A mobile push-to-talk network partition merging method based on dynamic coordination of right attribution domain and cross-domain connectivity evidence

Abstract

The application discloses a mobile multi-person intercom network partition merging method, system, equipment and storage medium, and belongs to the technical field of mobile self-organizing networks. The application aims at the coordination right merging problem when multiple network partitions independently providing intercom services are reconnected, and proposes a revocable transactional merging scheme: based on the evolution relationship of the coordination right declaration, the coordination right attribution domain is aggregated; cross-domain connectivity evidence is generated through multiple conditions of bidirectional confirmation, duration and link quality; the prepared coordination node synchronizes the members, topology and talk right state of each domain, generates a merging state snapshot and a merging effective marker, and determines a voice security switching boundary; after collecting the confirmation credentials of each attribution domain, the merging is submitted, the new coordination right and the new talk right state are uniformly adopted at the switching boundary; and the merging is revoked and the original partition service is maintained in response to the cross-domain connectivity evidence or the invalidation of the merging state snapshot before the boundary. The application reduces the invalid merging triggering probability and the service interruption risk.

Description

Technical Field

[0001] This invention relates to the fields of mobile ad hoc networks, wireless multi-person intercom, network partition merging, distributed coordination, and real-time voice service continuity. Specifically, it relates to a method for revocable merging after multiple network partitions capable of independently providing intercom services have regained connectivity, based on dynamic coordination right attribution domain, cross-domain connectivity evidence, pre-coordination node state synchronization, and voice security handover boundaries. Background Technology

[0002] A Mobile Ad Hoc Network (MANET) is a self-organizing network formed by a group of mobile devices via wireless links. It does not rely on fixed infrastructure and enables voice, data, and video transmission between any two devices. Multi-user mobile intercom networks are an important application of MANETs, ​​widely used in scenarios such as cycling communication, outdoor exploration, and emergency command. In such networks, device locations constantly change, resulting in a highly dynamic network topology. Network splits and re-merges can occur, placing stringent requirements on the partition merging mechanism.

[0003] Various networking and communication schemes for mobile ad hoc networks have been disclosed in the existing technology. For example, US Patent No. 10277748B2 (published April 2, 2019) discloses a mobile ad hoc voice network for motorcycle users, including a master node, synchronization node, relay node, neighbor topology, and digital voice transmission; US Patent No. 8942197B2 (published January 27, 2015) discloses delayed network entry, network splitting and merging methods for mobile ad hoc networks; US Patent No. 7532585B2 (published May 12, 2009) discloses a master node election method based on the total number of hops and the topology center position; US Patent No. 10028110B2 (published July 17, 2018) discloses technologies such as re-merging of split PTT subgroups, coexistence of multiple talk rights holding nodes, resynchronization of talk rights request queues, and dropping of low-priority voice calls; and US Patent No. 598... US Patent No. 7011A (published November 16, 1999) and US Patent No. 8238259B2 (published August 7, 2012) disclose methods for avoiding unstable wireless links from participating in routing by utilizing continuous reception, link stability, and hysteresis mechanisms; US Patent No. 8488558B2 (published July 16, 2013) and Canadian Patent No. CA2680523A1 (published April 29, 2010) respectively disclose uninterrupted handover of wireless coordinating nodes and session handover technology based on sequence number marking and shadow buffers; Chinese Invention Patent No. CN115665028A (published January 31, 2023) discloses a method for subnet fusion and subnet splitting in mobile ad hoc networks, which completes subnet fusion by selecting a control node in the subnet, searching for neighboring subnet synchronization signals, and constructing fusion control instructions.

[0004] However, the aforementioned existing technologies still have the following shortcomings: (1) Existing partition merging schemes are mostly based on static rules such as node identifier, partition size or version number for adjudication, without considering the historical evolution of coordination right declarations. For example, simply comparing node identifiers or link quality may lead to frequent invalid switching of coordination nodes, and when a new device joins an existing network, even if the original coordination node is stable, it may be replaced unreasonably because the new device identifier has the advantage in the preset sorting rules, resulting in unnecessary and frequent changes to coordination right declarations in the network.

[0005] (2) Existing technologies fail to effectively distinguish between brief wireless contacts and stable connectivity that can support merged services. Two partitions may only be briefly reachable due to edge members passing each other by momentarily. If merge preparation is triggered every time, it will result in a large number of invalid attempts, increasing network overhead and the risk of service interruption.

[0006] (3) The coordination rights of the original coordinating node are often terminated in advance during the merger preparation stage. If the merger conditions fail before they take effect (such as the departure of the only relay node), the business of the partition will be interrupted and difficult to recover.

[0007] (4) When multiple partitions each have a current voice right holder node, the existing technology lacks a perfect coordination mechanism and cannot complete the consistent switching of voice right status at the common voice boundary without interrupting the existing voice transmission.

[0008] (5) The boundaries of the merger are unclear, and the switching of coordination rights and the switching of voice rights may not be synchronized, which may easily lead to problems such as repeated playback, multiple voice rights being effective at the same time, or long-term silence.

[0009] Therefore, there is an urgent need for a partition merging method suitable for mobile multi-person intercom networks to solve the above problems. Summary of the Invention

[0010] Purpose of the invention The technical problem this invention aims to solve is: in a multi-user intercom network where nodes are constantly moving and wireless links may experience short-term interruptions, when multiple network partitions that can still independently provide intercom services re-establish wireless reachability, how to determine whether the relationship is sufficient to support the transmission of merged control information and voice data, how to prepare new coordinating nodes without interrupting the original partition's voice service, and how to complete a revocable consistent handover while voice frames are continuously transmitted, thereby reducing the probability of repeated playback and voice frame loss and limiting the duration of service interruption.

[0011] Specifically, this includes: distinguishing between short-term cross-domain reachability relationships formed by brief proximity of nodes and cross-domain connectivity relationships that have been continuously confirmed; identifying the original coordination right domain based on the evolution relationship of the coordination right declaration; avoiding premature termination of the original coordination right before the candidate coordination node obtains a complete state; handling situations with multiple current voice right holder nodes and different voice right rounds; ensuring that nodes participating in the merger adopt the same new coordination right declaration at the common voice boundary after meeting the confirmation conditions; and revoking the merger and maintaining the original partition to continue working when cross-domain connectivity fails before the merger takes effect.

