Dynamic networking communication method and system for heterogeneous intercoms based on virtual channel pool

Abstract

The application discloses a kind of heterogeneous intercom dynamic group networking communication method and system based on virtual channel pool, comprising: deploying distributed soft switching agent in intercom terminal, analog port, digital time slot is abstracted as virtual channel instance, and global virtual channel pool mapping table is constructed;Capture unstructured networking intention signal and report, cannot directly address when planning relay path, merge same proxy same system terminal to generate demand unit;Based on demand unit, candidate channel is screened, and comprehensive cost matrix is constructed, and improved hungarian algorithm is introduced to solve global optimal allocation scheme;According to scheme, subtask is issued, and each proxy is executed by system protocol state machine to parse into original signaling sequence, and terminal is attached to service channel to complete cross-system group call cut-in;Group call stage continuously detects and adapts coding and decoding format consistency, and dynamic recovery and media resource release are carried out after call ends.Heterogeneous intercom and on-demand scheduling are realized, and cross-system channel utilization and dynamic group networking flexibility are improved.

Description

Technical Field

[0001] This invention relates to the field of walkie-talkie intelligent communication technology, and in particular to a method and system for dynamic networking communication of heterogeneous walkie-talkies based on a virtual channel pool. Background Technology

[0002] In scenarios involving public safety, emergency command, and multi-departmental joint operations, different organizational units are typically equipped with walkie-talkies of different standards. Analog walkie-talkies, DMR or PDT digital walkie-talkies, and public network walkie-talkie systems based on LTE or 5G often coexist in the same location. To achieve cross-departmental collaborative communication, these heterogeneous terminals need to be temporarily grouped into a unified call group. Currently, the main way to achieve interoperability between heterogeneous walkie-talkie systems is to deploy protocol conversion gateways at the boundaries of each system. Through manual pre-configuration, specific analog voice ports are statically bound to specific digital time slots, forming a back-to-back hard bridge channel.

[0003] This approach has the following prominent problems. First, the networking relationship relies entirely on pre-planning. Each cross-system communication requires administrators to pre-define port mapping tables and call routing rules in the background, making it impossible for frontline commanders to initiate temporary networking in real time based on rapidly changing situations. This makes it difficult to meet the timeliness requirements of second-level dynamic grouping in emergency scenarios. Second, static binding means that regardless of whether there is an actual call demand, the switching channel and gateway port are continuously monopolized, resulting in serious waste of channel resources in scenarios with multiple groups rotating communications. The number of concurrent cross-system call groups that the system can support is limited by the number of physical ports, leading to a high call drop rate. Furthermore, fixed bridging paths cannot dynamically select routes based on factors such as real-time network transmission latency, gateway processing load, and voice quality attenuation caused by multi-level codec conversion. End-to-end voice quality is difficult to guarantee, especially when a call involves multiple inter-system cascading transfers, where accumulated codec distortion will lead to a severe decrease in voice clarity.

[0004] Furthermore, existing solutions generally lack precise synchronous control for establishing cross-system group calls. Differences in signaling interaction latency between different systems cause some terminals to experience voice loss at the start of a call, affecting the complete transmission of command instructions. Therefore, there is an urgent need for a dynamic networking communication method that can uniformly pool and schedule heterogeneous channel resources, intelligently allocate optimal transfer paths on demand based on on-site intentions, and ensure synchronous entry of cross-system terminals and reliable resource recovery. Summary of the Invention

[0005] This invention overcomes the shortcomings of the prior art and provides a method and system for dynamic networking communication of heterogeneous walkie-talkies based on virtual channel pools. Its important purpose is to realize the unified pooling and on-demand scheduling of heterogeneous walkie-talkie channels, thereby improving the cross-system channel utilization and dynamic networking flexibility.

[0006] To achieve the above objectives, the first aspect of this invention provides a method for dynamic networking communication of heterogeneous walkie-talkies based on a virtual channel pool, comprising: S102, deploy a distributed softswitch agent on the intercom terminal and collect the standard, capacity and real-time status of the physical channel, abstract the analog port into a virtual channel object with a capacity of one, abstract the digital time division multiple access time slot into a virtual channel instance, build a global resource pool and generate a global virtual channel pool mapping table. S104, when an unstructured networking intent signal is captured, it is converted into a networking intent description according to a predefined mapping rule and reported. For those that cannot be directly radio frequency addressed, a relay path is planned through breadth-first search. Multiple terminals with the same home agent and the same target standard are merged to generate a set of virtual channel demand units carrying capacity requirements. S106, Based on the set of virtual channel demand units, candidate virtual channels are selected from the global virtual channel pool mapping table and a multi-dimensional dynamic comprehensive cost matrix is ​​constructed. An improved Hungarian algorithm based on capacity awareness is introduced, and the global optimal channel allocation scheme is solved in combination with the multi-dimensional dynamic comprehensive cost matrix to finally obtain the virtual channel allocation scheme. S108, generate and distribute the subtask descriptions of each agent according to the virtual channel allocation scheme. Each agent parses the subtasks into native signaling sequences adapted to the local intercom system through the built-in standard protocol state machine and executes them. After the terminal is attached to the service channel, report the ready message to the corresponding agent to perform cross-standard terminal group call switching. S110 continuously detects the consistency between the encoding / decoding format of the voice frames output from the source virtual channel and the target virtual channel during the group call hold phase and performs media adaptive adaptation. Simultaneously, after the call ends, it triggers dynamic reclamation of the virtual channel based on silence detection or disconnection command, and uses the lease timeout mechanism to manage the release of media resources.

[0007] In this solution, S102 specifically includes: Deploy distributed softswitch agents in the intercom terminal and establish connections with the base stations or repeaters under its jurisdiction through standard interfaces to collect the system type, supported codec format list, physical port or time slot unique index, maximum concurrent communication capacity, current occupancy status, network round-trip latency and session load percentage for each physical channel. Read the standard type field of each physical channel, and split the physical channel into the analog channel abstract pipeline and the digital channel abstract pipeline according to the field value. For physical channels that are determined to be analog, call the analog channel modeling subroutine to generate independent virtual channel objects and assign them a globally unique identifier composed of proxy identifier and port index. In the simulated channel modeling subroutine, the system type is assigned to the simulated system enumeration value, the supported encoding and decoding formats are filled into the vocoder type bound to the simulated port, the concurrent communication capacity is fixed to the value one, and the current occupancy status, round-trip delay estimate, load percentage and home agent identifier are written. For digital channels determined to be using time division multiple access, the digital channel modeling subroutine is called to read the total number of time slots of the digital channel. Virtual channel instances are created for each independent time slot with time slot as the granularity. Each instance obtains a globally unique identifier composed of the proxy identifier, carrier index and time slot sequence number. In the digital channel modeling subroutine, the standard type of each virtual channel instance inherits the digital standard enumeration value of the original physical channel, the concurrent communication capacity is independently assigned the value of one, and the occupancy status, round-trip delay estimate, load percentage and home agent identifier are filled in according to the actual operation of each time slot. Once all physical channels have completed multi-dimensional attribute virtualization within this polling cycle, a global resource pool corresponding to all physical resources under the jurisdiction of this agent at the current moment is generated. All virtual channel objects and virtual channel instances in the global resource pool are encapsulated into registration update messages and sent to the coordinating controller through an encrypted control link to construct a global virtual channel pool mapping table.

