Distributed collaborative waveform switching method and system for networked computing

Abstract

This invention discloses a distributed cooperative waveform switching method and system for networked computing, primarily addressing the problems of excessive latency and high complexity in achieving network-wide consistency through distributed cooperative waveform switching in multi-hop ad hoc network environments. The implementation scheme includes a distributed coordination phase and a distributed decision-making consensus phase. In the distributed coordination phase, each node periodically exchanges state packets and cooperative data packets, and during this exchange, each node performs distributed waveform decision-making and latency information updates. In the distributed decision-making consensus phase, a unique central node is selected based on the cooperative data packets stored in the distributed coordination phase. This central node then makes the optimal waveform switching decision through majority voting based on the waveform decision information of each node stored internally. This invention reduces computational complexity, improves decision-making timeliness and accuracy, ensures network-wide distributed consistency, significantly reduces communication latency, and enhances system performance. It can be used for networked computing in wireless ad hoc networks.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology, specifically relating to a distributed cooperative waveform switching method and system, which can be used for networked computing in wireless ad hoc networks to improve the system's anti-interference capability and communication performance. Background Technology

[0002] Distributed consensus algorithms can be divided into two categories: strong consistency and weak consistency. Strong consistency algorithms include Paxos and Raft, while weak consistency algorithms include eventual consistency and temporary consistency algorithms.

[0003] Distributed collaborative strong consistency algorithms use synchronization mechanisms such as log replication and election protocols to synchronize and coordinate the data updates of each node, so that all participating nodes in the network can maintain a consistent state at any time.

[0004] Distributed collaborative weak consistency algorithms focus more on the high availability and scalability of the system. They allow temporary inconsistencies between some nodes, but the system design ensures that consistency can eventually be achieved. They are suitable for scenarios with high response time requirements and the ability to tolerate a certain degree of inconsistency.

[0005] In distributed collaborative waveform switching within a multi-hop ad hoc network, nodes must communicate under the same waveform to cooperate. Therefore, it is crucial to ensure decision-making consistency among all nodes within the cluster during waveform switching to avoid communication failures or data loss due to waveform inconsistencies. Consequently, a strong consistency mechanism is required to guarantee network-wide waveform consistency during the switching process.

[0006] Currently, the most widely used strong consistency distributed algorithm in the field of distributed collaborative consensus is the Raft algorithm. The Raft algorithm consists of five phases: random timeout election, initiating voting, waiting for responses, majority election, and decision dissemination. In the random timeout election phase, nodes initiate an election after a random timeout, becoming candidate nodes and competing for the leader position. Next, nodes vote for the latest candidate node based on their term number. The candidate node waits for responses from other nodes; if it receives more than half the support, it is elected as the leader. Finally, the leader node disseminates the waveform switching decision based on the majority opinion, ensuring network-wide consistency. While this distributed collaborative consensus algorithm can achieve consistent waveform switching decisions through information exchange and collaborative computation between nodes, improving network communication quality and anti-interference capabilities, the algorithm process is relatively cumbersome, involving multiple phases, resulting in long latency. The randomness of its election mechanism also makes the selection of the leader node less precise, further increasing the latency of decision information dissemination and affecting system performance. Therefore, in multi-hop ad hoc network environments, how to control latency and reduce complexity while ensuring waveform consistency remains a pressing technical challenge.

[0007] Patent document CN202410461259.X discloses a "Method and Apparatus for Building an Elastic Network Based on Distributed Cooperative Nodes". It performs time synchronization and channel quality measurement on multiple self-organizing network forwarding nodes, constructs a cooperative cloud based on these measurement results, and thus forms an elastic network. By achieving time synchronization of multiple self-organizing network forwarding nodes, it ensures that the time base of all nodes in the network is consistent. Through channel quality measurement, it obtains the communication status between nodes, thereby enabling intelligent resource allocation, routing selection, and communication strategy formulation. The elastic network built based on multiple cooperative clouds has high flexibility and scalability, and can dynamically adjust the network structure and resource configuration according to actual needs to cope with various emergencies and communication demands. This method achieves network resource allocation and adaptive adjustment, improving network flexibility. However, the shortcomings of this scheme are that the time synchronization and channel quality measurement processes have high overhead, especially in multi-hop self-organizing network environments. As the number of nodes increases, the latency also increases, affecting real-time performance. Simultaneously, as the number of nodes increases, the computational complexity rises, making it impossible to guarantee system reliability.