[0012] Technical solution To facilitate understanding of this invention, the following terms are defined as follows: Coordination Authority Declaration: A declaration published by a node in the network regarding its authority to act as a coordinating node. It includes evolutionary relationship information and must include at least the coordinating node instance identifier, declaration round, and parent declaration digest. The initial set of parent declaration digests for a coordination authority declaration is empty. Non-merged declarations may contain one parent declaration digest, while merged declarations contain the header declaration digests of the domains to which the coordination authority belongs during the merge.

[0013] Header Declaration: Within the set of valid coordination right declarations currently acquired and verified by the node, there are no other valid coordination right declarations whose parent declaration digest is this declaration digest. When multiple coordination right ownership domains are merged, each domain contributes its current header declaration, and the set of parent declaration digests for candidate new coordination right declarations includes all header declaration digests participating in the merge.

[0014] Coordination authority domain: A logical partition consisting of active nodes that currently recognize the same head declaration.

[0015] Active Nodes: In this invention, an active node is a node that meets the following conditions: (1) Continuously send member status reports within the preset time window; (2) The summary of its report’s claim to coordination authority is consistent with the current header claim of the domain to which it belongs; (3) The link quality parameters of the most recent connection are not lower than the preset active threshold; Nodes that do not meet the above conditions will not be included in the vote counting base for the merger confirmation, nor will they participate in the calculation of the member set in the merger status snapshot.

[0016] Baseline Coordination Authority Declaration Set: The set of header declarations for each coordinating authority domain participating in this round of merging, and the set of parent declarations or parent declaration summaries required to verify the evolution relationship of each header declaration.

[0017] Cross-domain connectivity evidence: A set of observation records used to characterize a bidirectional connectivity relationship between two or more coordination authority domains that meets a preset duration and quality threshold.

[0018] Preparatory Coordination Node: A node selected during the merger preparation phase to collect the status of all parties and generate a merger proposal, but which does not take over coordination authority until the merger takes effect.

[0019] Talking rights: In this application, talking rights refer to the speaking privileges in the intercom group. The node holding the talking rights is the node currently authorized to speak. The talking rights status includes information such as the node holding the talking rights, the talking rights round, and the voice frame sequence number.

[0020] Voice security switching boundary: The specific time or logical event point at which the merging takes effect. Before and after this boundary, nodes complete the switching according to the determined voice reception and transmission rules to reduce the probability of repeated playback, voice frame loss and service interruption.

[0021] Domain switching sequence mapping: Using the domain to which the coordination right belongs and the round of the old voice right as indexes, record the cutoff voice frame sequence number corresponding to each old voice stream at the switching boundary.

[0022] To achieve the above objectives, this invention processes the merging of mobile intercom network partitions as a revocable merging transaction that includes voice service status, rather than a single, instantaneous node election. Specifically, it includes the following steps: Step S1: Obtain multiple valid coordination right declarations within the same connected component, and aggregate the coordination right attribution domains according to the evolution relationship. Each established intercom network partition maintains a coordination right statement containing evolution association information, and the member status report carries a summary of the coordination right statement currently recognized by the node. After receiving multiple valid coordination right statements, the device determines the temporal evolution relationship between the statements based on the coordinating node instance identifier, statement round, and parent statement summary set in the statement, combined with the recognized summary in the member report, and identifies the currently valid header statement of each evolution branch. Before aggregating coordination right home domains, each node first filters active nodes based on the reception of member status reports. Specifically, for each candidate node, if it has not sent a member status report within a preset time window, or if the coordination right statement summary in its report is inconsistent with the current home domain header statement, or if the recent link quality parameter corresponding to the node is lower than a preset active threshold, then the node is marked as an inactive node and not included in the active member set of this home domain. Only active nodes participate in subsequent merge confirmation voting and member set calculation. The set of active nodes that recognize the same valid evolution relationship and belong to the same header statement are aggregated into a single coordination right home domain. Invalid, expired, or source instance mismatched statements are not included in the baseline coordination right statement set. The domain attribution of coordination rights is dynamically formed by the historical partitioning of the mobile network, and is not a pre-configured fixed domain.

[0023] Step S2: Obtain cross-domain connectivity observations at both ends of the cross-domain connected edge, and generate cross-domain connectivity evidence when the connectivity confirmation condition and the preset duration condition are met. When multiple valid coordination rights domains are detected, the device enters the discovery state and collects cross-domain connectivity observation records. Valid cross-domain connectivity evidence is generated when the cross-domain connectivity observations meet the following conditions: at least two endpoints of at least one cross-domain connected edge confirm bidirectional reachability; or at least two endpoints jointly form a bidirectional reachability relationship with different cross-domain connected edges; the continuous observation duration reaches a preset minimum duration threshold; the link quality is not continuously below the failure threshold; and the observation record is within the valid window. Merging admission conditions are not met if only a single broadcast, a one-way low-quality data packet, or an expired report is received.

[0024] Step S3: After the cross-domain connectivity evidence meets the merging admission criteria, based on the currently acquired topology observation information and node capability information, a preliminary coordinating node is selected according to the network topology of the connected components. Each original coordinating node continues to provide intercom services within its domain. Once the cross-domain connectivity evidence is valid, the node enters a preparation state and generates a round identifier for this merge. Nodes participating in the merge select a preliminary coordinating node from the cross-domain connectivity nodes according to a preset deterministic sorting rule. During the merge preparation period, if the current preliminary coordinating node still meets the preset service capability conditions and key topology indicators have not reached the preset degradation threshold, the current preliminary coordinating node is prioritized for maintenance. The preliminary coordinating node receives the member status, topology status, and talk control status of each home domain, generates a unified candidate status and a merge proposal, but must not invalidate the original coordination right declaration before formally acquiring coordination right. Each original coordinating node continues to process talk right requests and member management within its own domain.

[0025] Step S4: The pre-coordination node obtains the member status and voice session status of each home domain. The voice session status includes at least the current talk right holder node, the talk right round identifier, and the voice frame sequence number. Each domain with coordination rights propagates its member set, adjacent edge set, baseline coordination right declaration set, current talk right holder node, talk right round identifier, latest voice frame sequence number of the current voice burst, and summary of pending talk right requests to the pre-coordination node. If multiple domains have current talk right holder nodes, the pre-coordination node generates a unified talk right status according to deterministic rules, considering emergency priority, service priority, pre-configured role priority, and voice burst status in that order. If they are the same, the domain header declaration summary and the speaker node instance identifier determine the status. Voice streams that are not selected do not stop immediately during the preparation phase; the corresponding original domain can continue to transmit to the voice security handover boundary.

