Real-time voice communication system and method based on movement prediction and standby path mechanism

Abstract

The invention discloses a real-time voice communication system and method based on a movement prediction and standby path mechanism, and aims to solve the problems of voice packet loss, delay and call interruption caused by frequent link interruption, large route reconstruction delay and high control overhead when an existing outdoor dynamic small-scale team performs real-time voice communication in a high-speed movement scene. The system comprises a voice sending end, a voice receiving end, a path management module and a mobile prediction module. The path management module constructs and maintains a main path tree for data transmission and a pre-generated standby path tree; the movement prediction module indirectly predicts the movement trend of the node through the change trend of the link quality, and constructs a movement prediction quality parameter through the change of the link packet loss rate, the delay and the jitter so as to dynamically optimize the structure of the path tree; when the main path is subjected to link breakage or the communication quality, such as packet loss rate, delay and jitter deterioration, is below threshold values, the system can be seamlessly switched to the standby path tree, so that the continuity and real-time performance of voice communication are guaranteed; according to the invention, the lightweight design is adopted, the control packet overhead is reduced, the method is suitable for an outdoor real-time voice communication scene with a small number of nodes, high moving speed and limited bandwidth and power consumption, and the communication robustness and the user experience are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of data communication technology, and in particular to a real-time voice communication system and method based on mobility prediction and backup path mechanism. Background Technology

[0002] A lightweight real-time communication system supporting outdoor mobility requires real-time monitoring of link quality to ensure the real-time performance of the entire communication system, reduce interruptions, and minimize bandwidth consumption. Existing technical solutions suffer from limitations such as single-path limitations leading to voice interruptions, excessive control packets causing severe bandwidth consumption, and neglecting the system's mobility attributes by only considering parameters like packet loss rate, resulting in link updates failing to adapt to node movement.

[0003] For example, the invention with application number CN202010482983.2 and publication number CN111385196A discloses a routing optimization and adjustment method for communication networks oriented towards real-time channel faults. This routing adjustment method aims to minimize the adjustment of traffic volume and can provide an adjustment scheme that meets the operational requirements of the communication network. However, this technical solution relies on adjustment after link breakage and cannot provide early warning of link quality degradation; it relies on a large amount of global link information and service information, resulting in significant bandwidth consumption; and it is designed for power communication networks, with large topologies and fixed nodes, making it unsuitable for outdoor dynamic small-scale mobile networks.

[0004] The invention disclosed in application number CN202411857800.5 and publication number CN119676148A presents a real-time communication system based on dynamic broadcast routing. This system achieves data communication through application layer and system driver, and is implemented through three stages: network construction, path allocation, and data maintenance. However, this technical solution uses fixed path planning, making it impossible to predict link degradation in advance based on mobility trends; it uses single-path transmission, and once the link breaks, it needs to be rediscovered, during which voice packets are lost, causing significant interruptions. In mobile scenarios, the single-path mechanism is prone to failure.

[0005] The invention with application number CN202511065760.5 and publication number CN120602052A discloses a dual-mode communication method and system based on link redundancy check and dynamic hybrid routing. It constructs a dual-link parallel channel using HPLC and HRF for data synchronization and consistency comparison. In case of link anomalies, it dynamically elects a proxy coordinating node and constructs a cross-protocol hybrid link. However, this technical solution requires support from two sets of equipment at the physical layer and the link layer, making it unsuitable for portable terminal scenarios. Triple retransmission of data packets can cause significant delays, which is unfriendly to real-time voice communication. Furthermore, the intelligent routing algorithm used in existing technologies requires training samples and offline models, increasing engineering complexity.

[0006] Due to the limitations mentioned above, existing dynamic routing schemes are not suitable for outdoor dynamic real-time voice communication systems and require optimization through specific mechanisms. Summary of the Invention

[0007] To overcome the problems existing in the prior art, the present invention aims to provide a real-time voice communication system and method based on mobility prediction and backup path mechanisms. The system employs mobility prediction to assess link quality; establishes backup paths in advance for rapid path switching; combines node mobility trends, packet loss rate, and the number of forwarding failures to assess link quality; and utilizes cross-layer collaboration between the application layer and network layer to ensure that link quality reflects multi-dimensional information simultaneously. The mobility prediction-based path mechanism ensures that nodes forward only within the allowed range of the path tree. Furthermore, the multi-dimensional link assessment mechanism enables the path tree to be updated in a timely manner and maintain topological validity, avoiding rebroadcasting or erroneous forwarding due to link mutations.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A real-time voice communication system based on mobility prediction and alternative path mechanisms includes: The voice transmitter is used to collect and encapsulate real-time voice data into voice data packets, and transmit them to the voice receiver through the main path tree. The voice receiver is used to receive voice data packets and decode and play them. The path management module is deployed on all nodes in the network. Each node maintains a global routing table to build and maintain the primary path tree and backup path tree. The primary path tree is used for data transmission of the current voice communication, while the backup path tree serves as the switching path when the primary path fails. The path management module uses three types of signals to jointly determine the link status: mobility prediction quality, path history information (packet loss rate, latency), and continuous failure detection. These signals are divided into a mobility prediction layer, a link information layer, and a continuous failure detection layer. When a deterioration in link quality is detected, the path management module preheats the backup path. When a link interruption is detected in the primary path tree or the link quality deteriorates below a preset threshold, the path management module immediately activates the backup path tree to continue transmitting voice data packets, ensuring the continuity and real-time performance of voice communication. The mobility prediction module indirectly predicts node mobility trends by forecasting changes in link quality. It performs smoothing trend analysis on the real-time quality sequences of candidate next-hop nodes and dynamically updates the path tree based on the prediction results.