[0008] Patent document CN201911170210.4 discloses a "Waveform Design Method for Broadband Frequency Hopping Clustered Multi-Level Self-Organizing Network". It combines distributed, decentralized multi-hop frequency hopping synchronization, initial network establishment, dynamic gateway selection, physical layer modulation / demodulation and encoding / decoding, and link learning-based adaptive routing selection to achieve waveform design for a broadband frequency hopping clustered multi-level self-organizing network. Multi-hop frequency hopping synchronization does not require a preset synchronization center node or a special set of frequency hopping frequencies and patterns; synchronization time is related to factors such as the number of frequency points, hopping rate, and relay hop count. Each node probabilistically transmits a synchronization frame based on its local clock and a preset frequency hopping pattern. Other nodes maintain their synchronization frequency after receiving the frame header, completing frequency hopping synchronization. Link performance is evaluated through a learning strategy to achieve adaptive routing selection, ensuring dynamic adjustment when link quality changes in multi-hop links. While this method achieves flexible multi-hop synchronization and efficient data transmission path selection, improving the network's adaptability and communication efficiency, and avoiding centralized synchronization of the central node, its lack of strong waveform consistency guarantee may lead to waveform inconsistencies between different nodes, thus affecting communication stability. Summary of the Invention

[0009] The purpose of this invention is to address the shortcomings of the prior art by proposing a distributed collaborative waveform switching method and system for networked computing, so as to further improve network reliability, reduce computational complexity, reduce latency, improve decision-making timeliness and accuracy, and ensure network-wide distributed consistency.

[0010] The technical principle for achieving the objective of this invention is as follows: A distributed cooperative waveform switching method is designed by implementing a cooperative consensus algorithm and the OODA ring concept in multi-hop ad hoc networks. The technical solution includes:

[0011] 1. A distributed collaborative waveform switching method for networked computing, characterized in that it includes two stages: a distributed collaborative stage and a distributed decision consensus stage;

[0012] During the distributed collaboration phase, each node in the network periodically exchanges status packets and collaboration data packets to achieve real-time information collaboration and storage between nodes. During this exchange process, each node performs distributed waveform decision-making and latency information updates, dynamically adjusting waveform decision-making and latency data to optimize communication efficiency and improve anti-interference capabilities.

[0013] In the distributed decision-making consensus phase, a unique central node is selected through the collaborative data packet information stored in the distributed collaboration phase. This central node makes the best waveform switching decision by majority vote based on the waveform decision information of each node stored internally, and then broadcasts it to other nodes in the network. After receiving the decision information, other nodes adjust their own waveforms according to the central node's decision to ensure that the entire network achieves consistency in waveform switching, thereby improving the network's collaborative efficiency and reliability.

[0014] Furthermore, the status packet consists of a status packet header and status information, wherein the status packet header includes the status packet type, sequence number, and status information offset; the status information includes network status, interference status, and environmental status.

[0015] Furthermore, the cooperative data packet consists of a cooperative header and cooperative data. The cooperative header includes the cooperative data packet type, sequence number, and cooperative information offset. The cooperative data includes the node address, location information, average latency, maximum latency, waveform decision information, and the node's current role.

[0016] 2. A distributed cooperative waveform switching system for networked computing, characterized in that it comprises:

[0017] The node is configured with a state perception interaction module, which is used to periodically interact with the state packets and collaborative data packets of neighboring nodes. Specifically, it updates the storage of neighboring node state packets by comparing sequence numbers, collects the state information of the node at a period of 1 second and exchanges collaborative data packets with neighboring nodes at a period of 2 seconds.

[0018] The node is configured with a cognitive processing module to build an efficient system for processing state packets and cooperative data packets. This system processes neighboring node state packets and cooperative data packets by removing the state packet header and cooperative packet header, and then allocates them to a designated shared buffer.

[0019] The node is equipped with an anti-interference networking decision module, which takes the state information in the shared buffer as input and performs distributed waveform decision through the reinforcement learning model DQN to make the waveform switching strategy most suitable for the current network environment.

[0020] The node is configured with a dynamic control module for controlling the election of the central node and the final waveform decision. Specifically, the central node is selected from the collaborative data packet information stored by the nodes in the distributed collaboration phase, and the central node makes the best waveform switching decision by majority vote in the distributed decision consensus phase.