[0026] Step S5: The pre-coordinating node generates a merged state snapshot based on the network topology, valid coordination right declaration, cross-domain connectivity evidence, membership status, and voice session status. The pre-coordinating node generates a merge state snapshot summary according to deterministic sorting and filtering rules. This summary includes the intercom group identifier, merge round, normalized member and topology information, baseline coordination right declaration set, speech state summary for each domain, and cross-domain connectivity evidence summary. When changes occur that affect merge consistency, such as cross-domain connectivity evidence, the set of coordination right attribution domains, the candidate coordination node selection results, or key speech session states, the current merge state snapshot becomes invalid. The system increments the merge round identifier and re-collects the corresponding state. All confirmation messages generated based on the old merge state snapshot are deemed invalid by the node and ignored.

[0027] Step S6: Determine the new coordination right declaration and new voice right control status based on the merged state snapshot, and generate a merge effective flag, which is used to determine the voice security handover boundary. The pre-coordination node generates a merge effective marker, which includes at least the merge round identifier, the pre-coordination node identifier, the new coordination right declaration round, the authorized speaking node identifier, the new speaking right round identifier, the domain switching sequence mapping, and the effective time information; The domain switching sequence mapping is used to record the cutoff sequence number of the old voice stream in each original coordination right domain, and the effective boundary identifier is the control cycle number or the voice superframe number. The effective time is used to define when the new coordination right declaration and new voice control state corresponding to the merge effective mark officially replace the original state. The effective time is any one of the following: (1) An absolute time or control superframe number determined based on a common time reference, which is later than the expected propagation completion time of the merged submission certificate and earlier than the difference between the effective deadline of the cross-domain connectivity evidence and the preset clock error protection interval. (2) Logical barrier event, that is, after each coordination right domain completes the sealing of the old media stream corresponding to the domain switching sequence mapping and returns the switching ready certificate, the pre-coordination node broadcasts the final merge activation message, and each node is regarded as having reached the effective time when it receives the message; The voice security handover boundary is determined based on the voice rights status of each home domain: If no voice rights holder exists in any domain, the first satisfying control cycle boundary or voice superframe boundary after the current moment is used as the handover boundary to achieve the shortest handover delay; if one or more home domains have voice rights holders, the system waits for the corresponding voice burst to end normally, determines the cutoff sequence number of the old voice stream in each original coordination right home domain, forms a domain handover sequence mapping, and uses the first satisfying control cycle boundary or voice superframe boundary after all relevant voice bursts have ended as the handover boundary; if there are still voice bursts that have not ended after the preset maximum waiting time, the latest acknowledgment sequence number of each old voice stream up to that moment is determined as the corresponding cutoff sequence number, and the first satisfying control cycle boundary or voice superframe boundary after the preset maximum waiting time expires is used as the handover boundary. Before the handover boundary, the receiving node continues to receive voice frames under the original coordination right, and after the handover boundary, it only receives voice frames that match the new voice rights round identifier.

[0028] Step S7: Obtain confirmation messages from each coordination authority's domain, wherein the confirmation messages reference the same merge status snapshot and the same merge effective flag. The pre-coordination node sends a merge preparation message. Each node sends a home domain confirmation credential when the baseline coordination right declaration set, merge status snapshot summary, cross-domain connectivity evidence summary, pre-coordination node identifier, merge effectiveness flag, and the voice session status of its home domain are all consistent with its own calculation results. Merge submission requires the following conditions to be met: the number of confirmation messages reaches a preset overall threshold, calculated using the total number of active nodes in each home domain as the vote counting base (e.g., more than 50% of the total number of active nodes); each still valid original home domain has a node confirmation; each confirmation references the same merge round, pre-coordination node, and effectiveness flag; cross-domain connectivity evidence continuously meets the admission conditions from preparation to submission; and the current voice session status of each home domain has not changed to prevent handover.

[0029] Step S8: When the voice security handover boundary is reached, enable the nodes participating in the merging to adopt the new coordination right declaration and the new voice right control state. After the confirmation conditions are met, the pre-coordination node sends a merge submission message, and the node enters the pending-effectiveness state after verification. The validity of each state continues to be verified before the effective boundary is reached. At the moment the voice security handover boundary is reached, if the cross-domain connectivity evidence still meets the admission conditions, the handover is executed; otherwise, it is revoked. The state determination at the boundary moment is based on the updated cross-domain connectivity evidence up to that moment, and the cross-domain connectivity observation records within the valid window of that evidence. The handover execution specifically includes: adopting a new coordination right declaration; granting coordination authority to the pre-coordination node; blocking the speech right granting responses of each original coordination node, so that new speech right requests are uniformly routed to the pre-coordination node; updating the speech right round identifier and the authorized speaking nodes; and switching the group status and voice distribution rules.

[0030] Step S9: Before the voice security handover boundary is reached, in response to the expiration of cross-domain connectivity evidence or merged state snapshot, revoke the current merge round and keep all original coordination rights claims valid. The system enters revocation status if any of the following conditions occur before the effective boundary is reached: cross-domain connectivity evidence no longer meets the admission criteria; the baseline coordination right declaration set changes; the preparatory coordination node goes offline or the selection result changes; a home domain withdraws its confirmation or its declaration becomes invalid; the talk right status changes, causing the effective flag to be unsatisfactory; or a submission from the same round but with conflicting content is received. Upon revocation: a new coordination right declaration is not adopted; the original coordination right declaration is not terminated; each original coordination node continues to provide services in its respective partition; the preparatory status for this round is cleared. The pending effective status has a preset validity period; if it does not take effect within the time limit, it is locally revoked and the original coordination right declaration continues to be used. In addition, as a basic startup and access method: when a single machine starts up, it remains in an undeclared coordination right state; when at least two members that have not recognized a valid coordination right declaration form a stable connected component, an initial coordination node election is performed; when a new device joins a network with only one valid header declaration, the declaration is adopted after the authenticity and validity of the header declaration are verified, and a re-election is not triggered.