[0009] Both the primary path tree and the backup path tree are multi-hop tree structures. Each node contains a node address, a next-hop target address, and child node forwarding set information. The system supports multiple voice receivers receiving data from the same voice sender simultaneously. In this case, both the primary path tree and the backup path tree are constructed into tree topologies that support one-to-many broadcasting or multicasting.

[0010] The mobility prediction module indirectly predicts mobility trends based on time series data of link quality.

[0011] When constructing the backup path tree, the path management module prioritizes the pre-move prediction function score, the positive correlation between link stability and delay jitter, and selects relay nodes to form the path.

[0012] The communication quality degradation is determined based on at least one of the following: mobile prediction quality parameters, packet loss rate, latency, or number of consecutive transmission failures exceeding a preset threshold.

[0013] The voice data packets are encapsulated using a real-time transmission protocol, which supports data packet sequence number marking, timestamp annotation, and packet loss detection.

[0014] After the main path is restored, the path management module decides whether to switch back to the main path tree according to a preset strategy. The strategy includes: continuing to use the backup path until the session ends, or automatically switching back to the main path after the main path has been running stably for more than a preset time.

[0015] The relay nodes are uniquely identified based on Media Access Control addresses and Internet Protocol addresses, and dynamic routing and forwarding of paths are implemented at the network layer.

[0016] The prediction of the mobility trend uses link information such as packet loss rate, latency, and jitter, combined with smoothing trend analysis and linear extrapolation to construct a scoring function to predict the future mobility trend of the device; specifically, it uses the most recent N RTP data packets to form a sliding statistical window, and calculates the packet loss rate in the t-th window. Average latency Shaking Normalize the three indicators to the interval [0,1] to obtain the normalized values. , and The normalization upper limit is set according to the system tolerance threshold; the instantaneous link quality is defined as: ,in, , , For configurable weights, , , The sum of the parameters is 1 and , , All values ​​are ≥0; the current smoothing quality uses an exponential moving average. The trend of quality change is expressed using the first-order difference. Future quality prediction uses linear extrapolation. Utilizing current quality Future quality forecast and trends The three factors are used to construct a comprehensive scoring function: in, a,b,c These are non-negative weighting coefficients, and a + b + c = 1; The higher the score, the more stable the link is in the near future, and the nodes are relatively stationary or moving in a favorable direction; conversely, if the score continues to decline, it indicates that the nodes may be moving away from the communication range and the link is about to deteriorate; this score is used for next hop selection in the path planning phase and link health warning in the operation phase.

