A method for orchestrator convergence based on declarative leases, topology snapshots, and consistent commit in a mobile wireless group network

Abstract

A kind of mobile wireless group network based on declaration lease, topology snapshot and consistent convergence method of submitter of lease this application discloses a kind of convergence method of submitter in mobile wireless group network.Communication equipment determines the valid submitter declaration set in current communication component according to authentication result, node instance identifier, declaration serial number, declaration round and lease;When there is no valid declaration and active member is not less than two, or there are multiple valid declarations, node and edge observation of the same round are collected and normalized to form topology snapshot and topology abstract, and candidate submitter is determined according to the topology snapshot and topology abstract.Only when topology abstract, candidate result and original declaration set remain consistent within a stable window, node confirms the consistency proposal;After confirming set reaches quantity threshold, and in the multi-declaration scene covers each original declaration domain, submission message carrying submission certificate, new declaration round and lease is generated.Receiving device adopts the declaration after verifying certificate, abstract and round fence;Only one valid declaration is directly adopted.The submitter jitter caused by old message, local topology and short-time link fluctuation can be inhibited.

Description

Technical Field

[0001] This invention relates to the fields of mobile ad hoc networks, wireless group communication, distributed coordination, topology-aware role selection, and fault recovery. In particular, it relates to a method and device for converging the coordinator state within the current communication component under conditions of device startup, member access, physical partitioning, partition merging, message delay, and link fluctuation, through time-limited coordinator declarations, normalized topology snapshots, stability windows, and consistency commit certificates.

[0002] This invention can be used in multi-person intercom smart helmets, outdoor group communication terminals, mobile robots, field collaboration terminals, unmanned system formations, and other mobile wireless groups lacking fixed infrastructure. Background Technology

[0003] The location of nodes, obstruction relationships, and link quality in a mobile radio group are constantly changing. A single node may start before other nodes, and multiple nodes may start simultaneously within a short period of time; an existing group may split into multiple temporarily unreachable physical partitions due to distance or obstruction, and then reconnect; during the above process, messages may also be delayed, duplicated, or replayed after a node restarts.

[0004] Existing implementations can select a coordinating node within a group through heartbeat timeout, node identifier comparison, link quality comparison, topology centrality scoring, or general voting. However, forming a coordinating declaration immediately upon single-machine startup can easily lead to newly joined nodes simultaneously observing multiple short-lived declarations; switching coordinating nodes solely based on node identifier, single link quality, or a larger version number can easily cause newly joined nodes or transient messages to trigger meaningless preemption; and when each node calculates candidates based only on its own transient adjacency graph, it may also announce different results.

[0005] Even with leases, terms, or voting, old instance messages may still overwrite the current state late, unstable topologies may still lead to repeated switching, and when two separately running declaration domains are re-merged, simply reaching the general number of confirmations does not prove that every conflicting declaration domain has seen and accepted the same result. Therefore, a convergence mechanism is needed that links declaration validity, common topology inputs, stable commit conditions, and conflicting declaration domain witness relationships. Summary of the Invention

[0006] The technical problem this invention aims to solve is: in a mobile wireless group network without a fixed central node, how to avoid the jitter caused by local topology or instantaneous scoring in conditions such as single-machine startup, simultaneous startup of multiple nodes, new node access, physical partitioning, partition merging, old message arrival, and short-term link fluctuations, and to form verifiable coordinator submission results within the current stable communication components.

[0007] To address the aforementioned technical problems, this invention expresses currently recognized coordinators as coordinator declarations with declared leases. The communication device determines the validity of a declaration based on the authentication result, coordinator instance identifier, declaration sequence number, declaration round, declaration lease, and current reachability, and maintains a declaration round fence for the adopted submission chain. The declaration round is used to isolate older submitted results, rather than directly as the ranking priority for candidate coordinators.

[0008] The communication equipment counts the number of different valid coordinator declarations within the current communication component. When there are no valid declarations and only one active node, the system remains undeclared; when there are no valid declarations and at least two active nodes, the system enters the initial generation process; when there is only one valid declaration, it directly adopts that declaration; when there are multiple different valid declarations, the system enters the conflict convergence process.