[0031] Beneficial effects Compared with the prior art, the present invention has at least the following beneficial effects: By using a coordination right ownership domain identification mechanism based on the evolution relationship of coordination right declaration, the coordination right ownership status of each network partition can be identified, reducing unnecessary coordination node switching caused by simply comparing node identifiers.

[0032] The cross-domain connectivity evidence mechanism requires multiple conditions such as bidirectional confirmation, duration, and link quality to distinguish between short-term cross-domain reachability relationships and cross-domain connectivity relationships that meet preset connectivity confirmation conditions, thereby reducing the probability of triggering invalid merges.

[0033] By using a pre-coordination node mechanism, the original coordination rights declaration remains valid before the merger takes effect, ensuring that each network partition can still operate independently when the merger conditions fail, thus reducing the risk of business interruption due to merger failure.

[0034] By using a voice security switching boundary mechanism, the switching of coordination rights and the switching of talk rights status take effect at the same logical boundary, reducing the probability of repeated playback, multiple talk rights being effective at the same time, and abnormal silence lasting too long.

[0035] Through the multi-voice-rights conflict handling mechanism, when there are voice right holders in multiple network partitions, the unified voice right control status after merging is determined, and the cutoff sequence number of each old voice stream is determined separately to form a domain switching sequence mapping, so that existing voice bursts can continue to be transmitted within a preset maximum waiting time, reducing the probability of voice being abruptly cut off.

[0036] The merger confirmation mechanism requires that each still valid coordination authority domain be confirmed by a node, reducing the risk of triggering a network-wide merger if only a single coordination authority domain confirms the merger.

[0037] The revocable merging mechanism cancels the current merging process when the merging conditions fail before the voice security handover boundary is reached, while allowing each original network partition to continue operating, thus reducing the risk of the system entering an inconsistent state.

[0038] The above effects can be verified by metrics such as the number of invalid merge preparations or submissions, the duration of business interruption, the number of repeated voice frames, the number of lost voice frames, and the duration of multiple voice rights being effective simultaneously. Attached Figure Description

[0039] Figure 1A schematic diagram illustrating the topology changes of two mobile intercom network partitions splitting and reconnecting; Figure 2 A data structure diagram for evidence of dynamic coordination rights attribution and cross-domain connectivity; Figure 3 This is a state machine transition diagram of the partition merging method of the present invention; Figure 4 Sequence diagram of the pre-synchronization process for the state of the coordination node; Figure 5 A flowchart for merging the talk rights status when both coordination rights domains have nodes currently holding talk rights; Figure 6 This is a timing diagram for secure voice handover based on the voice handover sequence number; Figure 7 A flowchart illustrating the process by which cross-domain connectivity evidence becomes invalid and triggers revocation before the merger takes effect; Figure 8 A comparison chart showing the differences in redundant evidence between single cross-domain edges and multiple cross-domain edges; Figure 9 A schematic diagram illustrating an implementation of initiation without a valid reconciliation claim and access with a single valid reconciliation claim; Figure 10 This is the main flowchart of the mobile multi-person intercom network partition merging method of the present invention.

[0040] Explanation of reference numerals in the attached figures A, B, C, D, E – Mobile intercom device nodes; D1, D2 — Coordination authority domain (network partition). S1 to S9 – Processing steps of the partition merging method; The state names (idle state, discovery state, ready state, submit state, pending state, effective state, revoked state) correspond to the state machine nodes of the partition merging method of this invention. Detailed Implementation

[0041] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. The following embodiments are for illustrative purposes only and are not intended to limit the scope of the invention.

[0042] I. Detailed Definition of the Declaration of Coordination Authority and the Domain of Coordination Authority In this invention, each established intercom network partition maintains a coordination right declaration. The coordination right declaration includes at least the following fields: coordination right declaration identifier, coordination node identifier, coordination node instance identifier, declaration round, declaration sequence number, declaration validity period, intercom group identifier, parent declaration digest set, and authentication field. The coordination node instance identifier distinguishes different instances of the same node after different startup cycles or configuration changes. The declaration round identifies the generation of the coordination right declaration in the evolution relationship. The declaration round of a normal successor declaration can be obtained by incrementing the declaration round of its parent declaration. Each coordination right declaration forms a coordination right declaration evolution relationship through the parent declaration digest set.

[0043] The member status report carries a summary of the coordination right claims currently recognized by this node. Upon receiving multiple coordination right claims, the device verifies the validity of each claim based on the intercom group identifier, coordination node instance identifier, claim sequence number, claim validity duration, and authentication field. It then determines the ancestor-successor or branch relationship between claims based on the parent claim summary set. Finally, the device defines the set of active nodes currently recognizing the same header claim as a coordination right ownership domain. Coordination right claims that fail authentication verification, exceed their validity duration, fail freshness verification of their claim sequence number, or whose coordination node instance identifier does not match the authenticated identity, are not included in the baseline coordination right claim set.

[0044] If, within the set of valid coordination right claims currently acquired and verified by the device, there are no other valid coordination right claims whose parent statement digest is a certain coordination right claim digest, then that coordination right claim is determined as the header claim. For two valid header claims belonging to the same intercom group, if there is no ancestor-successor relationship between them, the active node sets of the two valid header claims are respectively recognized as different coordination right ownership domains. When merging multiple coordination right ownership domains, the set of parent statement digests for the candidate new coordination right claim includes the digests of all header claims participating in the merger.

[0045] II. Detailed Rules for Generating Cross-Domain Connectivity Evidence When multiple valid coordination rights domains are detected, the device enters the discovery state (see [link]). Figure 3 Collect cross-domain connectivity observation records. Each record includes at least: the observer node identifier, the coordination right declaration identifier recognized by the observer node, the observed node identifier, the coordination right declaration identifier recognized by the observed node, the observation direction, the first and most recent observation times, link quality parameters (such as received signal strength indicator, signal-to-noise ratio, packet loss rate, etc.), and the report sequence number.