[0017] A real-time voice communication method based on mobility prediction and alternative path mechanisms includes the following steps: Step S1: Establish a path tree and begin transmitting voice data; each node maintains a global routing table, which contains a list of directly connected nodes for each node; before a node begins transmitting voice data, path planning is performed based on the global routing table; a depth-first search is initiated from the source node, and candidate next hops are prioritized for expansion in descending order of mobility prediction score; during the expansion phase, a global set of covered targets is maintained as a termination condition; for each target, the path with the highest score is selected as the primary path; the path with the second highest score that is not the same as the primary path is recorded as the backup path for that target; for each relay node, its subtree information is aggregated to form the primary path next hop information and the backup path next hop information, ultimately forming a tree structure; Step S2: Collect link quality information of terminal devices in real time; through the queue of RTP data packets, statistically analyze packet loss rate, latency and jitter information at the network layer, so that the mobility prediction module and path management module can control and modify the link information. Step S3: Based on the statistical information from Step S2, perform mobility prediction and calculate the link quality function statistical score; indirectly predict the node's mobility trend by predicting the link quality change trend; perform smoothing trend analysis on the real-time quality sequence of candidate next-hop nodes; use the most recent N RTP packets as a statistical window, and calculate the packet loss rate in the t-th window. Average latency Shaking Normalize latency, packet loss, and jitter into , and Set an upper limit, and then combine the three items into an instantaneous mass. ,in, , , The sum of the parameters is 1; the current smoothing quality uses an exponential moving average. The trend of quality change is expressed using the first-order difference. Future quality prediction uses linear extrapolation. Utilizing current quality Future quality forecast and trends The three factors are used to construct a comprehensive scoring function: The motion prediction score is used as a quality parameter for the motion prediction of nodes. The next hop node with a high score is selected to avoid nodes that are about to move out of the range in advance; thus, the motion trend based on the nodes is obtained. Step S4: During RTP data transmission, monitor the communication quality of the main path. If an interruption or degradation occurs, seamlessly switch to the backup path tree to continue transmission. The path management module evaluates the link quality based on mobility prediction score, link quality statistics, and continuous failure detection. During the data packet transmission phase, the mobility prediction module determines the link status based on the node-based mobility trend obtained in Step 3. When the mobility prediction score is lower than a set threshold, it is determined that the link is about to break. The latency and packet loss rate of M consecutive data packets are used for subsequent judgment. When N consecutive data packet transmission failures occur, it is determined that the link has failed. The mobility prediction score, packet loss rate, and latency are complementary and form a three-level triggering mechanism according to their severity. Under normal conditions, none of the three mechanisms have reached the threshold, i.e., the mobility prediction score is higher than the set threshold, the packet loss rate and latency are lower than the set threshold, and there are no N consecutive data packet transmission failures, the main path is used. When the mobility prediction score continues to decrease and approaches the boundary value, and the packet loss rate continues to increase but has not reached the critical point, prepare for switching. When the mobility prediction score reaches the near value and both packet loss and latency exceed the set value, a forced switch is performed. When there are more than N consecutive transmission failures, a forced switch to the backup path is performed. Step S5: After the main path is restored, according to the predefined switching thresholds of mobility prediction score, packet loss rate and latency indicators, switching back is allowed when the link quality mobility prediction score, packet loss rate and latency indicators exceed the set thresholds and meet K statistical windows consecutively; during switching back, data is sent in parallel first, and the main path is switched back and the backup path is stopped after the link quality mobility prediction score, packet loss rate and latency indicators stabilize.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention uses mobility prediction during the routing planning stage to predict based on latency, packet loss, and jitter information, and selects a dual-path tree consisting of a primary path and a backup path. Combined with a three-layer evaluation mechanism of mobility prediction, link statistics, and continuous failure detection, it achieves path switching before link decay during dynamic movement, reduces voice playback interruptions, and realizes low-latency, uninterrupted path switching.

[0019] This invention performs mobility prediction based on link quality time series, calculating the short-term link quality change trend of nodes based on packet loss rate, latency, and jitter. The mobility prediction model of this invention is used in both the routing planning and dynamic link quality assessment stages. In the routing planning stage, it serves as a condition for selecting primary and backup nodes; in the link assessment stage, it combines historical link statistics and the number of consecutive failures to provide early warnings, significantly improving communication continuity and greatly reducing the probability of path breakage.

[0020] This invention employs a three-layer link health assessment mechanism, which collaborates across layers and considers multi-dimensional information to evaluate link quality, thereby improving the accuracy of link judgment. The three-layer triggering mechanism is set to be progressive, with multi-dimensional joint judgment, effectively avoiding frequent switching caused by single-parameter measurement.

[0021] In this invention, the RTP data stream itself can serve as information for link quality assessment. By relying on packet statistics, mobility prediction, and application layer feedback, the control overhead of link maintenance is reduced, and the communication load is reduced. This invention is suitable for lightweight, bandwidth-constrained outdoor team real-time communication scenarios.

[0022] In summary, this invention, through mechanisms such as motion prediction, cross-layer collaboration, and backup paths, can detect changes in link topology in advance and complete path switching before link quality anomalies occur, thereby reducing the overall packet loss rate and communication interruption and improving the real-time performance of the voice communication system; it also reduces redundant forwarding caused by replanning paths and lowers bandwidth consumption. Attached Figure Description

[0023] Figure 1 This is the overall system architecture diagram of the present invention.

[0024] Figure 2 This is the path planning flowchart of the present invention.

[0025] Figure 3 This is a data forwarding flowchart of the present invention.

[0026] Figure 4 This is a flowchart of the three-layer link quality detection process of the present invention.

[0027] Figure 5 This is a flowchart of the path switching process of the present invention.

[0028] Figure 6This is an example diagram of the node topology structure of an embodiment of the present invention. Figure 7 This is a statistical chart of data packet delivery rate in the simulation experiment of this invention.

