated network, and then the data packets are transmitted in the simulated environment, and the characteristic data of the data packets in various simulated scenarios are recorded. These data are used as a supplement to the training data set. At the same time, the training process of the deep learning model includes data preprocessing stage, data set division stage, model initialization stage, iterative training stage, model evaluation and adjustment stage and final model determination stage;. The data analysis module is installed in an independent data analysis server. Its hardware implementation includes the server's processor, memory, storage and other hardware resources. The server's high-performance processor can quickly process a large amount of data packets, the large-capacity memory is used for temporary storage and calculation of data, and the high-speed storage device is used to store analysis results and historical data. The protocol support and parsing module is equipped with a protocol parsing unit. The protocol parsing unit is used to quickly identify and decode the data packet structure under different protocols. Since the self-organizing network will adopt multiple network protocols, the structure and format of the data packet under different protocols are different. The protocol parsing unit can accurately identify the protocol type to which the data packet belongs and decode it through its specific algorithms and rules, so as to obtain the valid information in the data packet and ensure that the accuracy of the test results is not affected by the differences in network protocols. When identifying and decoding the data packet structure under different protocols, the protocol parsing unit adopts a hierarchical parsing method based on protocol features. During the parsing process, in order to cope with the continuous updating and expansion of the protocol, the protocol parsing unit will regularly obtain updated information of the protocol from the network and automatically update its parsing rule base.

[0020] See also Figure 2 , the ad hoc network voice transmission test method of the present invention comprises the following steps; S1 time synchronization; The time synchronization module is used to achieve time synchronization between the test software PC and the sending and receiving devices of the ad hoc network. The quantum clock interface and the multi-source time synchronization unit are used to ensure that the time synchronization accuracy is controlled within the level of hundreds of microseconds. When the multi-source time synchronization unit obtains multi-source time signals, a priority sorting mechanism is adopted to sort the time signals from different sources according to the pre-set priority. The priority setting is based on the historical stability, accuracy and adaptability of each time source to the ad hoc network environment. S2 program deployment; The special voice sending program is deployed to the voice sending end device of the self-organizing network through the voice program deployment module, and the special voice receiving program is deployed to the voice receiving end device of the self-organizing network; S3 timestamp records; The timestamp recording module on the voice sending device, receiving device and key relay node of the ad hoc network uses the high-precision timestamp on the device to record the sending and receiving time of the data packet respectively. The selection of key relay nodes is based on a dynamic evaluation mechanism, that is, the nodes in the network will conduct a self-evaluation at regular intervals. The evaluation parameters include the connection stability of the node, data traffic load, and data forwarding delay. S4 performance index calculation; The data analysis module is used to analyze the data packet sequence number to calculate the packet loss rate and disorder rate, and the delay and jitter results are calculated based on the time difference between packet sending and receiving; S5 protocol support and analysis; The protocol parsing unit uses a hierarchical parsing method based on protocol features to identify and decode the data packet structure under different protocols. During the parsing process, the protocol parsing unit regularly obtains updated protocol information from the network and automatically updates its parsing rule base to cope with the continuous updating and expansion of the protocol.