[0046] See Figure 8 Valid evidence of cross-domain connectivity is generated when cross-domain connectivity observations meet the following combination of conditions: (1) Single cross-domain connected edge scenario: The two ends of the connected edge report to each other, forming a strict two-way confirmation; (2) Redundant multi-connected edge scenario: There are at least two endpoint pairs with different cross-domain connected edges (see Figure 8 In scenario 2 (link 1 and link 2), the valid reports of each connected edge together cover the two communication directions between the two home domains, thus forming a bidirectional and verifiable cross-domain connectivity relationship; And it meets the following conditions: the continuous observation duration reaches the preset minimum duration threshold; the link quality is not continuously lower than the failure threshold; the observation time of the observation record is within the effective window, and the report sequence number passes the continuity and freshness verification; optionally, it is determined that the two home domains are not in a state of rapid distance from each other based on the link quality change trend.

[0047] It should be noted that if only a single cross-domain connected edge exists and only one-way reports are received (e.g., the first home domain node receives a broadcast from the second home domain node, but the second home domain node does not receive any frames from the first home domain node), even if the duration of this one-way report meets the preset minimum duration threshold, the merging admission condition is not met, and the system remains in the discovery state and does not enter the preparation state (see [link to relevant documentation]). Figure 3 Merge admission criteria are not met if only a single broadcast, one-way low-quality data packet, or expired report is received.

[0048] III. Detailed sorting rules for the election of preliminary coordinating nodes Once the cross-domain connectivity evidence is valid, the system enters a preparation state (see [link]). Figure 3 This generates the identifier for this merging round. Based on the currently acquired topology observation information and node capability information, preliminary coordination nodes are selected, with candidate nodes compared in the following order: Those with a larger number of reachable members will be given priority. Prioritize the path with the smaller maximum number of hops; The path with the smaller total number of hops is given priority. Priority will be given to those with a larger number of neighbors in a single jump; Prioritize the lowest path with the highest quality; Priority will be given to those with higher average link quality; The node identifier is only used as the final determination when all the aforementioned conditions are the same.

[0049] See Figure 1 Given the topology, if a stable cross-domain backbone link is established between device C belonging to domain D1 and device D belonging to domain D2, the system determines candidate devices among the cross-domain connected nodes according to the above sorting rules. For example, if the number of reachable members of device C (including 3 active nodes A, B, and C) is greater than the number of reachable members of device D (including 2 active nodes D and E), then C is preferentially selected as the reserve coordinating node.

[0050] Specifically, during the merge preparation period, if the current candidate coordinating node is still on the network and its reachable members and maximum path hops have not deteriorated beyond the preset thresholds (e.g., the number of members decreases by no more than 20% and the maximum hops increase by no more than 2 hops), the current candidate coordinating node will be maintained first, and a reselection will not be triggered due to other candidate nodes having a slight advantage in the above ranking, so as to avoid frequent oscillations in the merge preparation process.

[0051] The pre-coordinating node receives the membership status, topology status, and talk control status of each home domain, generates a unified candidate state and a merge proposal, but must not invalidate the original coordinating right declaration before formally acquiring coordinating right. Each original coordinating node (such as...) Figure 1 The respective coordination nodes (D1 and D2) continue to process requests for voice rights and member management within their respective domains.

[0052] IV. Detailed method for generating merged status snapshots The device generates a merged state snapshot summary according to deterministic sorting and filtering rules. The sorting rules ensure consistent normalized results when each node computes independently. The snapshot summary is generated as follows: merge_snapshot_digest = H(intercom_group_id, merge_round_id, pre_coordinator_id, normalized_members, normalized_edges, base_claim_set, voice_state_digests, cross_domain_evidence_digest) Wherein, H represents the preset anti-collision hash function (such as SHA-256), merge_snapshot_digest is the merge state snapshot digest, intercom_group_id is the intercom group identifier, merge_round_id is the merge round identifier, pre_coordinator_id is the pre-coordinator node instance identifier, normalized_members is the normalized member set sorted by node identifier, normalized_edges is the normalized adjacency edge set sorted by endpoint identifier, base_claim_set is the normalized representation of the base coordinator claim set, voice_state_digests is the digest set of voice session states for each home domain, and cross_domain_evidence_digest is the digest of cross-domain connectivity evidence. Each field is encoded using a preset field order, length identifier, and normalized encoding method; base_claim_set and voice_state_digests are sorted according to the claim digest and home domain identifier, respectively.

[0053] When changes occur that affect the consistency of the merge, such as cross-domain connectivity evidence, the set of domains to which the coordination authority belongs, the selection results of candidate coordination nodes, or the state of key voice sessions, the current merge state snapshot becomes invalid. The system increments the merge round identifier and re-collects the corresponding state. The node will determine the confirmation message that references the old merge round identifier or the old snapshot summary as invalid.

[0054] V. Detailed Determination Method of Merging Effective Marker and Voice Security Switching Boundary The pre-coordination node generates a merge effective marker, which includes at least the following fields: merge round identifier, pre-coordination node identifier, new coordination right declaration round, authorized speaking node identifier, new speaking right round identifier, domain switching sequence mapping, common effective boundary identifier, merge status snapshot summary, and cross-domain connectivity evidence summary.

[0055] Voice security switching boundaries are determined in the following ways: (1) If there is no node holding the current call right in each domain, the next control cycle boundary or the voice superframe boundary shall be used as the handover boundary to achieve the shortest handover delay. (2) If one or more home domains have a voice right holding node, wait for the corresponding voice burst to end normally, determine the cutoff sequence number of each old voice stream, form a domain switching sequence mapping, and take the first control cycle boundary or voice superframe boundary after the end of all related voice bursts as the common switching boundary. (3) If the waiting time reaches the preset maximum waiting time, the latest confirmation sequence number of each old voice stream up to that moment will be determined as the corresponding cutoff sequence number, and the first control cycle boundary or voice superframe boundary after the preset maximum waiting time expires will be used as the common switching boundary.

[0056] Before the boundary, the receiving node continues to receive voice frames whose sequence number under the original coordination right is not greater than the corresponding voice switching sequence number. After the boundary, it only receives voice frames that match the new voice right round identifier, thereby reducing the probability of voice interruption, repeated playback and voice frame loss.

[0057] VI. Detailed conditions for confirming message collection and merging submission The coordinating node sends a merge preparation message (see...) Figure 10 The message contains information such as the merge round identifier, merge status snapshot summary, cross-domain connectivity evidence summary, pre-coordination node identifier, and merge effective marker.