[0008] In this solution, S104 specifically includes: When the distributed softswitch agent deployed on the intercom terminal captures the unstructured networking intent signal, it extracts the target group identity, terminal list and source standard type according to the predefined mapping rule table, generates a networking intent description data packet by combining the current terminal registration information, and reports it to the coordinating controller through the encrypted control link. After receiving the network intention description, the collaborative controller parses the target terminal list, queries the terminal registration information database to obtain the home agent and radio frequency reachability status of each terminal, marks the terminals that are reachable by the home agent but not reachable by radio frequency as terminals that cannot be directly addressed, and performs relay path search one by one. The breadth-first search algorithm under the maximum hop count constraint is adopted. The network topology adjacency matrix between agents is used as the search space. The initiating agent is the root node and the agent to which the target terminal belongs is the target node. The algorithm expands and traverses the adjacent agents layer by layer and increases the hop count at each layer until the target agent is reached or the preset maximum relay hop count limit is reached. If one or more reachable paths are found within the hop count constraint, the path with the fewest hop count and the largest available relay channel margin of the path agent is selected as the relay path. The path agent sequence is recorded as the relay link description, and the required relay virtual channels on the path are included in the candidate resource set. If the search fails, the corresponding terminal is removed from the target terminal list. Obtain the updated target terminal list, perform dual clustering based on the home agent identifier and target system type, and traverse the updated target terminal list to group terminals with the same home agent and target system type into the same group. Count the number of terminals in the group and set it as the minimum service capacity of the virtual channel demand unit corresponding to the target group. A virtual channel requirement unit description body is generated for each group, including requirement unit identifier, target standard type, home agent identifier, minimum capacity, source standard codec type, target standard codec type, trunk link binding relationship and service priority. All virtual channel requirement unit description bodies are aggregated to generate a virtual channel requirement unit set.

[0009] In this solution, S106 specifically includes: Extract the target standard type, home agent identifier and relay link binding relationship of each virtual channel demand unit from the set of virtual channel demand units, and filter the virtual channels that are idle or half-occupied and whose standard and home are matched in the global virtual channel pool mapping table to obtain the candidate virtual channel subset corresponding to each demand unit. Based on the candidate virtual channel subsets corresponding to each demand unit, the comprehensive cost of each demand unit and candidate virtual channel is calculated pairwise, and a multi-dimensional dynamic comprehensive cost matrix is ​​constructed with demand units as rows and candidate virtual channels as columns. The comprehensive cost is obtained by weighted summation of transmission delay cost, encoding / decoding conversion distortion cost, load cost, and allocation / switching cost. An improved Hungarian algorithm with capacity awareness is introduced to solve the globally optimal channel allocation scheme. Candidate virtual channels with a capacity greater than a preset capacity threshold in the multidimensional dynamic integrated cost matrix are split into a corresponding number of virtual channel replica nodes with a capacity of one, and the original integrated cost value is inherited. A bipartite graph is constructed with each virtual channel demand unit as the left node, all virtual channel replica nodes as the right node, and the comprehensive cost inherited by the replica nodes as the weight of the connection edge from the demand unit to the replica node. The demand units are sorted in descending order according to the product of service priority and minimum capacity to generate a processing queue, and the node potential label and empty matching set are initialized. Demand units are retrieved one by one from the processing queue in priority queue order. An augmenting path is searched in the equal subgraph formed by all edges that satisfy the constraint of equal node potential labels. If no path is found, the labels are adjusted to expand the equal subgraph and the search is restarted. Once found, the matching edge is flipped along the path and the corresponding virtual channel replica node is marked as occupied. When a high-priority demand unit cannot find an available replica node after repeated searches in the equal subgraph after labeling adjustment, no more than two demand units are released from the matched low-priority demands and put back into the waiting queue with the currently blocked demand unit to perform augmenting path search and matching flip operation, thus completing limited backtracking redistribution; After all demand units in the queue are matched, the matching relationship of each virtual channel replica node is reverse mapped back to the original virtual channel. The occupancy status of multiple replica nodes under the same original channel is merged to update the remaining channel capacity, and a virtual channel allocation scheme is generated with the demand unit identifier as the key, the allocated virtual channel identifier and the occupied capacity as the value.

[0010] In this solution, S108 specifically includes: The virtual channel allocation scheme is obtained. The allocation records belonging to the same agent are aggregated into a subtask description by using the home agent identifier as the grouping key. Each subtask description is then sent to the corresponding distributed softswitch agent in parallel through the encrypted control signaling channel. After receiving the subtask description, the distributed softswitch agent sends it into the built-in standard protocol state machine. The subtask description includes the list of virtual channels that the target agent needs to activate, the call service type, the target terminal or group identity, the source and target encoding / decoding types, the upstream and downstream virtual channel binding relationship during relay forwarding, and the globally unique group call session identifier. The standard protocol state machine instantiates the corresponding protocol processing branch according to the target standard type in the subtask description, synchronously reads the call service type in the subtask description, and if the service type is group call establishment, it extracts the corresponding signaling framework from the preset message format template library, generates the native signaling sequence adapted to the local intercom system and injects it into the intercom system. The intercom system pagees the target terminal and allocates a wireless service channel based on the injected native signaling, and simultaneously monitors the feedback signaling response. When it receives the confirmation signaling that the target terminal has successfully attached to the service channel, it binds and records the corresponding virtual channel identifier and group call session identifier, and keeps the uplink voice right closed and the downlink media stream muted. Once the distributed softswitch agent confirms that all target terminals under its responsibility in the subtask description have been successfully attached to the service channel, it generates a ready message containing the agent's unique identifier, group call session identifier, and a list of ready virtual channels on its end, and sends it to the coordination controller through the encrypted control link. The collaborative controller maintains a ready status collection table using the group call session identifier as an index. When it is determined that all necessary agents under the target group call session identifier have reported ready messages, it generates an activation token containing the group call session identifier and activation timestamp within the same decision cycle, and broadcasts it to all ready agents in parallel through a low-latency control link. After each distributed softswitch agent receives the activation token and verifies the group call session identifier match, the agent responsible for the source direction opens uplink talk rights and starts forwarding media streams, while the agent responsible for the target receiving direction cancels downlink media stream mute and plays the received voice frames to the local terminal, thus enabling simultaneous entry into group calls across different terminal standards.

[0011] In this solution, S110 specifically includes: After the group call is established and enters the steady-state call phase, the distributed media processing unit continuously detects the encoding and decoding format of the voice frames output by the source virtual channel and compares it with the expected encoding and decoding format of the target virtual channel. If the formats are inconsistent, transcoding adaptation is triggered, and the encoder re-compresses the voice frames according to the target format and sends them to the target virtual channel. During the call hold, each group of call participants continuously detects voice activity on the assigned virtual channels under their jurisdiction. Using a fixed duration as a sliding window, they continuously calculate the average energy of audio frames within the window. When the average energy is lower than the preset silence threshold for a preset number of windows, a silence status flag is output. The silence timer starts timing when the silence status flag is detected for the first time. If no valid voice is detected within the preset timeout threshold, a silence release request is generated. At the same time, if the standard disconnect command of the local intercom system is captured, a disconnect notification is generated and the silence release request or disconnect notification is sent to the coordination controller. After receiving a silent release request or disconnection notification from any agent, the collaborative controller checks the virtual channel activity status of other agents under the same group call session identifier. If there is still voice activity on other channels, only the virtual channel of the requesting agent is marked as pending reclamation. If all channels have been silent or have received disconnection instructions, a global reclamation command is generated and sent to all participating agents. After receiving the global reclaim command, each agent sends a call release primitive to its local intercom system to disconnect the wireless service channel connection, updates the status of its virtual channels to idle and clears the group call session identifier binding, releases the media processing buffer and network port resources, and after completing the release, sends a release confirmation message containing the list of released virtual channels to the coordination controller. After the coordinating controller receives release confirmation messages from all agents, it updates the corresponding virtual channels to idle status in batches in the global virtual channel pool mapping table. At the same time, the agents and the coordinating controller maintain periodic heartbeat interactions, and the agents reset their local lease timers immediately upon receiving a valid message. If the lease timer expires, the connection is determined to be interrupted. The system will automatically set the allocated virtual channels under its jurisdiction to idle, release media resources and generate a recycling log. After the connection is restored, the log will be synchronized to the coordination controller to update the global state.