[0021] Furthermore, the node is equipped with a dynamic control module, including:

[0022] The node setting center node election submodule is used to elect the node with the shortest communication latency as the center node. Specifically, it compares the average latency of each node with the collaborative data packets of all nodes in the network stored within the node, and prioritizes the node with the lowest average latency. If multiple nodes have the same average latency, it further compares the maximum latency of each node, and prioritizes the node with the lowest maximum latency. If multiple nodes have the same maximum latency, it selects the node with the smallest node address as the unique center node.

[0023] The node sets the final waveform decision submodule, which is used to make the best waveform switching decision by majority vote. That is, after the central node is elected, the central node traverses the waveform decision information of each node stored internally, calculates the frequency of each waveform being selected, finds the waveform with the highest selection frequency as the candidate waveform, and if the candidate waveform is unique, it is selected as the final waveform decision; if it is not unique, the waveform with stronger anti-interference ability is selected as the final waveform decision.

[0024] Compared with the prior art, the present invention has the following advantages:

[0025] First, because the nodes of this invention periodically exchange state packets and coordination data packets through the sensing state interaction module during the distributed coordination phase, they can achieve an efficient coordination mechanism in multi-hop ad hoc networks, significantly improving coordination efficiency and adaptability, reducing communication overhead and system complexity, and enhancing network reliability and responsiveness.

[0026] Secondly, since the nodes of this invention can independently make waveform decisions while sharing decision information with other nodes during the distributed decision consistency stage, they avoid centralized control and excessive synchronization in traditional methods, enabling the system to make decisions quickly and improving the timeliness and accuracy of waveform decisions.

[0027] Third, the present invention ensures distributed consistency of the entire network with minimal latency by electing the node with the shortest communication latency as the central node, and the central node sends decision information to other nodes in the network.

[0028] Fourth, the nodes of this invention make the best waveform switching decision through majority decision, which reduces computational complexity and significantly improves the system's response speed in making waveform switching decisions when faced with changes in the external environment. Attached Figure Description

[0029] Figure 1 This is the overall flowchart of the distributed collaborative waveform switching method for networked computing proposed in this invention;

[0030] Figure 2 yes Figure 1 The distributed collaboration phase implementation sub-flowchart;

[0031] Figure 3 yes Figure 1 Flowchart of the distributed decision-making consensus phase implementation;

[0032] Figure 4 This is a structural block diagram of the distributed collaborative waveform switching system for networked computing according to the present invention;

[0033] Figure 5 The simulation diagrams show the communication delay comparison between the distributed cooperative waveform switching method of this invention and the existing distributed cooperative Raft algorithm for disseminating decision information to the entire network. Detailed Implementation

[0034] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0035] Example 1: A Distributed Cooperative Waveform Switching Method for Networked Computing

[0036] Reference Figure 1 The implementation of this example includes two phases: the distributed coordination phase and the distributed decision consensus phase.

[0037] 1. In the distributed collaboration phase, each node performs inter-node information collaboration and executes distributed waveform decisions and updates latency information.

[0038] Reference Figure 2In the distributed coordination phase, each node in the network periodically exchanges status packets and coordination data packets to achieve real-time information coordination and storage among nodes. During this exchange process, each node performs distributed waveform decision-making and latency information updates, dynamically adjusting the waveform decision-making and latency data to optimize communication efficiency and improve anti-interference capabilities. The steps include:

[0039] Step one: Each node performs inter-node information coordination, that is, periodically exchanging status packets and coordination data packets.

[0040] The status packet consists of a status packet header and status information. The status packet header includes the status packet type, sequence number, and status information offset. The status information includes network status, interference status, and environmental status.

[0041] The collaborative data packet consists of a collaborative header and collaborative data. The collaborative header includes the collaborative data packet type, sequence number, and collaborative information offset. The collaborative data includes the node address, location information, average latency, maximum latency, waveform decision information, and the node's current role.

[0042] The implementation of this step includes the following:

[0043] 1.1) State packet coordination between nodes:

[0044] 1.1.1) Each node collects its internal state information at the network layer every 1 second, merges it with the state information in the state packets of neighboring nodes within three hops stored in the node, and adds a state packet header to the merged information to generate a state packet to be sent.

[0045] 1.1.1) Each node sends status packets to neighboring nodes at a period of 2 seconds through the transmitting network layer-MAC layer-baseband waveform BB path, and receives neighboring node status packets through the baseband waveform BB-MAC layer-receiving network layer path, storing them in the storage space of its own node.