[0058] A node sends a home domain confirmation credential when all of the following conditions are met (see [link to relevant documentation]). Figure 10The baseline coordination right declaration set is consistent with the local calculation; the merged state snapshot summary is consistent with the local calculation result; the cross-domain connectivity evidence summary is consistent with the cross-domain connectivity observation record received and verified by the local machine; the pre-coordination node identifier and merge effective flag are consistent with the local verification result; and the voice session status of this domain is consistent with the status in the snapshot.

[0059] The merge commit must meet at least all of the following conditions: The number of confirmed messages reaches a preset overall threshold (e.g., more than 50% of the total number of active nodes). The number of confirmation nodes for each still valid home domain corresponding to the benchmark coordination right declaration set reaches a preset domain confirmation threshold, which is, for example, 30% of the number of valid nodes in the domain and not less than one node. When the number of active nodes in a certain domain is less than the preset minimum number of acknowledgments (e.g., less than 3), all active nodes in that domain must send acknowledgments in order to be considered to meet the acknowledgment conditions for that domain. Each confirmation references the same merge round, pre-coordination node, and effective flag; The cross-domain connectivity evidence continuously meets the admission criteria from preparation to submission; There are no changes in the current voice session status of each home domain that would prevent handover (such as an unexpected change in the turn order).

[0060] VII. Detailed Conditions and Procedures for Merging and Cancelling The revocation state is entered when any of the following conditions occur before the effective boundary is reached (see [link]). Figure 3 ): Cross-domain connectivity evidence no longer meets the admission criteria (see Figure 7 (The departure of the sole relay node caused the link quality to continuously fall below the failure threshold for more than a preset duration). The baseline coordination right declaration set changes (e.g., a coordinating node is re-elected in a certain domain). The pre-coordination node is offline or its key topology indicators (such as the number of reachable members and the maximum number of path hops) deteriorate beyond a preset threshold; A domain has withdrawn its confirmation or its coordination authority declaration has become invalid. The change in the talk right status makes it impossible to satisfy the already generated merge effective flag (e.g., a new urgent talk right request appears). Received merge commit messages from the same merge round but with conflicting content.

[0061] When revoking, perform the following actions: do not adopt a new coordination right declaration; do not terminate the original coordination right declaration; each original coordination node continues to provide services for this partition; clear the current round of preparation state and merge state snapshots; optionally set a cooldown time (e.g., 30 seconds, calculated from the time of revocation), during which time it will not respond to new merge triggers between the same pair of home domains to avoid repeated invalid attempts.

[0062] Pending effect status (see) Figure 3 It has a preset validity period, which is no earlier than the common validity boundary specified by the merge validity mark, and includes a preset protection duration; if it still does not take effect after the validity period expires, it will be revoked locally and the original coordination right declaration will continue to be used.

[0063] VIII. Detailed Description of Each Embodiment Example 1: Brief brushing shoulders without merging See Figure 1 and Figure 2 The coordination right home domain D1 includes devices A, B, and C (where C is an edge node of D1), and D2 includes devices D and E (where D is an edge node of D2). Devices C and D briefly approach each other at an intersection. C only receives two control frames from D, D does not receive a valid report from C, and the RSSI drops rapidly. Although the system detects another coordination right claim, the cross-domain connectivity observation does not meet the connectivity confirmation condition and the minimum duration threshold. Both home domains retain their original coordination nodes, do not select a backup coordination node, and do not change the voice service status. See also Figure 3 The state machine shown indicates that the system remains in the discovery state and does not enter the preparation state.

[0064] Example 2: After a single cross-domain connected edge stabilizes, the process enters the preparation stage. See Figure 4 D1 and D2 rejoin, and device C and device D form a unique cross-domain connected edge (see [link]). Figure 1 (Regarding the "stable cross-domain backbone link"). Devices C and D report to each other that the connection duration has exceeded the preset minimum duration threshold and the quality has not continuously fallen below the failure threshold. The system generates a merge round identifier and a baseline coordination right declaration set, and selects device C as the reserve coordination node according to the election sorting rules. The original coordination nodes D1 and D2 continue to maintain their respective services, and device C begins synchronizing the membership status and talk control status of the two domains. Figure 3 The state machine shown transitions the system from the discovery state to the preparation state.

[0065] Example 3: Both partitions have nodes currently holding the right to speak. See Figure 5During the merge preparation period, device A in D1 is speaking (the node holding the right to speak is A, the current round of speaking is 7, and the confirmed last voice sequence number is 420), and device E in D2 is speaking (the node holding the right to speak is E, the current round of speaking is 12, and the confirmed last voice sequence number is 188). The preparatory coordination node determines the final decision based on emergency priority, right to speak priority, voice burst status, time of right to speak acquisition, and final judgment criteria (see [link to relevant documentation]). Figure 5 (The judgment process) selects device A as the merged authorized speaking node. The pre-coordination node generates a new speaking round of 13 based on the current speaking rounds, determines the cutoff sequence number of the old voice stream in D1 as 428, and the cutoff sequence number of the old voice stream in D2 as 192, forms a domain switching sequence mapping, and determines the first control cycle or voice superframe boundary after the end of the two voice bursts as the common effective boundary. The two domains confirm the same merge effective marker. See Figure 6 Before the common effective boundary is reached, D2 continues to play E's voice frame up to sequence number 192; after the common effective boundary is reached, the node only receives the voice frame of the right-of-way round 13 and the node that is allowed to speak is device A.

[0066] Example 4: Chain Disruption During Merger Preparation See Figure 7 The system has selected a preliminary coordination node, but the unique cross-domain connectivity edge between device C and device D expires before the common effective boundary is reached (see branch 7, "Evidence Expiration, such as the departure of the unique relay"). The node sends or generates a merge revocation message locally; the new coordination right does not take effect, and D1 and D2 retain their original coordination node and talk rights status. Since the original coordination right was not terminated prematurely, there is no need to rebuild the original coordination state. Figure 3 The state machine shown indicates that the system has entered the cancel state.

[0067] Example 5: Redundant Evidence for Multiple Cross-Domain Connected Edges See Figure 8 In scenario two, two home domains form two cross-domain connectivity edges: one between device C and device D (link 1), and the other between device B and device E (link 2). The system generates cross-domain connectivity evidence based on these two independent connectivity edges. Even if link 1 (CD) fails temporarily, the merging process can continue as long as link 2 (BE) alone meets the strict bidirectional confirmation and duration requirements, and the link quality meets the admission criteria.