[0029] Figure 8 This is a simulation experiment delay statistics chart of the present invention. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0031] Reference Figure 1 A real-time voice communication system based on mobility prediction and alternative path mechanisms, the overall architecture of which includes: The voice transmitter is used to collect and encapsulate real-time voice data into voice data packets, and transmit them to the voice receiver through the main path tree. The voice receiver is used to receive voice data packets and decode and play them. The mobility prediction module indirectly predicts node mobility trends by forecasting changes in link quality. It performs smoothing trend analysis on the real-time quality sequences of candidate next-hop nodes and dynamically updates the path tree based on the prediction results. The path management module is deployed on all nodes in the network. Each node maintains a global routing table to build and maintain the primary path tree and backup path tree. The primary path tree is used for data transmission in the current voice communication, while the backup path tree serves as the switching path when the primary path fails. The path management module uses three types of signals to jointly determine the link status: the mobility prediction layer, the link information layer, and the continuous failure detection layer. When a deterioration in link quality is detected, the path management module pre-warms up the backup path. When a link interruption is detected in the primary path tree or the link quality deteriorates below a preset threshold, the path management module immediately activates the backup path tree to continue transmitting voice data packets, ensuring the continuity and real-time performance of voice communication. The mobility prediction module is responsible for predicting mobility trends based on packet loss rate, latency, and jitter. The link statistics module monitors packet loss and latency of voice data packets in real time at the network layer. The path planning module simultaneously builds the primary path tree and backup path tree based on information from the first two modules. The link quality assessment module is responsible for determining the link status based on comprehensive parameters to facilitate subsequent route switching.

[0032] During the network establishment phase, the network needs to build a complete routing table. First, the nodes send probe packets carrying their own MAC address and IP address. The nodes that receive the probe packets return the complete path table of the network to the newly connected nodes. Next, the connection information of the two nodes is synchronized with other nodes in the network. Finally, all nodes maintain the complete path table of the entire network, which contains information on all single-hop direct data paths in the network.

[0033] Reference Figure 2 During the path planning phase, the root node obtains a list of destination nodes from which data needs to be sent. By parsing the global routing table and combining link information, mobility prediction is performed to select an optimal node and a second-best node, which are then used to expand to the next layer. If a direct path to a node is found for the first time, it is added to the main routing tree; if it is found a second time, it is added to the backup path tree. Finally, the path tree is serialized and distributed to all nodes, and each node determines whether to forward the data based on its role in the path tree. Mobility prediction uses packet loss rate, latency, and jitter information over a period of time to construct a comprehensive link quality scoring function. For each path, a candidate score is calculated, and the order of the primary and alternative paths is determined based on the score. A higher score indicates a more stable path and greater suitability for real-time voice transmission. Paths are sorted from highest to lowest score. The path with the highest score is selected as the primary path (for normal data forwarding). The path with the second-highest score is selected as the backup path (for normal monitoring, and to take over when the main path is abnormal).

[0034] Reference Figure 3 After receiving the path tree, each node determines its role in the forwarding task based on the roles recorded in the path tree. Roles include primary path relay nodes, primary path edge nodes, backup path nodes, and unrelated nodes. For primary path relay nodes, the system forwards data packets according to the predefined next-hop information in the path tree, enabling multi-hop real-time voice transmission. For primary path edge nodes, the system only receives and plays audio data, without forwarding. Backup path nodes do not participate in data relay; they only passively monitor data packets and link status, seamlessly taking over forwarding tasks when link quality deteriorates beyond a critical threshold. Unrelated nodes discard received data packets to reduce unnecessary broadcast overhead. The entire forwarding process does not rely on Hello packets or link probe packets in traditional ad hoc network protocols, but instead achieves forwarding control entirely through the path tree structure, significantly reducing communication load and improving real-time performance.

[0035] Reference Figure 4In the mobility prediction module, the system first performs continuous failure detection. If the number of consecutive failures exceeds a set threshold N, for example, 10, it is judged as a link break. If no failure is detected at this stage, the next step is to perform statistics, including mobility prediction and link history information statistics. Mobility prediction establishes a function based on link information, which includes the current link quality, future quality, and quality change trend, constructing a comprehensive link score. Link history information includes the packet loss rate and latency of the most recent M data packets. When the prediction score is lower than the set threshold, and there are many packet losses, but none have reached the boundary value, the link status is set to WARN, and the backup path is preheated. When the prediction score exceeds the threshold and both the packet loss rate and latency exceed the threshold, the link status is set to FAIL, and a forced switch to the backup path is performed. If no transmission failure occurs, and the mobility prediction and link statistics are within the normal range, the link status is set to NORMAL, and data forwarding is performed according to the main path tree.

[0036] Reference Figure 5 Path switching first requires obtaining the link evaluation result. The necessary actions are determined based on the current link status (NORMAL, WARN, FAIL). If the link status is normal, the primary path continues to be used for forwarding. If the link status is abnormal but not broken, the backup path is pre-warmed, including checking the backup path link status, setting the backup path flag `isBcakupValid` to true, and adding the backup path forwarding table at the network layer. If the link status is FAIL, link switching immediately occurs, swapping the roles of the primary and backup paths, updating the network layer next hop, and the network layer automatically uses the backup path for forwarding, achieving seamless switching.