[0046] Through the aforementioned periodic state information coordination process among nodes, each node can obtain the state information of itself and its neighbors within three hops.

[0047] 1.2) Inter-node data packet coordination:

[0048] 1.2.1) Each node exchanges collaborative data packets with a period of 1 second. In this collaborative data, the average delay reflects the overall communication efficiency between the node and other nodes, the maximum delay represents the longest transmission delay from the node to the farthest node, and the node's current role reflects whether the node is a central node or an ordinary node.

[0049] 1.2.2) Each node integrates the collaborative data of all nodes in each cycle, adds it to the collaborative packet header, and exchanges collaborative data packets between nodes according to the sending path of the originating network layer-MAC layer-baseband waveform BB and the receiving path of the baseband waveform BB-MAC layer-receiving network layer.

[0050] Step 2: Within a node, distributed waveform decision-making is performed using the collected state information of neighboring nodes within three hops.

[0051] The distributed waveform decision-making process involves collecting and storing the state packets of neighboring nodes within three hops, and then using a reinforcement learning model (DQN) within the node, which includes an input layer, a hidden layer, and an output layer, to obtain the distributed waveform decision-making result of the current node in the output layer.

[0052] The specific implementation of this step includes the following:

[0053] 2.1) Set up a two-dimensional structure array within the node as storage space to collect and store neighbor node status packets. The first dimension of the two-dimensional structure array represents the node number, and the second dimension represents the time.

[0054] 2.2) When a node receives a status packet from a neighboring node, it updates its storage based on whether the neighboring node's status packet exists in the two-dimensional structure array:

[0055] If the two-dimensional structure array does not have neighboring node state packets, then add the neighboring node state packets to the two-dimensional structure array;

[0056] If the two-dimensional structure array has neighboring node state packets, then compare the sequence number of the neighboring node state packet A in the two-dimensional structure array with the received neighboring node state packet B:

[0057] If the sequence number of B is greater than the sequence number of A, then add the neighboring node's state packet to the two-dimensional structure array;

[0058] Otherwise, ignore the state packet of the neighboring node;

[0059] The sequence number is set to a range of [0, 255]. As a special case, if the sequence number count starts to cycle, that is, the sequence number of the previous neighboring node status packet is 255 and the sequence number of the currently received neighboring node status packet is 0, then the sequence number comparison rule is 1>0>255. The current neighboring node status packet is added to the two-dimensional structure array according to this rule.

[0060] 2.3) Remove the headers of the received neighbor node state packets in the two-dimensional structure array, and allocate the state information to the shared buffer built into the node according to the node number. Maximize the concurrent performance of state information reading and writing by utilizing the efficiency of data storage and access through the shared buffer.

[0061] 2.4) Using the node state information stored in the shared buffer within each node as input, waveform decision calculation is performed based on the DQN reinforcement learning model to select the waveform switching strategy most suitable for the current network environment:

[0062] 2.4.1) Obtain the state information within the shared buffer of the node, including the frame error rate in the network state, the interference frequency domain occupancy in the interference state, and the transmission rate in the environment state, so as to reflect the reliability of data transmission, the degree of interference in the spectrum, and the transmission rate of each node, respectively.

[0063] 2.4.2) Construct a reinforcement learning model DQN consisting of an input layer, a hidden layer, and an output layer. Its input layer takes the state information of each node stored in the shared buffer within the node as input, and the output layer outputs the predicted Q value of each waveform.

[0064] 2.4.3) Define the reward function in the reinforcement learning model DQN: transmission rate * (1 - frame error rate) + interference penalty. This reward function is used to guide the DQN model to learn the optimal waveform switching strategy to adapt to the dynamically changing interference environment.

[0065] 2.4.4) By using the network state information in the shared buffer within the node, the input required for the reward function is obtained, including the current waveform, transmission rate, and frame error rate, and the reward value calculated by the reward function under the current waveform is obtained;

[0066] 2.4.5) Using the reward value as input to the reinforcement learning model DQN, the target Q value is updated using the Q-learning algorithm:

[0067] The Q-learning algorithm is a value-iterative reinforcement learning algorithm that makes decisions by learning the expected future reward of each state-action pair. Specifically, it calculates the immediate reward based on the immediate reward obtained by taking an action in the current state, and updates the target Q-value based on the immediate reward, progressively optimizing the waveform selection strategy. The update formula is as follows:

[0068]

[0069] Where Q(s,a) is the current Q value when waveform selection action a is taken in state s; α is the learning rate, used to control the speed of Q value updates; R is the immediate reward after taking the current waveform selection action; and γ is the discount factor, representing the impact of future rewards. It is the maximum Q value of all possible actions a' under the new state s'.