[0012] A second aspect of the present invention provides a heterogeneous walkie-talkie dynamic networking communication system based on a virtual channel pool, the system comprising a distributed softswitch agent, a cooperative controller, and a distributed media processing unit: The distributed softswitch agent is deployed at the boundary of each intercom system to abstract the physical channels under its jurisdiction into virtual channel objects to build a global resource pool, capture unstructured networking intent signals and convert them into networking intent descriptions, translate the received subtask descriptions into native signaling sequences adapted to the local intercom system through the built-in standard protocol state machine, report ready messages to the coordination controller and respond to activation tokens, and perform silent detection and lease timeout recovery. The collaborative controller communicates with each distributed softswitch agent to maintain a global virtual channel pool mapping table, generate a set of virtual channel demand units based on the received network intent description, construct a comprehensive cost matrix and run a capacity-aware improved Hungarian algorithm to solve the virtual channel allocation scheme, issue subtask descriptions to each agent, issue activation tokens after collecting ready messages, and generate a global recycling command based on release requests. The distributed media processing unit works in conjunction with the cooperative controller and each distributed softswitch agent to detect the consistency of the vocoders between the source virtual channel and the target virtual channel during the group call hold phase, and to perform real-time transcoding adaptation by decoding and re-encoding media streams with mismatched vocoders. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments or examples of the present invention, the drawings used in the embodiments or examples will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained according to these drawings without creative effort.

[0014] Figure 1 A flowchart of a heterogeneous walkie-talkie dynamic networking communication method based on a virtual channel pool is provided in an embodiment of the present invention; Figure 2 This is a flowchart of a communication period control method for dynamic networking of heterogeneous walkie-talkies provided in an embodiment of the present invention; Figure 3 This is a block diagram of a heterogeneous walkie-talkie dynamic networking communication system based on a virtual channel pool, provided in an embodiment of the present invention. The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0015] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0016] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0017] Figure 1 A flowchart of a heterogeneous walkie-talkie dynamic networking communication method based on a virtual channel pool is provided in an embodiment of the present invention; like Figure 1 As shown in the figure, the present invention provides a flowchart of a heterogeneous walkie-talkie dynamic networking communication method based on a virtual channel pool, including: S102, deploy a distributed softswitch agent on the intercom terminal and collect the standard, capacity and real-time status of the physical channel, abstract the analog port into a virtual channel object with a capacity of one, abstract the digital time division multiple access time slot into a virtual channel instance, build a global resource pool and generate a global virtual channel pool mapping table. S104, when an unstructured networking intent signal is captured, it is converted into a networking intent description according to a predefined mapping rule and reported. For those that cannot be directly radio frequency addressed, a relay path is planned through breadth-first search. Multiple terminals with the same home agent and the same target standard are merged to generate a set of virtual channel demand units carrying capacity requirements. S106, Based on the set of virtual channel demand units, candidate virtual channels are selected from the global virtual channel pool mapping table and a multi-dimensional dynamic comprehensive cost matrix is ​​constructed. An improved Hungarian algorithm based on capacity awareness is introduced, and the global optimal channel allocation scheme is solved in combination with the multi-dimensional dynamic comprehensive cost matrix to finally obtain the virtual channel allocation scheme. S108, generate and distribute the subtask descriptions of each agent according to the virtual channel allocation scheme. Each agent parses the subtasks into native signaling sequences adapted to the local intercom system through the built-in standard protocol state machine and executes them. After the terminal is attached to the service channel, report the ready message to the corresponding agent to perform cross-standard terminal group call switching. S110 continuously detects the consistency between the encoding / decoding format of the voice frames output from the source virtual channel and the target virtual channel during the group call hold phase and performs media adaptive adaptation. Simultaneously, after the call ends, it triggers dynamic reclamation of the virtual channel based on silence detection or disconnection command, and uses the lease timeout mechanism to manage the release of media resources.

[0018] Furthermore, in a preferred embodiment of the present invention, step S102 specifically includes: Deploy distributed softswitch agents in the intercom terminal and establish connections with the base stations or repeaters under its jurisdiction through standard interfaces to collect the system type, supported codec format list, physical port or time slot unique index, maximum concurrent communication capacity, current occupancy status, network round-trip latency and session load percentage for each physical channel. Read the standard type field of each physical channel, and split the physical channel into the analog channel abstract pipeline and the digital channel abstract pipeline according to the field value. For physical channels that are determined to be analog, call the analog channel modeling subroutine to generate independent virtual channel objects and assign them a globally unique identifier composed of proxy identifier and port index. In the simulated channel modeling subroutine, the system type is assigned to the simulated system enumeration value, the supported encoding and decoding formats are filled into the vocoder type bound to the simulated port, the concurrent communication capacity is fixed to the value one, and the current occupancy status, round-trip delay estimate, load percentage and home agent identifier are written. For digital channels determined to be using time division multiple access, the digital channel modeling subroutine is called to read the total number of time slots of the digital channel. Virtual channel instances are created for each independent time slot with time slot as the granularity. Each instance obtains a globally unique identifier composed of the proxy identifier, carrier index and time slot sequence number. In the digital channel modeling subroutine, the standard type of each virtual channel instance inherits the digital standard enumeration value of the original physical channel, the concurrent communication capacity is independently assigned the value of one, and the occupancy status, round-trip delay estimate, load percentage and home agent identifier are filled in according to the actual operation of each time slot. Once all physical channels have completed multi-dimensional attribute virtualization within this polling cycle, a global resource pool corresponding to all physical resources under the jurisdiction of this agent at the current moment is generated. All virtual channel objects and virtual channel instances in the global resource pool are encapsulated into registration update messages and sent to the coordinating controller through an encrypted control link to construct a global virtual channel pool mapping table.

[0019] It should be noted that in the existing technology, analog ports and digital time slots have fundamental differences in call establishment methods, multiplexing capabilities and media carrying characteristics. If network scheduling is performed directly on the original physical resources, the scheduling logic must be written separately for each standard. Not only does the system complexity increase linearly with the type of access standard, but it is also difficult to optimize heterogeneous resources in a unified manner from a global perspective.

[0020] To address the aforementioned issues, a distributed softswitch agent deployed at the system boundary is first used to continuously collect the static capability attributes and dynamic operational status of each physical channel. Among the collected metrics, the standard type determines the applicable call control protocol and media encapsulation format for the channel; the codec format list specifies the acceptable voice coding methods for the channel; the unique physical port or timeslot index provides the hardware location basis for subsequent addressing operations; the maximum concurrent communication capacity identifies the number of independent calls the physical resource can handle simultaneously; the current occupancy status reflects resource availability; and network round-trip latency and session load percentage characterize the real-time transmission quality and congestion level of the channel. In the abstraction processing phase, the physical channels are routed to different processing pipelines based on their standard type. Analog channels lack time slot allocation capabilities; a single analog port can only carry one call at a time. Therefore, they are directly abstracted as virtual channel objects with a fixed capacity of one. Digital time-division multiple access (DDMA) channels, however, divide the communication time of each carrier into several periodically repeating independent time slots. Each time slot can be independently allocated and carry one call. Therefore, each time slot must be independently abstracted as a virtual channel instance with a capacity of one, rather than treating the entire carrier as a single scheduling unit. This abstraction method, based on the finest-grained independent schedulable unit, ensures that every logical resource in the virtual channel pool has the characteristics of occupancy and release, avoiding capacity ambiguity in subsequent allocation algorithms when handling digital channels. For example, in a DMR repeater channel configured with two time slots, if the entire carrier is abstracted as a virtual channel object with a capacity of two, the allocation algorithm cannot precisely specify which time slot to occupy when handling a request requiring only one time slot, which is detrimental to subsequent native signaling generation and time slot assignment operations. After splitting the carrier into two independent virtual channel instances with a capacity of one, each instance can be independently allocated, occupied, and reclaimed, eliminating ambiguity in capacity management. After abstraction, the agent encapsulates the virtual channel objects and instances generated in this round into a registration update message and sends it to the coordinating controller. The controller then maintains a cross-standard, cross-system virtual channel pool mapping table using a globally unique identifier as the primary index. This mapping table provides a unified resource query view for demand matching, cost calculation, and channel allocation in subsequent dynamic networking decisions.