[0068] Example 6: Fast handover when there is no node holding the current talk rights (achieving the shortest handover latency) Both home domains are silent, and the voice session state summary indicates that there is no speaker. The system writes the identifier of the next control cycle or voice superframe boundary into the merge effective flag, without waiting for the voice burst to end before entering the confirmation and submission process (see [link]). Figure 10 (The "No-voice holder" branch in China).

[0069] Example 7: Normal Startup and Access See Figure 9 When a single node starts up, it remains in an undeclared coordination right state. When at least two devices that have not recognized a valid coordination right declaration form a connected component that meets the preset stable connectivity conditions, an initial coordination node election is performed. When a new device joins a network with only one valid coordination right declaration, the declaration is adopted after the coordination right declaration is verified to be valid, without triggering the full partition merging process.

[0070] Example 8: Device Implementation Example Mobile multi-person intercom equipment includes the following components: The wireless communication interface is used to send and receive network control frames and voice frames, and provides link quality parameters such as received signal strength indicators; The processor executes the aforementioned steps, maintains member states, network topology, and coordination authority domains, and runs the cross-domain connectivity evidence generation and merging state machine. The processor updates voice reception conditions and voice control state before and after the effective boundary, achieving synchronous switching between the control plane and data plane. The audio acquisition and playback module is used to generate and receive voice frames with source identifiers, turn order, and sequence numbers; The time reference module is used to provide the control cycle number, voice superframe number, or synchronized network time reference, and to determine the effective boundary of the merge based on this. The memory is used to store the coordination rights declaration, merge status snapshot, effective flag, pending effective status, and duplicate frame detection window.

[0071] IX. Industrial Applicability This invention is applicable to scenarios requiring real-time voice communication among multiple people and with highly dynamic network topology, such as mobile self-organizing intercom systems, cycling communication, outdoor exploration, and emergency command. It can be implemented by software, hardware, or a combination of both, and has broad industrial application prospects.

Claims

1. A method for merging partitions in a mobile multi-person intercom network, executed by a pre-coordination device, characterized in that: include: Obtain the header declaration and domain state of each of the multiple coordination right attribution domains in the same intercom group, wherein the header declaration includes a set of parent declaration summaries, the multiple coordination right attribution domains correspond to valid header declarations that do not have an ancestor-successor relationship with each other, and the domain state includes the talk status of each coordination right attribution domain; Obtain cross-domain connectivity evidence, which is used to prove that the domain-level directed graph with the multiple coordination right attribution domains as vertices satisfies bidirectional reachability between the multiple coordination right attribution domains, and has a valid evidence deadline. The header declarations, domain states, and cross-domain connectivity evidence of the multiple coordination rights attribution domains are normalized and encoded to generate a merged state snapshot; Based on the merged state snapshot, candidate new coordination right declarations, new voice right states, and domain switching sequence mappings are generated. The parent declaration summary set of the candidate new coordination right declarations includes the header declaration summaries of each of the multiple coordination right ownership domains. The domain switching sequence mapping records the old voice right round identifier, the speaking node instance identifier, the voice burst identifier, and the termination sequence number in the domain sequence space for each coordination right ownership domain with an old voice burst. Send a merge preparation message and obtain confirmation credentials covering the multiple coordination rights ownership domains respectively. Each ownership domain confirmation credential references the same merge status snapshot, the same candidate new coordination rights declaration, the same new talk rights status, the same domain handover sequence mapping, and the same merge effective event. When the multiple coordination rights domains are all covered by the domain confirmation certificate, and the cross-domain connectivity evidence remains valid from the time the merge preparation message is sent to the time the merge submission certificate is generated, a merge submission certificate is generated and sent. The merge submission certificate is bound to the merge status snapshot, the domain confirmation certificate, the candidate new coordination rights declaration, the new talk rights status, the domain switching sequence mapping, and the merge effective event. The merge submission certificate is used to ensure that the nodes participating in the merge maintain the validity of their original coordination rights claims before the merge effective event, and when the merge effective event occurs, the nodes adopt the candidate new coordination rights claims and the new talk rights status to stop accepting old media streams from the domains to which each coordination right belongs, according to the domain switching sequence mapping. Prior to the occurrence of the merge effective event, in response to the expiration of the cross-domain connectivity evidence, the merge status snapshot, or the merge effective event, the current merge round is revoked and all original coordination rights claims remain valid.

2. The method according to claim 1, characterized in that, The header declaration also includes the intercom group identifier, coordinating node identifier, coordinating node instance identifier, coordinating authority declaration round, declaration sequence number, validity period, and authentication field; The set of parent statement summaries for a normal coordinating right succession statement includes a parent statement summary, and the set of parent statement summaries for a candidate new coordinating right statement includes all header statement summaries involved in the merging; When two valid header declarations share a common ancestor and neither can reach the other along a parent declaration relationship, the two valid header declarations are determined to be different coordination domains.

3. The method according to claim 1, characterized in that, The valid header declaration, cross-domain connectivity observation, home domain confirmation certificate, and merged submission certificate each include a digital signature or message authentication code, and include the sending node instance identifier, message sequence number, and validity period; The receiving node determines its validity based on the verification results of the anti-replay window, instance matching, validity period, and digital signature or message authentication code.

4. The method according to claim 1, characterized in that, The cross-domain connectivity evidence is generated based on a request-response message with a probe random number and includes a directed path proof constituting the bidirectional reachability, a set of observation record summaries, a continuous satisfaction time, a packet loss rate or probe success rate, and an evidence validity deadline; received signal strength is not used alone as proof of bidirectional reachability.

5. The method according to claim 4, characterized in that, When the cross-domain connectivity evidence is configured as redundant evidence, the domain-level directed graph has at least two bidirectional cross-domain edges whose endpoints are not shared, or at least a pair of directed paths that cover two communication directions respectively and whose internal nodes do not intersect.

6. The method according to claim 1, characterized in that, The preliminary coordination device is a candidate device determined by the participating nodes according to a deterministic ranking tuple. The deterministic ranking tuple includes at least a deterministic ranking tuple comparing candidate devices based on the number of reachable active members, the maximum number of path hops, the path cost, and the node identifier. If the topology index change of the current pre-coordination device does not reach the preset replacement threshold, the current pre-coordination device is maintained. After the preset replacement threshold is reached and the change continues for a preset time, a new merging round is created and a new pre-coordination device is selected.