[0070] 2.4.6) After updating the target Q value using the Q-learning algorithm update formula, the error between the predicted Q value output by the output layer of the reinforcement learning model DQN and the target Q value output by the Q-learning algorithm update formula is calculated. The weights of the model are adjusted by the calculated error to obtain the optimal waveform selection under different state information inputs, which is the waveform decision result of the reinforcement learning model DQN.

[0071] 2.5) After storing the waveform decision results of the reinforcement learning model DQN in the waveform decision information of the local node's collaborative data, waveform switching control is performed according to the current node role in the local node's collaborative data:

[0072] If the current node is not the central node, it will adjust its own waveform after receiving the waveform decision information from the central node.

[0073] If the current node is the central node, then the majority decision is made to determine the best waveform switching method and broadcast it to the entire network.

[0074] Step 3: Update the latency information within the node.

[0075] 3.1) Obtain the maximum latency and average latency of all node-coordinated data packets stored within the node. The maximum latency represents the longest transmission latency from the node to the farthest node, and the average latency represents the overall communication efficiency between this node and other nodes.

[0076] 3.2) Iterate through the maximum latency of all nodes stored within the node, compare the maximum latency of each node with the maximum latency of the current node, check if there is a larger latency value, and update the maximum latency of the current node accordingly:

[0077] If a larger latency value is found, the maximum latency of this node is updated to that larger value;

[0078] If no larger delay value is found, no changes will be made;

[0079] 3.3) During the process of traversing the maximum latency of all nodes stored in the node, the maximum latency information of all nodes stored in the node is accumulated, the average latency is calculated by weighted averaging, and the average latency value of the current node is updated to the average latency calculated by weighted averaging.

[0080] II. In the distributed decision-making consensus phase, each node elects a central node and the central node makes the optimal waveform decision.

[0081] Reference Figure 3In the distributed decision-making consensus phase, a unique central node is selected based on the collaborative data packet information stored in the distributed collaboration phase. This central node makes the optimal waveform switching decision through majority voting using the waveform decision information of each node stored internally, and broadcasts this decision to all other nodes in the network. After receiving this decision information, other nodes adjust their own waveforms according to the central node's decision to ensure that the entire network achieves consistency in waveform switching, thereby improving the network's collaborative efficiency and reliability. The implementation steps include:

[0082] Step four: Elect a central node by using the collaborative data packet information stored in the distributed collaborative phase.

[0083] To effectively avoid efficiency issues caused by excessive latency and improve the overall response speed and decision-making timeliness of the system, it is necessary to select the central node with the shortest communication latency across the entire network by using the maximum latency, average latency, and node address stored in the collaborative data packets of each node. This process includes:

[0084] 4.1) Find the node with the lowest average latency among the collaborative data stored within the node:

[0085] If the average latency of this node is the minimum and unique, then this node becomes the central node and the process ends.

[0086] If the average latency of this node is not the minimum, terminate the process.

[0087] If multiple nodes have the same average latency, proceed to step 4.2.

[0088] 4.2) Compare the maximum latency in the collaborative data stored within each node:

[0089] If the maximum latency of this node is the smallest and unique, then this node becomes the central node and the process ends.

[0090] If the maximum latency of this node is not the minimum, terminate the process.

[0091] If multiple nodes have the same maximum latency, proceed to step 4.3.

[0092] 4.3) Compare the node addresses in the collaborative data stored within each node:

[0093] If the node address of this node is the smallest, then this node becomes the central node and the process ends;

[0094] Otherwise, it means that this node cannot become the central node, so no changes are made and the process ends.

[0095] Step 5: The central node makes the best waveform switching decision by majority vote based on the waveform decision information of each node stored internally.

[0096] 5.1) The central node traverses the waveform decision information of each node stored internally, counts the frequency of each waveform being selected, and finds the waveform with the highest selection frequency as the candidate waveform.

[0097] 5.2) Determine the waveform switching decision based on the attributes of the candidate waveform:

[0098] If there is only one candidate waveform, then that waveform will be used as the final result of the majority decision waveform switching decision.