[0021] Furthermore, in a preferred embodiment of the present invention, step S104 specifically includes: When the distributed softswitch agent deployed on the intercom terminal captures the unstructured networking intent signal, it extracts the target group identity, terminal list and source standard type according to the predefined mapping rule table, generates a networking intent description data packet by combining the current terminal registration information, and reports it to the coordinating controller through the encrypted control link. After receiving the network intention description, the collaborative controller parses the target terminal list, queries the terminal registration information database to obtain the home agent and radio frequency reachability status of each terminal, marks the terminals that are reachable by the home agent but not reachable by radio frequency as terminals that cannot be directly addressed, and performs relay path search one by one. The breadth-first search algorithm under the maximum hop count constraint is adopted. The network topology adjacency matrix between agents is used as the search space. The initiating agent is the root node and the agent to which the target terminal belongs is the target node. The algorithm expands and traverses the adjacent agents layer by layer and increases the hop count at each layer until the target agent is reached or the preset maximum relay hop count limit is reached. If one or more reachable paths are found within the hop count constraint, the path with the fewest hop count and the largest available relay channel margin of the path agent is selected as the relay path. The path agent sequence is recorded as the relay link description, and the required relay virtual channels on the path are included in the candidate resource set. If the search fails, the corresponding terminal is removed from the target terminal list. Obtain the updated target terminal list, perform dual clustering based on the home agent identifier and target system type, and traverse the updated target terminal list to group terminals with the same home agent and target system type into the same group. Count the number of terminals in the group and set it as the minimum service capacity of the virtual channel demand unit corresponding to the target group. A virtual channel requirement unit description body is generated for each group, including requirement unit identifier, target standard type, home agent identifier, minimum capacity, source standard codec type, target standard codec type, trunk link binding relationship and service priority. All virtual channel requirement unit description bodies are aggregated to generate a virtual channel requirement unit set.

[0022] It should be noted that this step transforms the unstructured networking intent triggered by frontline commanders through terminal operations into a structured set of virtual channel demand units. This is a core step in realizing the transition from "manual static configuration" to "software-defined, on-demand connectivity." In existing technologies, the establishment of cross-system call groups relies on administrators pre-defining port mappings and call routes in the background, preventing commanders from initiating dynamic networking in real time based on the situation on the ground.

[0023] To address the aforementioned issues, during the intent acquisition phase, the distributed softswitch agent continuously monitors multiple input channels from terminals. Dual-tone multi-frequency (DTMF) code sequences are typically generated by the commander inputting specific function code combinations via a walkie-talkie keypad. Programmable function keys can be pre-programmed with one-click network setup commands, while the voice command channel receives short voice commands processed by front-end noise reduction. The agent's internal intent parsing engine converts these unstructured signals into structured network setup intent descriptions based on a predefined mapping rule table. This rule table pre-associates each DTMF code sequence or key combination with a target group identity and its corresponding terminal member list, thus automatically mapping a simple key press by the commander into a network setup task containing a complete terminal list and service priorities.

[0024] After receiving the network deployment intention description, the collaborative controller first performs terminal reachability verification and relay path planning. In multi-department joint operation scenarios, some target terminals may not be directly paged by the system's base station due to being located in different wireless coverage areas, but their home agent and the initiating agent have IP network connectivity. For these terminals, the controller uses a breadth-first search algorithm under the maximum hop count constraint to automatically search for the optimal relay path forwarded by a third-party agent in the inter-agent network topology, and includes the required relay virtual channels on the path in the candidate resource set. If the search fails, the unreachable terminal is removed from the list and logged to avoid wasting resources by reserving channels for unreachable terminals in the subsequent allocation phase. This allows terminals that were originally limited by wireless coverage boundaries and unable to participate in the network to be flexibly included, significantly expanding the coverage of dynamic networking. After completing the relay planning, a double clustering and merging operation is performed on the updated target terminal list. The basis for this is that multiple terminals belonging to the same agent and using the same target standard can achieve downlink broadcast or multicast coverage through a single virtual channel under the agent's jurisdiction, without needing to allocate a separate channel for each terminal. Therefore, grouping terminals that meet the dual consistency condition into the same group and counting the number of terminals as the minimum service capacity of that group not only reduces the number of virtual channels occupied, but also provides accurate capacity requirement parameters for subsequent optimization algorithms. The final generated set of virtual channel requirement units organizes the target terminals according to their home agent and target standard. Each requirement unit carries complete information such as minimum capacity, codec type, relay link binding relationship, and service priority, allowing subsequent cost calculation and channel allocation to be directly based on this set for global optimization calculations without needing to query the original network request again.

[0025] Furthermore, in a preferred embodiment of the present invention, S106 specifically includes: Extract the target standard type, home agent identifier and relay link binding relationship of each virtual channel demand unit from the set of virtual channel demand units, and filter the virtual channels that are idle or half-occupied and whose standard and home are matched in the global virtual channel pool mapping table to obtain the candidate virtual channel subset corresponding to each demand unit. Based on the candidate virtual channel subsets corresponding to each demand unit, the comprehensive cost of each demand unit and candidate virtual channel is calculated pairwise, and a multi-dimensional dynamic comprehensive cost matrix is ​​constructed with demand units as rows and candidate virtual channels as columns. The comprehensive cost is obtained by weighted summation of transmission delay cost, encoding / decoding conversion distortion cost, load cost, and allocation / switching cost. An improved Hungarian algorithm with capacity awareness is introduced to solve the globally optimal channel allocation scheme. Candidate virtual channels with a capacity greater than a preset capacity threshold in the multidimensional dynamic integrated cost matrix are split into a corresponding number of virtual channel replica nodes with a capacity of one, and the original integrated cost value is inherited. A bipartite graph is constructed with each virtual channel demand unit as the left node, all virtual channel replica nodes as the right node, and the comprehensive cost inherited by the replica nodes as the weight of the connection edge from the demand unit to the replica node. The demand units are sorted in descending order according to the product of service priority and minimum capacity to generate a processing queue, and the node potential label and empty matching set are initialized. Demand units are retrieved one by one from the processing queue in priority queue order. An augmenting path is searched in the equal subgraph formed by all edges that satisfy the constraint of equal node potential labels. If no path is found, the labels are adjusted to expand the equal subgraph and the search is restarted. Once found, the matching edge is flipped along the path and the corresponding virtual channel replica node is marked as occupied. When a high-priority demand unit cannot find an available replica node after repeated searches in the equal subgraph after labeling adjustment, no more than two demand units are released from the matched low-priority demands and put back into the waiting queue with the currently blocked demand unit to perform augmenting path search and matching flip operation, thus completing limited backtracking redistribution; After all demand units in the queue are matched, the matching relationship of each virtual channel replica node is reverse mapped back to the original virtual channel. The occupancy status of multiple replica nodes under the same original channel is merged to update the remaining channel capacity, and a virtual channel allocation scheme is generated with the demand unit identifier as the key, the allocated virtual channel identifier and the occupied capacity as the value.

[0026] It should be noted that the selection of cross-system bridging channels relies on manual configuration or simple idle-priority assignment, which cannot comprehensively consider multi-dimensional factors such as latency, distortion, and load, nor can it achieve globally optimal allocation under multiple concurrent demands. This step addresses these problems by constructing a quantifiable multi-dimensional cost model and introducing a capacity-constrained combinatorial optimization algorithm, upgrading channel allocation from static manual decision-making to dynamic global optimization decision-making.