[0021] This solution can achieve time synchronization with an accuracy of hundreds of microseconds between the test software PC and the self-organizing network equipment through the time synchronization module. The internal multi-source time synchronization unit adopts a priority sorting mechanism to obtain multi-source time signals, and its priority is set based on multiple factors. There is also a quantum clock interface to improve accuracy. The deviation prediction model construction unit periodically samples and collects multi-level data to build a model. The data verification mechanism ensures data validity. The clock frequency adjustment unit hierarchically adjusts the clock frequency to adapt to different time drift situations. Secondly, the added voice program deployment module can deploy special voice transceiver programs. Its performance monitoring unit monitors multiple device parameters every second. When the device is over-temperature, it can notify the voice program deployment module to adjust the voice program parameter set. The timestamp recording module records timestamps at the sending end, receiving end and key relay nodes. The weighted algorithm fusion is used in data analysis, which helps to accurately obtain the sending and receiving time information of the data packet. The deep learning model training unit in the data analysis module uses a hybrid model and adjusts the hyperparameters through cross-validation. After multiple iterative training, it accurately learns the complex relationship between performance indicators and improves the accuracy of performance indicator calculation. Finally, the protocol parsing unit in the protocol support and parsing module uses a hierarchical parsing method to identify the structure of the decoded data packet, and can also regularly update the parsing rule base to cope with protocol updates and expansions to ensure that the test results are not affected by network protocol differences. These aspects work together to ensure accuracy, reliability, adaptability and other aspects of self-organizing network voice transmission testing; The time synchronization module added in this solution: realizes the time synchronization between PC and self-organizing network devices at the level of 100 microseconds, and has a machine learning algorithm module inside. The multi-source time synchronization unit obtains multi-source time signals according to priority, the deviation prediction model construction unit periodically samples multi-level data to build a model and has a data verification mechanism, the clock frequency adjustment unit hierarchically adjusts the clock frequency, the voice program deployment module: deploys a special voice transceiver program, and its performance monitoring unit monitors multiple device parameters every second. When the device is overheated, the notification module adjusts the voice program parameters, the timestamp recording module: records timestamps at the sending end, the receiving end and the key relay nodes, and uses a weighted algorithm to fuse node timestamps, the data analysis module: calculates indicators such as packet loss rate, disorder rate, delay and jitter, and the deep learning model training unit uses a hybrid model and cross-validation to adjust hyperparameters and protocol support and analysis module: the protocol analysis unit uses a hierarchical analysis method to identify the structure of the decoded data packet and regularly updates the analysis rule base.

[0022] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0023] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. Ad hoc network voice transmission test platform, characterized by: include: A time synchronization module, which is used to achieve time synchronization between the test software PC and the sending and receiving devices of the ad hoc network, and ensure that the time synchronization accuracy is controlled within the level of hundreds of microseconds. The time synchronization module is internally integrated with a machine learning algorithm module, a quantum clock interface and a multi-source time synchronization unit. The machine learning algorithm module is internally equipped with a time deviation prediction model construction unit and a clock frequency adjustment unit; A voice program deployment module, which is used to deploy a special voice sending program to a voice sending terminal device of an ad hoc network on the one hand, and to deploy a special voice receiving program to a voice receiving terminal device of an ad hoc network on the other hand. A performance monitoring unit is added to the voice program deployment module, and a parameter set that can be adjusted is built into the voice program inside the voice program deployment module; A timestamp recording module, which is installed on the voice sending end device and the receiving end device of the ad hoc network, and uses the high-precision timestamp on the device to record the sending time and receiving time of the data packet; A data analysis module, which is used to analyze the data packet sequence number to calculate the packet loss rate and disorder rate, and obtain the delay and jitter results based on the time difference between packet sending and receiving. The data analysis module is internally equipped with a deep learning model training unit and an additional prediction unit; The protocol support and analysis module has a protocol analysis unit installed inside the module, and the protocol analysis unit is used to quickly identify and decode data packet structures under different protocols to ensure that the accuracy of the test results is not affected by differences in network protocols.

2. The ad hoc network voice transmission test platform according to claim 1, characterized in that: The deviation prediction model construction unit adopts the historical time synchronization data collection method as periodic sampling collection. Time synchronization data is collected every 5 minutes according to the preset time period, and the collected data needs to cover multiple levels, including but not limited to the frequency deviation and phase deviation of the clock and the impact of environmental factors on time synchronization. These data are used to construct the time deviation prediction model. At the same time, through the analysis of a large amount of periodically collected data, the validity and integrity of the data are ensured. In the data collection process, a data verification mechanism is also provided to perform real-time verification on the collected data and remove abnormal data to improve the accuracy of model construction.