[0099] If there are multiple candidate waveforms, the waveform with stronger anti-interference capability will be selected as the final result of the majority decision waveform switching.

[0100] Example 2: Distributed Cooperative Waveform Switching System for Networked Computing

[0101] Reference Figure 4 This example system includes: a node setting perception status interaction module 1, a cognitive processing module 2, an anti-interference networking decision module 3, and a dynamic control module 4, wherein the dynamic control module 4 includes a central node election submodule 41 and a final waveform decision submodule 42.

[0102] The perception state interaction module 1 is used to periodically interact with the state packets and cooperative data packets of neighboring nodes. Specifically, it collects the state information of the current node with a period of 1 second and exchanges cooperative data packets with neighboring nodes with a period of 2 seconds. It uses a two-dimensional structure array set within the node as the storage space for the state packets and cooperative data packets, and updates the storage by comparing the sequence numbers of the state packets and cooperative data packets.

[0103] The cognitive processing module 2 is used to construct an efficient state packet and cooperative data packet processing system. It processes the neighbor node state packets and cooperative data packets stored in the two-dimensional structure array by removing the state packet header and cooperative packet header, and allocates the neighbor node state packets and cooperative data packets to the set shared buffer.

[0104] The anti-interference networking decision module 3 takes the state information in the shared buffer as input, performs distributed waveform decision through the reinforcement learning model DQN, makes the waveform switching strategy most suitable for the current network environment, and stores the decision results in the cooperative data packet.

[0105] The dynamic control module 4 is used for the election of the central node and the control of the final waveform decision. It selects a unique central node based on the information of the collaborative data stored in the two-dimensional structure array during the distributed collaboration phase. The central node then makes the optimal waveform switching decision through majority voting during the distributed decision consensus phase. Wherein:

[0106] The node with the shortest communication latency is elected as the central node through the node setting central node election submodule 41. That is, the average latency of each node is compared with the collaborative data in the collaborative data packets of all nodes in the network stored in the node, and the node with the smallest average latency is selected first. If there are multiple nodes with the same average latency, the maximum latency of each node is further compared and the node with the smallest maximum latency is selected first. If there are multiple nodes with the same maximum latency, the node with the smallest node address is selected as the unique central node.

[0107] The final waveform decision submodule is set by nodes to make the best waveform switching decision through majority judgment. That is, after the central node is elected, the central node traverses the waveform decision information of each node stored internally, calculates the frequency of each waveform being selected, finds the waveform with the highest selection frequency as the candidate waveform, and if the candidate waveform is unique, the waveform is selected as the final waveform decision; if it is not unique, the waveform with stronger anti-interference ability is selected as the final waveform decision.

[0108] The effects of this invention can be further illustrated by the following simulation results:

[0109] I. Simulation Conditions

[0110] A container platform is built within a virtual machine using Docker containers to simulate multi-hop networks, and related code is written within this container platform.

[0111] The settings include parameters for network size, network topology, and interference environment simulation.

[0112] By testing the effectiveness of majority decision waveform switching in the face of dynamic environmental changes under different network scales, and comparing it with the existing distributed cooperative Raft algorithm, the differences in communication latency performance between the two methods are evaluated.

[0113] II. Simulation Content and Results

[0114] Under the above simulation conditions, the waveform switching of this invention and the existing distributed cooperative Raft algorithm were performed respectively in a dynamic environment, and the communication delay performance of the two was compared. The results are as follows: Figure 5 As shown.

[0115] from Figure 5 It is evident that existing distributed cooperative Raft algorithms suffer from high communication latency in transmitting waveform decisions from the elected central node to the entire network due to the randomness of the election phase. In contrast, this invention, by electing the central node with the shortest communication latency for transmitting decision information to the entire network, significantly reduces the communication latency of the central node transmitting decision information to the entire network, thereby improving the overall efficiency and stability of the network and ensuring distributed consistency across the entire network.

[0116] It should be noted that the step numbers in the specification and claims of this invention are only for the purpose of clearly describing the embodiments of this invention and facilitating understanding, and their order is not limited.