[0027] Specifically, during the candidate channel selection phase, the collaborative controller uses the target standard type and home agent identifier of each demand unit as joint selection criteria, performing exact matching in the global virtual channel pool mapping table. The selection logic takes idle or half-occupied status as an availability prerequisite. A half-occupied channel refers to a digital time slot resource that has already carried one call but still has remaining capacity; including it in the candidate set can further improve resource reuse. If the demand unit carries a trunk link binding relationship, the available virtual channels under the jurisdiction of each path agent on the trunk path are included in the candidate subset to ensure that the intermediate channel resources required for trunk forwarding are synchronously reserved in subsequent allocations. The overall cost is obtained by weighted summation of cost components from four dimensions. Transmission latency cost reflects the sum of network round-trip latency from the initiating agent to the candidate channel's home agent and the encoding / decoding conversion processing latency; the lower the latency, the lower the cost. Encoding / decoding conversion distortion cost is obtained by looking up a pre-built encoding / decoding concatenated distortion mapping database. This database, calibrated through large-scale auditory evaluation experiments, specifies the voice quality attenuation value corresponding to each encoding format conversion combination. The attenuation value obtained from the table is linearly mapped to the distortion cost component. Load cost maps the current load rate of the candidate channel to the cost using a non-linear function. This function increases gradually in the low-load range and steeply in the high-load range, thus guiding the allocation algorithm to prioritize lightly loaded channels and avoid load concentration. Allocation handover cost generates a fixed penalty value when allocation requires interrupting non-urgent existing communication; otherwise, it is zero, to avoid frequent interruptions of existing calls to establish new group calls. The weight coefficients of the four components are dynamically correlated with the service priority of the network request. High-priority requests automatically increase the weight of latency and distortion, ensuring rapid response and clear voice in emergency communication scenarios.

[0028] In the optimal allocation solution phase, a capacity-aware improved Hungarian algorithm is introduced. The standard Hungarian algorithm can only handle one-to-one perfect matching problems and cannot cope with the real-world constraints of virtual channels having different concurrent capacities. This algorithm uses a capacity replica splitting strategy to split a channel with a capacity of N into N independent replica nodes with a capacity of one, so that each replica can be independently occupied, thus transforming the combinatorial optimization problem with capacity constraints into a standard bipartite graph minimum cost matching problem. The replica nodes inherit the comprehensive cost of the original channel, ensuring the consistency of edge weights after splitting. The algorithm arranges the processing queue in descending order of the product of service priority and minimum capacity, so that demand units with high resource consumption and high priority get matching opportunities first. During the iterative search for augmenting paths, the expansion and contraction range of the equal subgraph is controlled by dynamically adjusting the node potential labels, ensuring that each match is the optimal individual allocation for the current demand unit. When a high-priority demand is blocked because resources are occupied by low-priority demands, a finite backtracking redistribution mechanism allows at most two low-priority demands to be released from the already matched ones and the search to be restarted, exchanging controllable computational overhead for a further reduction in the overall global cost. The design rationale for setting the maximum backtracking depth to two is that, under typical network scales, the cost improvement brought by backtracking beyond two layers approaches zero, while the computation time increases exponentially. Therefore, this constraint achieves a balance between optimization accuracy and solution efficiency. After allocation, the replica-level matching is restored to the physical channel-level scheme through inverse mapping and same-source merging operations, generating a virtual channel allocation table that can be directly executed.

[0029] Furthermore, in a preferred embodiment of the present invention, S108 specifically includes: The virtual channel allocation scheme is obtained. The allocation records belonging to the same agent are aggregated into a subtask description by using the home agent identifier as the grouping key. Each subtask description is then sent to the corresponding distributed softswitch agent in parallel through the encrypted control signaling channel. After receiving the subtask description, the distributed softswitch agent sends it into the built-in standard protocol state machine. The subtask description includes the list of virtual channels that the target agent needs to activate, the call service type, the target terminal or group identity, the source and target encoding / decoding types, the upstream and downstream virtual channel binding relationship during relay forwarding, and the globally unique group call session identifier. The standard protocol state machine instantiates the corresponding protocol processing branch according to the target standard type in the subtask description, synchronously reads the call service type in the subtask description, and if the service type is group call establishment, it extracts the corresponding signaling framework from the preset message format template library, generates the native signaling sequence adapted to the local intercom system and injects it into the intercom system. The intercom system pagees the target terminal and allocates a wireless service channel based on the injected native signaling, and simultaneously monitors the feedback signaling response. When it receives the confirmation signaling that the target terminal has successfully attached to the service channel, it binds and records the corresponding virtual channel identifier and group call session identifier, and keeps the uplink voice right closed and the downlink media stream muted. Once the distributed softswitch agent confirms that all target terminals under its responsibility in the subtask description have been successfully attached to the service channel, it generates a ready message containing the agent's unique identifier, group call session identifier, and a list of ready virtual channels on its end, and sends it to the coordination controller through the encrypted control link. The collaborative controller maintains a ready status collection table using the group call session identifier as an index. When it is determined that all necessary agents under the target group call session identifier have reported ready messages, it generates an activation token containing the group call session identifier and activation timestamp within the same decision cycle, and broadcasts it to all ready agents in parallel through a low-latency control link. After each distributed softswitch agent receives the activation token and verifies the group call session identifier match, the agent responsible for the source direction opens uplink talk rights and starts forwarding media streams, while the agent responsible for the target receiving direction cancels downlink media stream mute and plays the received voice frames to the local terminal, thus enabling simultaneous entry into group calls across different terminal standards.

[0030] It should be noted that heterogeneous system group call establishment typically involves each system independently initiating a call, followed by the gateway directly bridging the media stream. Due to significant differences in processing time required for signaling interactions such as paging terminals, channel allocation, and access confirmation across different systems, some terminals experience voice loss of hundreds of milliseconds or even seconds at the start of the call. This step aims to resolve this inherent problem of asynchronous cross-system group call establishment time.

[0031] Specifically, during the subtask generation and distribution phase, the collaborative controller uses the attribution agent identifier as the grouping key to aggregate all virtual channel allocation records involving the same agent in the allocation scheme into a single subtask description. This agent-based aggregation design ensures that each agent only needs to process a subset of tasks relevant to its own end, avoiding the redundant information transmission and parsing overhead associated with distributing the global scheme to agents. The globally unique group call session identifier carried in the subtask description is used throughout the subsequent readiness reporting and activation token association matching process, serving as a key association field to ensure operational consistency between the distributed agents and the centralized controller.

[0032] Furthermore, due to the fundamental differences in message format, parameter encoding, and interaction timing between call setup signaling of different intercom systems, the state machine translates the abstract operational semantics in the subtask description into specific signaling sequences recognizable by the peer system through a pre-set message format template library for each system. For DMR systems, the state machine generates single-call or group-call setup request signaling conforming to the DMR air interface protocol and completes time slot assignment; for public network intercom systems, it generates session invitation requests conforming to SIP or proprietary protocols. After injecting the generated native signaling into the intercom system, the agent continuously monitors the signaling responses returned by the system. Upon receiving confirmation signaling that the terminal has successfully attached to the service channel, it binds and records the virtual channel identifier corresponding to the terminal with the group-call session identifier. At this time, the agent maintains a pre-activation waiting state with uplink talk rights disabled and downlink media stream muted. The significance of this is that although the terminal has completed the establishment of the wireless link and resides on the service channel, it has not yet obtained speaking authorization and cannot hear any media stream, and is in an intermediate state of "accessed but not activated". During the synchronous activation phase, the coordination controller maintains a readiness status collection table indexed by the group call session identifier, verifying the readiness messages reported by each agent one by one. Once it is determined that all necessary agents under the group call session identifier are ready, the controller generates an activation token and broadcasts it in parallel within the same decision cycle. The parallel broadcasting of the activation token ensures that each agent receives the activation command at almost the same moment. Upon receiving the token, the agent immediately performs either voice access granting or media stream release, thereby enabling synchronous entry into the group call at the user perception level for terminals of different standards. This solves the problem of voice truncation caused by signaling interaction latency differences in traditional solutions, ensuring the complete transmission of command instructions.