[0037] A real-time voice communication method based on mobility prediction and alternative path mechanisms includes the following steps: Step S1: Establish a path tree and begin transmitting voice data; each node maintains a global routing table, which contains a list of directly connected nodes for each node; before a node begins transmitting voice data, path planning is performed based on the global routing table; a depth-first search is initiated from the source node, and candidate next hops are prioritized for expansion in descending order of mobility prediction score; during the expansion phase, a global set of covered targets is maintained as a termination condition; for each target, the path with the highest score is selected as the primary path; the path with the second highest score that is not the same as the primary path is recorded as the backup path for that target; for each relay node, its subtree information is aggregated to form the primary path next hop information and the backup path next hop information, ultimately forming a tree structure; Step S2: Collect link quality information of terminal devices in real time; through the queue of RTP data packets, statistically analyze packet loss rate, latency and jitter information at the network layer, so that the mobility prediction module and path management module can control and modify the link information. Step S3: Perform mobility prediction based on statistical information and calculate the link quality function statistical score; indirectly predict the node's mobility trend by predicting the link quality change trend; perform smoothing trend analysis on the real-time quality sequence of candidate next-hop nodes; use the most recent N RTP packets as a statistical window, and calculate the packet loss rate in the t-th window. Average latency Shaking Normalize latency, packet loss, and jitter into , and Set an acceptable upper limit, such as the maximum acceptable packet loss rate. Normalize it to get Then combine the three items into an instantaneous quality. ,in, , , The sum of the parameters is 1; specific values ​​can be optimized based on specific experiments; the current smoothing quality uses an exponential moving average. The trend of quality change is expressed using the first-order difference. Future quality prediction uses linear extrapolation. Utilizing current quality Future quality forecast and trends The three factors are used to construct a comprehensive scoring function: The motion prediction score is used as a quality parameter for the motion prediction of nodes, selecting more stable next-hop nodes and avoiding nodes that are about to move out of the range in advance. Step S4: During RTP data transmission, monitor the communication quality of the main path. If an interruption or degradation occurs, seamlessly switch to the backup path tree to continue transmission. The path management module evaluates the link quality based on mobility prediction score, link quality statistics, and continuous failure detection. During the data packet transmission phase, the mobility prediction module determines the link status based on the node-based mobility trend obtained in Step 3. When the mobility prediction score is lower than a set threshold, it is determined that the link is about to break. The latency and packet loss rate of M consecutive data packets are used for subsequent judgment. When N consecutive data packet transmission failures occur, it is determined that the link has failed. The mobility prediction score, packet loss rate, and latency are complementary and form a three-level triggering mechanism according to their severity. Under normal conditions, none of the three mechanisms have reached the threshold, i.e., the mobility prediction score is higher than the set threshold, the packet loss rate and latency are lower than the set threshold, and there are no N consecutive data packet transmission failures, the main path is used. When the mobility prediction score continues to decrease and approaches the boundary value, and the packet loss rate continues to increase but has not reached the critical point, prepare for switching. When the mobility prediction score reaches the near value and both packet loss and latency exceed the set value, a forced switch is performed. When there are more than N consecutive transmission failures, a forced switch to the backup path is performed. Step S5: After the main path is restored, decide whether to switch back according to the strategy; define the switchback threshold, and allow switchback when the link quality exceeds the set threshold and meets K statistical windows consecutively; during switchback, send in parallel for a period of time first, and switch back to the main path and stop using the backup path after the indicators stabilize.

[0038] In this invention, mobility prediction is used not only in the path planning stage, but also in the dynamic link quality assessment stage. Through the mobility prediction model, the network can detect the deterioration trend of the link in advance, complete the handover in advance, and maintain the relative stability of voice transmission.

[0039] A dual-path routing system with a primary path and a backup path is adopted. During the path planning phase, the system generates a primary path tree and a backup path tree for each source node. The primary path tree is used for normal forwarding, while the backup path is used to warm up when the primary path shows signs of deterioration and to activate when the primary path fails. The tree structure determines which nodes participate in intermediate forwarding, reducing unnecessary broadcasts. The dual-path design enables the voice communication system to perform uninterrupted switching and achieve stable voice transmission.