Claims

1. A distributed collaborative waveform switching method for networked computing, characterized in that, include: The process consists of two phases: distributed collaboration and distributed decision consensus. During the distributed collaboration phase, each node in the network periodically exchanges status packets and collaboration data packets to achieve real-time information collaboration and storage between nodes. During this exchange process, each node performs distributed waveform decision-making and latency information updates, dynamically adjusting waveform decision-making and latency data to optimize communication efficiency and improve anti-interference capabilities. In the distributed decision consensus phase, a unique central node is selected by using the collaborative data packet information stored in the distributed collaboration phase. This central node makes the best waveform switching decision by majority vote using the waveform decision information of each node stored internally, and then broadcasts it to other nodes in the network. After receiving the decision information, other nodes adjust their own waveforms according to the central node's decision to ensure that the entire network achieves consistency in waveform switching, thereby improving the network's collaborative efficiency and reliability. The status packet consists of a status packet header and status information. The status packet header includes the status packet type, sequence number, and status information offset. The status information includes network status, interference status, and environmental status. The cooperative data packet consists of a cooperative header and cooperative data. The cooperative header includes the cooperative data packet type, sequence number, and cooperative information offset. The cooperative data includes the node address, location information, average latency, maximum latency, waveform decision information, and the node's current role. The distributed waveform decision-making process performed within the node includes the following implementation: (4a) The node collects and stores the perceived state of the state packets, and updates the storage based on whether the neighboring node's state packet exists in the storage space when it receives a neighboring node's state packet: If the storage space does not have the neighbor node status packet, then add the neighbor node status packet to the storage space; If the storage space contains neighbor node status packets, compare the sequence numbers of neighbor node status packet A in the storage space with the received neighbor node status packet B: If the sequence number of B is greater than the sequence number of A, then add the neighboring node's state packet to the storage space; Otherwise, ignore the state packet of the neighboring node; (4b) Remove the header of the neighbor node status packet received in (4a) and allocate the status information to the set shared buffer according to the node number to maximize the concurrent performance of data reading and writing; (4c) Using the state information in the shared buffer in (4b) as input, waveform decision calculation is performed based on the DQN reinforcement learning model to select the waveform switching strategy most suitable for the current network environment; (4d) Store the decision result of (4c) in the waveform decision information in the collaborative data, and perform waveform switching control according to the current node role in the collaborative data: If the current node is not the central node, it will adjust its own waveform after receiving the waveform decision information from the central node. If the current node is the central node, then the majority decision is made to determine the best waveform switching method and broadcast it to the entire network.

2. The method according to claim 1, characterized in that: The nodes in the network periodically exchange state packets and coordination packets, which include: (3a) Nodes exchange state packets by periodically sending state packets: (3a1) Each node collects its internal state information at the network layer with a period of 1 second, merges it with the state information in the state packets of neighboring nodes within three hops stored in its internal storage, and adds a state packet header to generate a state packet to be sent. (3a2) Each node sends status packets to neighboring nodes in a 2s period through the transmitting network layer-MAC layer-baseband waveform BB path, and receives neighboring node status packets through the baseband waveform BB-MAC layer-receiving network layer path, and stores them in the storage space of its own node. (3b) Nodes exchange cooperative data packets by periodically sending cooperative data packets: (3b1) Each node exchanges collaborative data packets with a period of 1 second; (3b2) Each node integrates the collaborative data of all nodes in each cycle, adds a collaborative packet header, and exchanges collaborative data packets between nodes according to the same packet sending and receiving path as (3a2).

3. The method according to claim 1, characterized in that: In step (4c), the reinforcement learning model DQN is used for... The implementation of line waveform decision calculation includes: (4c1) Obtain the state information in the shared buffer in (4b), including the frame error rate in the network state, the interference frequency domain occupancy in the interference state, and the transmission rate in the environment state, so as to reflect the reliability of data transmission, the degree of interference in the spectrum, and the transmission rate of each node, respectively. (4c2) Construct a reinforcement learning model DQN with the state information from step (4c1) as input and the predicted Q value for each waveform as output; (4c3) Define the reward function as effective throughput: transmission rate * (1 - frame error rate) + interference penalty, which is used to guide the DQN model to learn the optimal waveform switching strategy to maximize effective throughput; (4c4) Based on the currently selected waveform and the calculated reward, the DQN model updates its target Q value using the Q-learning algorithm. The update formula is as follows: ; Where Q(s,a) is the current Q value after taking action a in state s; α is the learning rate, used to control the speed of Q value updates; R is the immediate reward after taking the current action, i.e., the effective throughput; γ is the discount factor, representing the impact of future rewards; It is the maximum Q value of all possible actions a' under the new state s'; (4c5) After updating the target Q value in (4c4), calculate the error between the output of the reinforcement learning model DQN in (4c2) and the target Q value. Adjust the weights of the model through backpropagation to obtain the optimal waveform selection strategy under different state information inputs and improve the effective throughput of the system.