[0033] Figure 2 This is a flowchart of a communication period control method for dynamic networking of heterogeneous walkie-talkies provided in an embodiment of the present invention; like Figure 2 As shown, the present invention provides a flowchart of a communication period control method for dynamic networking of heterogeneous walkie-talkies, including: After the group call is established and enters the steady-state call phase, the distributed media processing unit continuously detects the encoding and decoding format of the voice frames output by the source virtual channel and compares it with the expected encoding and decoding format of the target virtual channel. If the formats are inconsistent, transcoding adaptation is triggered, and the encoder re-compresses the voice frames according to the target format and sends them to the target virtual channel. During the call hold, each group of call participants continuously detects voice activity on the assigned virtual channels under their jurisdiction. Using a fixed duration as a sliding window, they continuously calculate the average energy of audio frames within the window. When the average energy is lower than the preset silence threshold for a preset number of windows, a silence status flag is output. The silence timer starts timing when the silence status flag is detected for the first time. If no valid voice is detected within the preset timeout threshold, a silence release request is generated. At the same time, if the standard disconnect command of the local intercom system is captured, a disconnect notification is generated and the silence release request or disconnect notification is sent to the coordination controller. After receiving a silent release request or disconnection notification from any agent, the collaborative controller checks the virtual channel activity status of other agents under the same group call session identifier. If there is still voice activity on other channels, only the virtual channel of the requesting agent is marked as pending reclamation. If all channels have been silent or have received disconnection instructions, a global reclamation command is generated and sent to all participating agents. After receiving the global reclaim command, each agent sends a call release primitive to its local intercom system to disconnect the wireless service channel connection, updates the status of its virtual channels to idle and clears the group call session identifier binding, releases the media processing buffer and network port resources, and after completing the release, sends a release confirmation message containing the list of released virtual channels to the coordination controller. After the coordinating controller receives release confirmation messages from all agents, it updates the corresponding virtual channels to idle status in batches in the global virtual channel pool mapping table. At the same time, the agents and the coordinating controller maintain periodic heartbeat interactions, and the agents reset their local lease timers immediately upon receiving a valid message. If the lease timer expires, the connection is determined to be interrupted. The system will automatically set the allocated virtual channels under its jurisdiction to idle, release media resources and generate a recycling log. After the connection is restored, the log will be synchronized to the coordination controller to update the global state.

[0034] It should be noted that existing cross-system interconnection gateways only have simple media pass-through capabilities and cannot automatically adapt to format differences between different vocoders; resource reclamation relies on a single timeout or manual release, lacking fault tolerance protection for extreme anomalies such as link interruptions, which can easily lead to idle channel resources. Therefore, to solve the above problems, this solution proposes a full lifecycle management mechanism from the establishment of a coverage group call to the maintenance of steady-state calls and the final reclamation of resources.

[0035] Specifically, during the media adaptation phase, the distributed media processing unit continuously compares the encoding and decoding formats of each cross-standard media path. Since different intercom systems use different vocoders—for example, analog systems commonly use G.711 encoding, DMR systems use AMBE+2 vocoders, and public network intercom systems often use Opus encoding—if a compressed voice frame of one format is directly transmitted to a terminal of another standard, the target terminal will be unable to parse it, resulting in silence or noise. The media processing unit automatically identifies the source format by parsing the frame header identifier field of the source voice frame and compares it with the desired encoding and decoding format recorded by the target virtual channel during the allocation phase. When the formats are inconsistent, a decoding and re-encoding pipeline is triggered, first restoring the source voice frame to a pulse code modulation audio sequence and then recompressing it according to the target format. When the formats are consistent, it is directly transmitted, minimizing unnecessary processing delays.

[0036] During the call termination detection phase, each call group and participating agent operates a dual mechanism in parallel: silent detection and disconnection command monitoring. Silence detection employs a sliding window energy average algorithm, requiring that the energy average of multiple consecutive windows fall below a threshold before starting the silence timer. The call is only considered to have ended naturally when the silence duration exceeds a preset timeout threshold, thus achieving a balance between timely termination and preventing false alarms. Disconnection command monitoring ensures that call termination initiated by an authorized terminal is responded to immediately. Upon receiving a release request, the coordination controller first checks the channel activity status of other agents in the same call group session to prevent the entire group from being mistakenly disconnected due to single-end silence. In extreme cases where network interruptions prevent the delivery of termination commands between the agent and the controller, periodic heartbeat interactions are maintained between the agent and the controller. Each time an agent receives a valid message, it resets its local lease timer. If no valid message is received within a preset duration, the connection is considered interrupted, and the agent autonomously forcibly sets its assigned virtual channels to idle and releases media resources. After the connection is restored, the agent synchronously reclaims logs to the coordination controller to update the global state. This mechanism ensures that virtual channel resources can be recovered in a timely and complete manner under any abnormal operating conditions, achieving consistent release and zero residue of heterogeneous resources.

[0037] Furthermore, the heterogeneous walkie-talkie dynamic networking communication method based on virtual channel pools provided by this invention also includes the following steps: The distributed softswitch agent deployed on the commander terminal captures cross-group emergency call interruption commands triggered by preset function keys or specific dual-tone multi-frequency codes, extracts the target group call identity identifier from them, and encapsulates the current group call session identifier, the commander's agent identifier, and the target group call identity identifier into temporary bridging request data and sends it to the coordination controller. After receiving a temporary bridging request, the coordination controller queries the active group call status table for the group call context record corresponding to the current group call session identifier and the target group call identity identifier, extracts the source virtual channel and home agent of the commander's uplink voice, as well as the list of target virtual channels and home agents received by each terminal of the target group call downlink, and generates a bridging link endpoint description. The collaborative controller reads the network topology connectivity and available forwarding port resource status between the source agent and each target agent in the global virtual channel pool mapping table. After determining that the resource conditions are met, it configures temporary media forwarding mapping rules between the media processing units of the source agent and each target agent to establish a one-way temporary media bridging link from the source group call to the target group call. After the temporary bridging link is established, the source agent will send the commander's voice frame to the current group call downlink channel and the temporary bridging link at the same time. After each target agent receives the voice frame through the temporary bridging link, it will perform time-stamp alignment and amplitude weighting superposition with the downlink voice frame generated locally in the group call in the audio mixing buffer to generate a merged downlink voice frame and send it to each target terminal through the virtual channel allocated in the group call. When the collaborative controller receives a bridge teardown command or detects that the continuous silence duration on the temporary bridge link exceeds the preset bridge silence teardown threshold, it issues a bridge teardown command to the source agent and each target agent. Each agent deletes the temporary media forwarding mapping rule and releases the forwarding port and mixing buffer resources, and the two groups of calls resume independent call status.

[0038] It should be noted that, in response to the urgent need for cross-group emergency communication in multi-team collaborative combat scenarios, a temporary bridging method for cross-group calls is proposed, extending the existing dynamic networking framework. In complex emergency operations such as high-rise building fires and geological disaster rescue, multiple independently operating dynamic communication groups often exist simultaneously on-site, each maintaining internal communication according to its task assignments. When a group discovers a sudden emergency and needs to immediately notify other groups, the traditional method requires the commander to manually initiate a merged network request involving all relevant groups, or for the commander to manually relay information between two groups. The former requires a complete networking process, from intent capture and resource allocation to synchronization activation, taking at least several seconds, which cannot meet the requirement of second-level response; the latter adds intermediate links in information transmission, easily introducing relay errors and decision-making delays.