3. The ad hoc network voice transmission test platform according to claim 1, characterized in that: When the clock frequency adjustment unit dynamically adjusts the clock frequency based on the prediction results, the adjustment strategy adopted is hierarchical adjustment. When the predicted time drift trend is small, a fine-tuning strategy is adopted. The amplitude of each adjustment of the clock frequency is 0.01%-0.05% of the original frequency. When the time deviation is large, the adjustment amplitude is increased to 0.05%-0.1% of the original frequency.

4. The ad hoc network voice transmission test platform according to claim 1, characterized in that: The monitoring parameters of the performance monitoring unit include CPU utilization, memory usage, network bandwidth, device temperature and battery power. The performance monitoring unit monitors these parameters once per second. When it is detected that the device temperature exceeds a preset threshold, a signal is sent to the voice program deployment module. The voice program deployment module adjusts the adjustable parameter set in the voice program according to the received signal and the built-in strategy.

5. The ad hoc network voice transmission test platform according to claim 1, characterized in that: The timestamp recording module is also installed on key relay nodes in the ad hoc network. When data packets pass through these relay nodes, timestamps will also be recorded. During data analysis, a weighted algorithm is used to fuse the timestamps recorded by each node.

6. The ad hoc network voice transmission test platform according to claim 1, characterized in that: The deep learning model training unit adopts a hybrid model that combines a convolutional neural network and a recurrent neural network. During the model training process, the deep learning model training unit uses a cross-validation method to evaluate the performance of the model, and adjusts the parameters of the model according to the verification results. These parameters include the size of the convolution kernel and the number of layers of the neural network. After multiple iterative training, the model can accurately learn the complex relationship between the packet loss rate, disorder rate, delay and jitter performance indicators from the input data, thereby improving the accuracy of the performance indicator calculation.

7. The ad hoc network voice transmission test platform according to claim 1, characterized in that: The protocol parsing unit adopts a hierarchical parsing method based on protocol features when identifying and decoding data packet structures under different protocols. During the parsing process, in order to cope with the continuous updating and expansion of the protocol, the protocol parsing unit will regularly obtain protocol update information from the network and automatically update its parsing rule base.

8. A method for testing voice transmission in an ad hoc network, comprising a test platform for voice transmission in an ad hoc network as claimed in any one of claims 1 to 7, characterized in that: The steps include: S1 time synchronization; The time synchronization module is used to achieve time synchronization between the test software PC and the sending and receiving devices of the ad hoc network. The quantum clock interface and multi-source time synchronization unit are used to ensure that the time synchronization accuracy is controlled within the hundreds of microseconds level. S2 program deployment; The special voice sending program is deployed to the voice sending end device of the self-organizing network through the voice program deployment module, and the special voice receiving program is deployed to the voice receiving end device of the self-organizing network; S3 timestamp records; The timestamp recording module on the voice sending device, receiving device and key relay node of the ad hoc network uses the high-precision timestamp on the device to record the sending time and receiving time of the data packet respectively; S4 performance index calculation; The data analysis module is used to analyze the data packet sequence number to calculate the packet loss rate and disorder rate, and the delay and jitter results are calculated based on the time difference between packet sending and receiving; S5 protocol support and analysis; The protocol parsing unit uses a hierarchical parsing method based on protocol features to identify and decode the data packet structure under different protocols. During the parsing process, the protocol parsing unit regularly obtains updated protocol information from the network and automatically updates its parsing rule base to cope with the continuous updating and expansion of the protocol.

9. The ad hoc network voice transmission test method according to claim 8, characterized in that: The multi-source time synchronization unit in step S1 adopts a priority sorting mechanism when acquiring multi-source time signals, and sorts the time signals from different sources according to a preset priority, and the priority is set based on the historical stability, accuracy and adaptability of each time source to the self-organizing network environment.

10. The ad hoc network voice transmission test method according to claim 8, characterized in that: The selection of key relay nodes in step S3 is based on a dynamic evaluation mechanism, that is, the nodes in the network will perform a self-evaluation at regular intervals, and the evaluation parameters include the connection stability, data traffic load and data forwarding delay of the nodes.

Images