4. The method according to claim 1, characterized in that: During the exchange process, the node performs a delay information update, which includes: (6a) Obtain the maximum and average latency of all node-coordinated data packets stored inside the node, which represent the longest transmission latency from the node to the farthest node and the overall communication efficiency between the node and other nodes, respectively. (6b) Iterate through the maximum latency of all nodes stored in the node, compare it with the maximum latency of the current node, check if there is a larger latency value, and update the maximum latency of the current node: If a larger latency value is found, the maximum latency of this node is updated to that larger value; If no larger delay value is found, no changes will be made; (6c) During the traversal in (6b), the maximum latency information of all nodes stored in the node is accumulated, the average latency is calculated by weighted averaging, and the average latency value of the current node is updated.

5. The method according to claim 1, characterized in that: In the distributed decision consensus phase, the selection of a unique central node through the collaborative data packet information stored in the distributed collaboration phase includes the following implementation: (8a) Find the node with the minimum average latency stored within the node: If the average latency of this node is the minimum and unique, then this node becomes the central node and the process ends. If the average latency of this node is not the minimum, terminate the process. If multiple nodes have the same average latency, execute (8b); (8b) Compare the maximum latency of each node stored within the node: If the maximum latency of this node is the smallest and unique, then this node becomes the central node and the process ends. If the maximum latency of this node is not the minimum, terminate the process. If multiple nodes have the same maximum latency, execute (8c); (8c) Compare the node addresses of each node stored within the node: If the node address of this node is the smallest, then this node becomes the central node and the process ends; Otherwise, end the process.

6. The method according to claim 1, characterized in that: The central node uses internally stored nodes... Waveform decision information is used to make the optimal waveform switching decision through majority voting, including: (8a) The central node traverses the waveform decision information of each node stored internally, calculates the frequency of each waveform being selected, and finds the waveform with the highest selection frequency as the candidate waveform. (8b) Determine the waveform switching decision based on the candidate waveform attributes in (8a): If there is only one candidate waveform, then that waveform will be used as the waveform switching decision result. If there are multiple candidate waveforms, the waveform with stronger anti-interference capability will be selected as the waveform switching decision result.

7. A system for implementing the method of claim 1, characterized in that, include: The node is configured with a state perception interaction module, which is used to periodically interact with the state packets and collaborative data packets of neighboring nodes. Specifically, it updates the storage of neighboring node state packets by comparing sequence numbers, collects the state information of the node at a period of 1 second and exchanges collaborative data packets with neighboring nodes at a period of 2 seconds. The node is configured with a cognitive processing module to build an efficient system for processing state packets and cooperative data packets. This system processes neighboring node state packets and cooperative data packets by removing the state packet header and cooperative packet header, and then allocates them to a designated shared buffer. The node is equipped with an anti-interference networking decision module, which takes the state information in the shared buffer as input and performs distributed waveform decision through the reinforcement learning model DQN to make the waveform switching strategy most suitable for the current network environment. The node is configured with a dynamic control module for controlling the election of the central node and the final waveform decision. Specifically, the central node is selected from the collaborative data packet information stored by the nodes in the distributed collaboration phase, and the central node makes the best waveform switching decision by majority vote in the distributed decision consensus phase.

8. The system according to claim 7, characterized in that, The node is equipped with a dynamic control module, including: The node setting center node election submodule is used to elect the node with the shortest communication latency as the center node. Specifically, it compares the average latency of each node with the collaborative data packets of all nodes in the network stored within the node, and prioritizes the node with the lowest average latency. If multiple nodes have the same average latency, it further compares the maximum latency of each node, and prioritizes the node with the lowest maximum latency. If multiple nodes have the same maximum latency, it selects the node with the smallest node address as the unique center node. The node sets the final waveform decision submodule, which is used to make the best waveform switching decision by majority vote. That is, after the central node is elected, the central node traverses the waveform decision information of each node stored internally, calculates the frequency of each waveform being selected, finds the waveform with the highest selection frequency as the candidate waveform, and if the candidate waveform is unique, it is selected as the final waveform decision; if it is not unique, the waveform with stronger anti-interference ability is selected as the final waveform decision.

Images