[0039] The core improvement of this scheme lies in establishing a temporary media bridging link at the media layer, based on the forwarding capabilities of existing virtual channels, instead of rebuilding group calls or allocating new channels to achieve cross-group communication. Upon receiving a temporary bridging request, the coordination controller queries the context records of the current and target group calls in the active group call status table, extracts the virtual channels and home agent information of the source and target ends, and combines them into a bridging link endpoint description. Subsequently, the controller reads the network topology connectivity and available forwarding port status between the source agent and each target agent in the global virtual channel pool mapping table. When resource conditions are met, it directly configures temporary media forwarding mapping rules to establish a unidirectional temporary media bridging link from the source group call to the target group call. The source agent generates a bypass output from the commander's uplink voice frame through internal replication, and sends it to each target agent after encoding, decoding, and adaptation. The target agent timestamps and weights the voice frames received by the bridging link with the local downlink voice frames of the group call in the mixing buffer, merges them, and sends them to each target terminal through the virtual channels allocated to the group call. The target group's intra-group terminal perceives simultaneously the original content of its own group and the commander's interjection from another group within the same call, without requiring any terminal operation. While maintaining the integrity of the original dynamic networking framework, it achieves instant voice penetration across groups with minimal resource cost, improving the communication flexibility and emergency response efficiency of heterogeneous intercom systems in multi-group collaborative combat scenarios.

[0040] Figure 3 A heterogeneous walkie-talkie dynamic networking communication system 3 based on a virtual channel pool is provided as an embodiment of the present invention. The system includes a distributed softswitch agent, a cooperative controller, and a distributed media processing unit. The distributed softswitch agent is deployed at the boundary of each intercom system to abstract the physical channels under its jurisdiction into virtual channel objects to build a global resource pool, capture unstructured networking intent signals and convert them into networking intent descriptions, translate the received subtask descriptions into native signaling sequences adapted to the local intercom system through the built-in standard protocol state machine, report ready messages to the coordination controller and respond to activation tokens, and perform silent detection and lease timeout recovery. The collaborative controller communicates with each distributed softswitch agent to maintain a global virtual channel pool mapping table, generate a set of virtual channel demand units based on the received network intent description, construct a comprehensive cost matrix and run a capacity-aware improved Hungarian algorithm to solve the virtual channel allocation scheme, issue subtask descriptions to each agent, issue activation tokens after collecting ready messages, and generate a global recycling command based on release requests. The distributed media processing unit works in conjunction with the cooperative controller and each distributed softswitch agent to detect the consistency of the vocoders between the source virtual channel and the target virtual channel during the group call hold phase, and to perform real-time transcoding adaptation by decoding and re-encoding media streams with mismatched vocoders.

[0041] 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 variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for dynamic networking communication of heterogeneous walkie-talkies based on a virtual channel pool, characterized in that, include: S102, deploy a distributed softswitch agent on the intercom terminal and collect the standard, capacity and real-time status of the physical channel, abstract the analog port into a virtual channel object with a capacity of one, abstract the digital time division multiple access time slot into a virtual channel instance, build a global resource pool and generate a global virtual channel pool mapping table. S104, when an unstructured networking intent signal is captured, it is converted into a networking intent description according to a predefined mapping rule and reported. For those that cannot be directly radio frequency addressed, a relay path is planned through breadth-first search. Multiple terminals with the same home agent and the same target standard are merged to generate a set of virtual channel demand units carrying capacity requirements. S106, Based on the set of virtual channel demand units, candidate virtual channels are selected from the global virtual channel pool mapping table and a multi-dimensional dynamic comprehensive cost matrix is ​​constructed. An improved Hungarian algorithm based on capacity awareness is introduced, and the global optimal channel allocation scheme is solved in combination with the multi-dimensional dynamic comprehensive cost matrix to finally obtain the virtual channel allocation scheme. S108, generate and distribute the subtask descriptions of each agent according to the virtual channel allocation scheme. Each agent parses the subtasks into native signaling sequences adapted to the local intercom system through the built-in standard protocol state machine and executes them. After the terminal is attached to the service channel, report the ready message to the corresponding agent to perform cross-standard terminal group call switching. S110 continuously detects the consistency between the encoding / decoding format of the voice frames output from the source virtual channel and the target virtual channel during the group call hold phase and performs media adaptive adaptation. Simultaneously, after the call ends, it triggers dynamic reclamation of the virtual channel based on silence detection or disconnection command, and uses the lease timeout mechanism to manage the release of media resources.

2. The heterogeneous walkie-talkie dynamic networking communication method based on a virtual channel pool according to claim 1, characterized in that, S102 specifically includes: Deploy distributed softswitch agents in the intercom terminal and establish connections with the base stations or repeaters under its jurisdiction through standard interfaces to collect the system type, supported codec format list, physical port or time slot unique index, maximum concurrent communication capacity, current occupancy status, network round-trip latency and session load percentage for each physical channel. Read the standard type field of each physical channel, and split the physical channel into the analog channel abstract pipeline and the digital channel abstract pipeline according to the field value. For physical channels that are determined to be analog, call the analog channel modeling subroutine to generate independent virtual channel objects and assign them a globally unique identifier composed of proxy identifier and port index. In the simulated channel modeling subroutine, the system type is assigned to the simulated system enumeration value, the supported encoding and decoding formats are filled into the vocoder type bound to the simulated port, the concurrent communication capacity is fixed to the value one, and the current occupancy status, round-trip delay estimate, load percentage and home agent identifier are written. For digital channels determined to be using time division multiple access, the digital channel modeling subroutine is called to read the total number of time slots of the digital channel. Virtual channel instances are created for each independent time slot with time slot as the granularity. Each instance obtains a globally unique identifier composed of the proxy identifier, carrier index and time slot sequence number. In the digital channel modeling subroutine, the standard type of each virtual channel instance inherits the digital standard enumeration value of the original physical channel, the concurrent communication capacity is independently assigned the value of one, and the occupancy status, round-trip delay estimate, load percentage and home agent identifier are filled in according to the actual operation of each time slot. Once all physical channels have completed multi-dimensional attribute virtualization within this polling cycle, a global resource pool corresponding to all physical resources under the jurisdiction of this agent at the current moment is generated. All virtual channel objects and virtual channel instances in the global resource pool are encapsulated into registration update messages and sent to the coordinating controller through an encrypted control link to construct a global virtual channel pool mapping table.

3. The heterogeneous walkie-talkie dynamic networking communication method based on a virtual channel pool according to claim 1, characterized in that, S104 specifically includes: When the distributed softswitch agent deployed on the intercom terminal captures the unstructured networking intent signal, it extracts the target group identity, terminal list and source standard type according to the predefined mapping rule table, generates a networking intent description data packet by combining the current terminal registration information, and reports it to the coordinating controller through the encrypted control link. After receiving the network intention description, the collaborative controller parses the target terminal list, queries the terminal registration information database to obtain the home agent and radio frequency reachability status of each terminal, marks the terminals that are reachable by the home agent but not reachable by radio frequency as terminals that cannot be directly addressed, and performs relay path search one by one. The breadth-first search algorithm under the maximum hop count constraint is adopted. The network topology adjacency matrix between agents is used as the search space. The initiating agent is the root node and the agent to which the target terminal belongs is the target node. The algorithm expands and traverses the adjacent agents layer by layer and increases the hop count at each layer until the target agent is reached or the preset maximum relay hop count limit is reached. If one or more reachable paths are found within the hop count constraint, the path with the fewest hop count and the largest available relay channel margin of the path agent is selected as the relay path. The path agent sequence is recorded as the relay link description, and the required relay virtual channels on the path are included in the candidate resource set. If the search fails, the corresponding terminal is removed from the target terminal list. Obtain the updated target terminal list, perform dual clustering based on the home agent identifier and target system type, and traverse the updated target terminal list to group terminals with the same home agent and target system type into the same group. Count the number of terminals in the group and set it as the minimum service capacity of the virtual channel demand unit corresponding to the target group. A virtual channel requirement unit description body is generated for each group, including requirement unit identifier, target standard type, home agent identifier, minimum capacity, source standard codec type, target standard codec type, trunk link binding relationship and service priority. All virtual channel requirement unit description bodies are aggregated to generate a virtual channel requirement unit set.