[0040] This invention uses three types of signals to jointly determine the link status. The motion prediction layer, based on the node's movement trend, determines that the link is about to break when the motion prediction quality score falls below a set threshold. The delay and packet loss rate of M consecutive data packets are used for subsequent judgment. The continuous failure detection layer determines that the link has failed when N consecutive data packet transmission failures occur. These three mechanisms complement each other, forming a three-level triggering mechanism according to severity. In normal conditions, the main path is used. When the prediction quality approaches the boundary value and packet loss is high but has not reached the critical value, the backup path is preheated in preparation for switching. When the motion prediction reaches an imminent value and both packet loss and delay exceed the set threshold, a forced switch is triggered. A forced switch is triggered after more than N consecutive transmission failures.

[0041] A cross-layer collaborative routing mechanism between the application and network layers. The application layer is responsible for the node mobility prediction model, while the network layer is responsible for link evaluation and path switching. When a link anomaly is detected, the network layer automatically switches to a backup path. This cross-layer collaboration enables early switching before the link breaks, so the upper-layer voice transmission module does not need to interrupt or rebuild the link.

[0042] Example This embodiment uses a multi-hop voice communication system as an example. In this embodiment, the terminal nodes perform multi-hop voice communication through a self-organizing network, and the network topology is as follows: Figure 6 Example topology diagram shown.

[0043] Node A is the voice data transmitter, nodes B and C are A's direct neighbor nodes, nodes D and E are next-hop relay nodes, and node F is the voice receiving edge node.

[0044] Each node maintains a global routing table, recording its directly connected neighbor list and reachability information. At the start of communication, the system first constructs a path tree using the global routing information from the source node. When source node A prepares to send voice data, it first plans a path based on the global routing table. During the search process, mobility prediction scoring is introduced, calculating the mobility prediction scores for next-hop nodes B and C. The system prioritizes expanding to next-hop nodes with higher scores. Assuming a higher link score in the direction of node B, the system constructs the path A->B->D->F as the primary path tree and A->C->E->F as the backup path tree. Ultimately, a primary path tree and a backup path tree are generated for each source node, forming a dual-path routing system. The primary path tree is used for normal forwarding, while the backup path is used to warm up when the primary path shows signs of deterioration. Nodes C and E are in a standby listening state, initiating forwarding when the primary path fails. The tree structure determines which nodes participate in intermediate forwarding, reducing unnecessary broadcasts. This dual-path design allows for uninterrupted switching in the voice communication system, achieving stable voice transmission.

[0045] During voice transmission, the network layer collects link information in real time and feeds the results back to the mobility prediction module and routing control module. Each node obtains packet loss rate, latency, and jitter metrics by statistically analyzing the received RTP packets. The system maintains an RTP queue window for each voice data stream to record the sequence numbers and arrival times of the most recent N RTP packets, and calculates relevant information based on the packet sequence numbers and inter-packet latency differences. In this process, the application layer and network layer collaborate across layers. The application layer is responsible for the node's mobility prediction model, while the network layer is responsible for link evaluation and path switching. When a link anomaly is detected, the network layer automatically switches to a backup path. This cross-layer collaboration enables early switching before the link breaks, eliminating the need for the upper-layer voice transmission module to interrupt or rebuild the link.

[0046] Upon receiving a data packet, a comprehensive score is calculated using a mobility prediction model to assess link quality. In this invention, mobility prediction is used not only in the path planning phase but also in the dynamic link quality assessment phase. Through the mobility prediction model, the network can detect link deterioration trends in advance, complete handover early, and maintain relatively stable voice transmission.

[0047] During data transmission, based on link information statistics, three types of signals jointly determine the link status. The motion prediction layer, based on the node's movement trend, determines that the link is about to break when the motion prediction quality score is lower than a set threshold; the latency and packet loss rate of M consecutively detected data packets are used for subsequent judgment; the continuous failure detection layer determines that the link has failed when N consecutive data packets fail to be sent.

[0048] During path switching, the three mechanisms complement each other, forming a three-level triggering mechanism according to the severity. In normal conditions, the main path is used. For example, in the initial stage, the main path A->B->D->F has good link quality, and RTP voice packets are forwarded along the main path. As the network operates and nodes move, when node B's link statistics module detects that the packet loss rate and latency are constantly increasing, and the mobility prediction score is close to the boundary value, it enters an early warning state, warms up the backup path in advance, and prepares for switching. When the mobility prediction score reaches the near value and the packet loss and latency exceed the set value, a forced switch to the backup path A->C->E->F is performed. If there are more than N consecutive transmission failures, a forced switch will also be performed.

[0049] After the primary path is restored, a switchback is allowed once the link quality exceeds a set threshold and meets K statistical window requirements consecutively. During the switchback, parallel transmission is performed for a period of time, and the primary path is switched back and the backup path is discontinued once the metrics stabilize.