7. The method according to claim 1, characterized in that, The standardized encoding specifies character encoding, field length or length prefix, numerical unit, set deduplication and sorting, empty field representation, and protocol version; The merge status snapshot is also bound to the participating domain header declaration set, active member set, adjacency relationship, cross-domain connectivity evidence digest and evidence validity deadline, pre-coordination device instance and pending talk request digest for each domain; confirmation messages referencing old merge rounds or different merge status snapshots are not used to generate the merge commit certificate.

8. The method according to claim 1, characterized in that, When determining the new talk right status, a cross-domain comparison is performed according to the emergency level, business priority and pre-configured role priority uniformly defined within the intercom group, and when the priorities are the same, a deterministic determination is made according to the domain header declaration summary and the speaking node instance identifier; The local call round identifier and uncalibrated call acquisition time of each coordination right domain are not used for cross-domain numerical sorting.

9. The method according to claim 1, characterized in that, The termination sequence number in the domain switching sequence mapping is determined according to any of the following methods: The last sent sequence number is determined by the message sent by the speaking node to release the right of speech or to end the voice. For selected speaking nodes that continue to speak after the merger, the plan termination sequence number is determined by the generational grant message issued by the original coordinating node; Alternatively, after the maximum waiting time expires, the original coordinating node may determine the number of frames to be protected based on the latest confirmed sequence number, the maximum number of frames in transit, and the number of protected frames. In this process, the termination sequence numbers of different coordination authority domains are not compared with each other to form a unified maximum sequence number.

10. The method according to claim 9, characterized in that, The selected speaking node sends media frames using the old speaking round identifier before the termination sequence number corresponding to its coordination right domain, and sends subsequent media frames using the new speaking round identifier in the new speaking right state after the merger takes effect event. Unselected speaking nodes stop sending media frames with the old speaking round identifier after explicit release of speaking rights, explicit end of voice, or forced end of timeout; voice activity detection results are used to assist in the determination of silence timeout or abnormal exit.

11. The method according to claim 1, characterized in that, The merger taking effect event is any one of the following: The control superframe or effective time is represented by a common time reference with an upper bound on the error, the effective time being later than the expected propagation completion time of the merged submission certificate and earlier than the evidence validity deadline minus the clock error protection interval. Alternatively, after each coordination authority domain completes the sealing of the old media stream corresponding to the domain switching sequence mapping and returns the switching ready certificate, a logical barrier event is received carrying the final merge activation message with the merge submission certificate.

12. The method according to claim 1, characterized in that, Each home domain confirmation credential is jointly generated by the current coordinating node and at least one different node in the corresponding coordinating domain, or generated by the aggregation of node signature shares that have reached a preset confirmation threshold within the corresponding coordinating domain; the merged submission certificate includes a summary of the home domain confirmation credential set covering all participating domain header declarations, certificate serial number, and authentication field.

13. The method according to claim 1, characterized in that, Before the merger takes effect event occurs, in response to changes in the set of participating domain header declarations, invalidation of cross-domain connectivity evidence or insufficient remaining validity period, changes in the pre-coordination equipment, invalidation of the domain switching sequence mapping due to the talk power status, or different snapshot digests appearing in the same merger round, a merger revocation message is sent and the pre-state of the current round is cleared. After the merge submission certificate is generated, the current merge is not revoked based on the local link quality report of a single node; before the common effective boundary is reached, it is only revoked when the updated cross-domain connectivity evidence as a whole no longer meets the admission conditions.

14. The method according to claim 1, characterized in that, The plurality of coordination rights attribution domains are three or more coordination rights attribution domains, the domain-level directed graph is strongly connected on the subgraph formed by the three or more coordination rights attribution domains, the set of parent declaration digests of the candidate new coordination rights declarations covers all header declarations of the three or more coordination rights attribution domains, and the set of attribution domain confirmation credentials respectively contains attribution domain confirmation credentials corresponding to all the header declarations.

15. A method for merging mobile multi-person intercom network partitions, executed by mobile multi-person intercom devices participating in the merging, characterized in that, include: Receive a merge preparation message, which includes merge round, merge status snapshot, cross-domain connectivity evidence, candidate new coordination right declaration, new voice right status, domain switching sequence mapping and merge effective event. The domain switching sequence mapping records the termination sequence number in the domain sequence space for each coordination right ownership domain with old voice bursts. Verify that the merged state snapshot contains the header claims recognized by this device and the talk status of the coordination right domain to which this device belongs; verify that the cross-domain connectivity evidence is used to prove bidirectional reachability between the coordination right domains participating in the merge; and verify that the set of parent claims digests of the candidate new coordination right claims covers all header claims participating in the merge. In response to successful verification, a home domain confirmation credential referencing the merge round, the merge status snapshot, the candidate new coordination right declaration, the new talk right status, the domain switching sequence mapping, and the merge effective event is sent. Receive and verify the merge submission certificate, which is bound to all domain confirmation credentials for all coordinating rights involved in the merge, the merge status snapshot, the candidate new coordinating rights declaration, the new talk rights status, the domain switching sequence mapping, and the merge effective event; Before the merger takes effect, the original coordination right declaration of the coordination right domain to which this device belongs continues to be recognized, and control messages and media frames are processed in accordance with the original coordination right declaration and the original voice right status. The old media frames that match the old voice right round identifier, speaking node instance identifier and voice burst identifier of the corresponding domain in the domain switching sequence mapping, and whose voice frame sequence number is not greater than the corresponding termination sequence number in the domain switching sequence mapping are received. When the merge takes effect, the candidate new coordination right declaration and the new talk right status are adopted, and after the merge takes effect, the control messages of the old coordination right declaration round and the media frames of the old talk right round identifier are rejected; Specifically, before the merge submission certificate verification is passed, the local preparatory state is revoked in response to the merge state snapshot or the cross-domain connectivity evidence becoming invalid. After the merge submission certificate is verified, the merge submission certificate will not be automatically revoked due to the local link quality report of this device.

16. A mobile multi-person intercom network partition merging system, characterized in that, include: Multiple mobile intercom devices, including a pre-coordination device and a merging device; The pre-coordination device is used to perform the method as described in any one of claims 1 to 14; The participating merging device is used to perform the method as described in claim 15.

17. A computer-readable storage medium, characterized in that, The device contains a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 15.

Images