4. The heterogeneous walkie-talkie dynamic networking communication method based on a virtual channel pool according to claim 1, characterized in that, S106 specifically includes: Extract the target standard type, home agent identifier and relay link binding relationship of each virtual channel demand unit from the set of virtual channel demand units, and filter the virtual channels that are idle or half-occupied and whose standard and home are matched in the global virtual channel pool mapping table to obtain the candidate virtual channel subset corresponding to each demand unit. Based on the candidate virtual channel subsets corresponding to each demand unit, the comprehensive cost of each demand unit and candidate virtual channel is calculated pairwise, and a multi-dimensional dynamic comprehensive cost matrix is ​​constructed with demand units as rows and candidate virtual channels as columns. The comprehensive cost is obtained by weighted summation of transmission delay cost, encoding / decoding conversion distortion cost, load cost, and allocation / switching cost. An improved Hungarian algorithm with capacity awareness is introduced to solve the globally optimal channel allocation scheme. Candidate virtual channels with a capacity greater than a preset capacity threshold in the multidimensional dynamic integrated cost matrix are split into a corresponding number of virtual channel replica nodes with a capacity of one, and the original integrated cost value is inherited. A bipartite graph is constructed with each virtual channel demand unit as the left node, all virtual channel replica nodes as the right node, and the comprehensive cost inherited by the replica nodes as the weight of the connection edge from the demand unit to the replica node. The demand units are sorted in descending order according to the product of service priority and minimum capacity to generate a processing queue, and the node potential label and empty matching set are initialized. Demand units are retrieved one by one from the processing queue in priority queue order. An augmenting path is searched in the equal subgraph formed by all edges that satisfy the constraint of equal node potential labels. If no path is found, the labels are adjusted to expand the equal subgraph and the search is restarted. Once found, the matching edge is flipped along the path and the corresponding virtual channel replica node is marked as occupied. When a high-priority demand unit cannot find an available replica node after repeated searches in the equal subgraph after labeling adjustment, no more than two demand units are released from the matched low-priority demands and put back into the waiting queue with the currently blocked demand unit to perform augmenting path search and matching flip operation, thus completing limited backtracking redistribution; After all demand units in the queue are matched, the matching relationship of each virtual channel replica node is reverse mapped back to the original virtual channel. The occupancy status of multiple replica nodes under the same original channel is merged to update the remaining channel capacity, and a virtual channel allocation scheme is generated with the demand unit identifier as the key, the allocated virtual channel identifier and the occupied capacity as the value.

5. The heterogeneous walkie-talkie dynamic networking communication method based on a virtual channel pool according to claim 1, characterized in that, S108 specifically includes: The virtual channel allocation scheme is obtained. The allocation records belonging to the same agent are aggregated into a subtask description by using the home agent identifier as the grouping key. Each subtask description is then sent to the corresponding distributed softswitch agent in parallel through the encrypted control signaling channel. After receiving the subtask description, the distributed softswitch agent sends it into the built-in standard protocol state machine. The subtask description includes the list of virtual channels that the target agent needs to activate, the call service type, the target terminal or group identity, the source and target encoding / decoding types, the upstream and downstream virtual channel binding relationship during relay forwarding, and the globally unique group call session identifier. The standard protocol state machine instantiates the corresponding protocol processing branch according to the target standard type in the subtask description, synchronously reads the call service type in the subtask description, and if the service type is group call establishment, it extracts the corresponding signaling framework from the preset message format template library, generates the native signaling sequence adapted to the local intercom system and injects it into the intercom system. The intercom system pagees the target terminal and allocates a wireless service channel based on the injected native signaling, and simultaneously monitors the feedback signaling response. When it receives the confirmation signaling that the target terminal has successfully attached to the service channel, it binds and records the corresponding virtual channel identifier and group call session identifier, and keeps the uplink voice right closed and the downlink media stream muted. Once the distributed softswitch agent confirms that all target terminals under its responsibility in the subtask description have been successfully attached to the service channel, it generates a ready message containing the agent's unique identifier, group call session identifier, and a list of ready virtual channels on its end, and sends it to the coordination controller through the encrypted control link. The collaborative controller maintains a ready status collection table using the group call session identifier as an index. When it is determined that all necessary agents under the target group call session identifier have reported ready messages, it generates an activation token containing the group call session identifier and activation timestamp within the same decision cycle, and broadcasts it to all ready agents in parallel through a low-latency control link. After each distributed softswitch agent receives the activation token and verifies the group call session identifier match, the agent responsible for the source direction opens uplink talk rights and starts forwarding media streams, while the agent responsible for the target receiving direction cancels downlink media stream mute and plays the received voice frames to the local terminal, thus enabling simultaneous entry into group calls across different terminal standards.

6. The heterogeneous walkie-talkie dynamic networking communication method based on a virtual channel pool according to claim 1, characterized in that, S110 specifically includes: After the group call is established and enters the steady-state call phase, the distributed media processing unit continuously detects the encoding and decoding format of the voice frames output by the source virtual channel and compares it with the expected encoding and decoding format of the target virtual channel. If the formats are inconsistent, transcoding adaptation is triggered, and the encoder re-compresses the voice frames according to the target format and sends them to the target virtual channel. During the call hold, each group of call participants continuously detects voice activity on the assigned virtual channels under their jurisdiction. Using a fixed duration as a sliding window, they continuously calculate the average energy of audio frames within the window. When the average energy is lower than the preset silence threshold for a preset number of windows, a silence status flag is output. The silence timer starts timing when the silence status flag is detected for the first time. If no valid voice is detected within the preset timeout threshold, a silence release request is generated. At the same time, if the standard disconnect command of the local intercom system is captured, a disconnect notification is generated and the silence release request or disconnect notification is sent to the coordination controller. After receiving a silent release request or disconnection notification from any agent, the collaborative controller checks the virtual channel activity status of other agents under the same group call session identifier. If there is still voice activity on other channels, only the virtual channel of the requesting agent is marked as pending reclamation. If all channels have been silent or have received disconnection instructions, a global reclamation command is generated and sent to all participating agents. After receiving the global reclaim command, each agent sends a call release primitive to its local intercom system to disconnect the wireless service channel connection, updates the status of its virtual channels to idle and clears the group call session identifier binding, releases the media processing buffer and network port resources, and after completing the release, sends a release confirmation message containing the list of released virtual channels to the coordination controller. After the coordinating controller receives release confirmation messages from all agents, it updates the corresponding virtual channels to idle status in batches in the global virtual channel pool mapping table. At the same time, the agents and the coordinating controller maintain periodic heartbeat interactions, and the agents immediately reset their local lease timers whenever they receive a valid message. If the lease timer expires, the connection is determined to be interrupted. The system will automatically set the allocated virtual channels under its jurisdiction to idle, release media resources and generate a recycling log. After the connection is restored, the log will be synchronized to the coordination controller to update the global state.

7. A heterogeneous walkie-talkie dynamic networking communication system based on a virtual channel pool, characterized in that, The system includes a distributed softswitch agent, a coordination controller, and a distributed media processing unit: The distributed softswitch agent is deployed at the boundary of each intercom system to abstract the physical channels under its jurisdiction into virtual channel objects to build a global resource pool, capture unstructured networking intent signals and convert them into networking intent descriptions, translate the received subtask descriptions into native signaling sequences adapted to the local intercom system through the built-in standard protocol state machine, report ready messages to the coordination controller and respond to activation tokens, and perform silent detection and lease timeout recovery. The collaborative controller communicates with each distributed softswitch agent to maintain a global virtual channel pool mapping table, generate a set of virtual channel demand units based on the received network intent description, construct a comprehensive cost matrix and run a capacity-aware improved Hungarian algorithm to solve the virtual channel allocation scheme, issue subtask descriptions to each agent, issue activation tokens after collecting ready messages, and generate a global recycling command based on release requests. The distributed media processing unit works in conjunction with the cooperative controller and each distributed softswitch agent to detect the consistency of the vocoders between the source virtual channel and the target virtual channel during the group call hold phase, and to perform real-time transcoding adaptation by decoding and re-encoding media streams with mismatched vocoders.

Images