[0050] Simulation Experiment To evaluate the performance of this invention, simulation verification was performed: Simulation verification was performed using the NS3 network simulation platform. A multi-node self-organizing network simulation environment was constructed, in which the links adopted the IEEE 802.11 standard Ad-hoc mode, and the maximum communication radius was set to 200m; the routing protocol was a custom real-time communication protocol of this invention, an improvement on the AODV protocol. To simulate the application effect of this invention in an outdoor multi-node environment, this embodiment uses the RandomWaypointMobilityModel built into ns-3 as the node mobility model. This model can simulate the random movement behavior of terminals in outdoor scenarios, including random destination selection, random speed changes, and pause times.

[0051] The simulation experiment simulated a self-organizing network with 9 nodes. To examine the performance of the invention under different network densities, connectivity, and mobility conditions, four sets of movement ranges were selected for comparative experiments. The node speed was set to 1-5 m / s to simulate walking speed, and the four movement ranges were set as square ranges with side lengths of 50m, 100m, 150m, and 200m, respectively. The simulation time was set to 100s.

[0052] To comprehensively evaluate the protocol performance of this invention, this embodiment selects the following core indicators for measurement: end-to-end latency and data packet delivery rate. (See also...) Figure 7 The packet delivery rate refers to the ratio of the number of received packets to the total number of sent packets, while the end-to-end latency refers to the entire time interval from when a packet is generated by the sending node (or sent from the network layer) to when it arrives at the receiving node and is successfully received by the upper layer.

[0053] like Figure 8 As shown in the experimental results, with the increase in the node's movement range, the network topology becomes more dynamic, and the probability of link failure increases, resulting in an increasing trend in both packet loss rate and end-to-end latency. However, the improved routing mechanism proposed in this invention still maintains an overall packet delivery rate of over 90% and an end-to-end latency of less than 2 ms even in high-mobility scenarios, such as a maximum movement range of 200 m × 200 m.

[0054] The above results demonstrate that the present invention can effectively reduce the negative impact of topology changes in complex outdoor mobile scenarios, possesses high real-time performance and stability, and can meet the stringent requirements of voice communication for continuity and low latency. The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A real-time voice communication system based on a mobile prediction and backup path mechanism, characterized by, include: The voice transmitter is used to collect and encapsulate real-time voice data into voice data packets, and transmit them to the voice receiver through the main path tree. The voice receiver is used to receive voice data packets and decode and play them. The path management module is deployed on all nodes in the network. Each node maintains a global routing table to build and maintain the primary path tree and backup path tree. The primary path tree is used for data transmission of the current voice communication, while the backup path tree serves as the switching path when the primary path fails. The path management module uses three types of signals to jointly determine the link status: mobility prediction quality, path history information (packet loss rate, latency), and continuous failure detection. These signals are divided into a mobility prediction layer, a link information layer, and a continuous failure detection layer. When a deterioration in link quality is detected, the path management module preheats the backup path. When a link interruption is detected in the primary path tree or the link quality deteriorates below a preset threshold, the path management module immediately activates the backup path tree to continue transmitting voice data packets, ensuring the continuity and real-time performance of voice communication. The mobility prediction module indirectly predicts the mobility trend of nodes by predicting the changing trend of link quality, performs smoothing trend analysis on the real-time quality sequence of candidate next-hop nodes, and dynamically updates the path tree based on the prediction results.

2. The real-time voice communication system based on the mobile prediction and standby path mechanism according to claim 1, characterized in that, Both the primary path tree and the backup path tree are multi-hop tree structures, with each node containing a node address, a next-hop target address, and child node forwarding set information; both the primary path tree and the backup path tree are constructed as tree topologies that support one-to-many broadcast or multicast.

3. The real-time voice communication system based on the mobile prediction and standby path mechanism according to claim 1, characterized in that, The mobility prediction module indirectly predicts mobility trends based on time series data of link quality.

4. The real-time voice communication system based on the mobile prediction and standby path mechanism according to claim 1, characterized in that, When constructing the backup path tree, the path management module prioritizes the pre-move prediction function score, the positive correlation between link stability and delay jitter, and selects relay nodes to form the path.

5. The real-time voice communication system based on mobile prediction and standby path mechanism according to claim 1, characterized in that, The communication quality degradation is determined based on at least one of the following: mobile prediction quality parameters, packet loss rate, latency, or number of consecutive transmission failures exceeding a preset threshold.

6. The real-time voice communication system based on mobile prediction and standby path mechanism according to claim 1, characterized in that, The voice data packets are encapsulated using a real-time transmission protocol, which supports data packet sequence number marking, timestamp annotation, and packet loss detection.

7. The real-time voice communication system based on mobile prediction and standby path mechanism according to claim 1, characterized in that, After the main path is restored, the path management module decides whether to switch back to the main path tree according to a preset strategy. The strategy includes: continuing to use the backup path until the session ends, or automatically switching back to the main path after the main path has been running stably for more than a preset time.

8. The real-time voice communication system based on mobile prediction and standby path mechanism according to claim 1, characterized in that, The relay nodes are uniquely identified based on Media Access Control addresses and Internet Protocol addresses, and dynamic routing and forwarding of paths are implemented at the network layer.

9. The real-time voice communication system based on mobile prediction and standby path mechanism according to claim 3, characterized in that, The prediction of the movement trend uses link information such as packet loss rate, latency, and jitter, combined with smoothing trend analysis and linear extrapolation to construct a scoring function to predict the future movement trend of the device. Specifically: with the last N RTP packets to form a sliding statistical window, in the tth window to count the packet loss rate , average delay , jitter ; The three indicators are normalized to the interval [0, 1] to obtain normalized values , and , where the upper limit of normalization is set according to the system tolerance threshold; define the instantaneous link quality as: , where, , , is a configurable weight, , , the sum of the parameters is 1 and , , all ≥ 0; the current smoothed quality uses an exponential moving average , the quality change trend uses a first-order difference , and the future quality prediction uses linear extrapolation Utilize the current quality , the future quality prediction , and the change trend to construct a comprehensive score function: wherein, a,b,c are non-negative weight coefficients, and a + b + c = 1; the higher the score, the more stable the link is in the near future, the node is relatively stationary or moving in a favorable direction; on the contrary, if the score continues to decline, it indicates that the node may be moving away from the communication range, and the link will soon deteriorate; This score is used for next-hop selection during the path planning phase and for link health warnings during the operation phase.

10. A communication method for the real-time voice communication system according to any one of claims 1 to 8, characterized by, Includes the following steps: Step S1: Establishing a path tree and starting to transmit voice data; each node maintains a global routing table containing a list of directly connected nodes of each node; before the node starts to send voice data, path planning is performed according to the global routing table; a depth-first search is initiated from a source node, and candidate next hops are preferentially expanded in descending order of mobile prediction score; in the expansion stage, a global covered target set is maintained as a termination condition; For each target, the path with the highest score is preferred as the main path; The path with the second highest score and different from the main path is recorded as the backup path of the target; For each relay node, the information of its sub-tree is summarized to form next-hop information of the main path and next-hop information of the backup path, and finally a tree structure is formed; Step S2: Collecting link quality information of terminal equipment in real time; Through the queue of RTP data packets, packet loss rate, delay and jitter information are counted at the network layer; Step S3: Based on the statistical information of step S2, the link quality function statistical score is calculated by predicting the change trend of the link quality, and the mobile trend of the node is indirectly predicted; Smooth trend analysis on real-time quality sequence of candidate next-hop node; take the latest N RTP packets as a statistical window, and count the packet loss rate in the tth window , average delay , jitter ; delay, packet loss, and jitter are normalized as , and , set an upper limit, and then the three items are synthesized into an instant quality , wherein, , , the sum of the parameters is 1; the current smooth quality uses exponential moving average , the quality change trend adopts first-order difference , and the future quality prediction adopts linear extrapolation The current quality , the future quality prediction and the change trend are used to construct a comprehensive score function: The mobile prediction score value is used as the mobile prediction quality parameter of the node, and the next hop node with a high score is selected to avoid nodes that will soon move out of the range in advance; The node-based mobile trend is obtained; Step S4: In the process of RTP data transmission, the communication quality of the main path is monitored, and if interruption or degradation occurs, seamless switching to the backup path tree is performed for continuous transmission, and the link quality is evaluated by the path management module according to the mobile prediction score, link quality statistics and continuous failure detection; In the data packet sending stage, the mobile prediction module judges the link state according to the node-based mobile trend obtained in step 3; when the mobile prediction score is lower than the set threshold, it is judged that the link will soon be broken; The delay and packet loss rate of M data packets are used for subsequent judgment; when N consecutive data packet sending failures occur, it is judged that the link has failed; the three mechanisms of mobile prediction score, packet loss rate and delay are complementary, and a three-level triggering mechanism is formed according to the severity; when the mobile prediction score is higher than the set threshold, the packet loss rate and the delay are less than the set threshold, and there is no continuous N data packet sending failure, the main path is used; when the mobile prediction score continuously decreases and approaches the boundary value, the packet loss rate continuously increases but does not reach the critical value, the switching is prepared; when the mobile prediction score reaches the critical value and the packet loss and delay exceed the set value, the switching is forced; when the continuous sending failure is more than N times, the switching is forced to the backup path; Step S5: After the main path is restored, the switching threshold of the pre-defined mobile prediction score, packet loss rate and delay index is allowed to switch back when the link quality mobile prediction score, packet loss rate and delay index exceed the set threshold and continuously meet K statistical windows; when switching back, data is sent in parallel first, and after the link quality mobile prediction score, packet loss rate and delay index are stable, the main path is switched back and the backup path is stopped.

Images