[0009] In the initial generation or conflict convergence process, nodes collect fresh and validated node and edge observations from the same snapshot round, normalize, sort, and filter them, and calculate a topology summary along with the original declaration set and snapshot parameters. Each node uses the same topology summary and predetermined candidate sorting rules to determine a candidate coordinator, and only sends or confirms a consensus proposal after the topology summary, candidate results, original declaration set, and key edge set have remained consistent for a stable window.

[0010] The submitter collects confirmations for the same topology summary, candidate coordinator, and next claim round. In the initial generation scenario, confirmations reach the initial generation threshold; in the conflict convergence scenario, confirmations not only reach the quantity threshold but also include at least one confirmation for each valid claim in the original claim set's corresponding claim domain. A coordinator submit message is then generated, containing a submit certificate, topology summary, next claim round, and claim lease. The receiving device verifies the submit certificate, topology summary, and claim round fence before adopting the new coordinator claim.

[0011] Compared to schemes relying solely on heartbeats, node identifiers, single topology scoring, or general voting, this invention offers at least the following advantages: single-machine operations do not generate short-lived claims; a single valid claim can be inherited by a new node during the lease period; instance identifiers, claim sequence numbers, claim leases, and claim round fences collectively suppress replay and expired messages; normalized topology snapshots provide comparable inputs for candidate computation; a stability window makes topology stability a prerequisite for submission; claim domain coverage confirmation allows all original claim domains to jointly witness the merge result; and submission certificates enable receiving nodes that did not directly participate in the proposal to verify the basis for adoption. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the functional structure of a communication device according to an embodiment of the present invention.

[0013] Figure 2 This is a schematic diagram of the coordinator's state transition according to an embodiment of the present invention.

[0014] Figure 3 This is a schematic diagram of the coordinator declaration field and validity fence according to an embodiment of the present invention.

[0015] Figure 4 This is a schematic diagram of the topology snapshot formation and candidate calculation process according to an embodiment of the present invention.

[0016] Figure 5 This is a schematic diagram illustrating the timing of the stable window, consensus proposal, confirmation, and coordinator submission in an embodiment of the present invention.

[0017] Figure 6 This is a schematic diagram illustrating the re-merging and overriding confirmation of two original declaration fields according to an embodiment of the present invention.

[0018] Figure 7 This is a schematic diagram illustrating the topology change reset stabilization window and the exception handling of old submissions being filtered, according to an embodiment of the present invention. Detailed Implementation

[0019] The present invention will be further described below with reference to the accompanying drawings and embodiments. It should be understood that the following embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention. Where there is no conflict, the various embodiments and their technical features can be combined with each other.

[0020] like Figure 1 As shown, the communication device 100 may include a wireless transceiver 170, an observation and maintenance module 110, a lifecycle declaration module 120, a snapshot round module 130, a candidate calculation module 140, a consistency commit module 150, a status output module 160, and a processor and a memory 180. These modules can be implemented by the processor executing programs in the memory, or by a combination of software, firmware, programmable logic, or dedicated hardware. The wireless transceiver 170 can transmit group messages via broadcast, multicast, forwarding, or on-demand unicast; this invention does not limit specific wireless protocols or chips.

[0021] The observation and maintenance module 110 maintains the node identifier, node instance identifier, active status, one-hop neighbor or reachability observation, observation time, link quality, and observation source for nodes in the same group. The node instance identifier is used to distinguish messages from the same node in different startup cycles, key rotation cycles, or identity instances. The current communication component can be formed based on time-fresh and verified neighbor or reachability observations and includes the local machine.

[0022] The lifecycle claim module 120 maintains the coordinator claim. The coordinator claim may include the group identifier (group_id), coordinator identifier (coordinator_id), coordinator instance identifier (coordinator_instance_id), claim epoch, claim sequence number (claim_seq), claim lease duration or expiration time, snapshot round identifier (snapshot_round_id), topology digest (topology_digest), candidate score digest (candidate_score_digest), submission certificate digest, and authentication tag. The claim identifier (claim_id) can be jointly determined by the coordinator identifier, coordinator instance identifier, and claim epoch.

[0023] like Figure 3 As shown, the receiving device determines a coordinator's claim as valid when the group identifier and authentication information are valid, the coordinator instance identifier is consistent with the current instance, the claim sequence number is a fresh sequence number relative to the same claim identifier, the claim lease has not expired, the coordinator is still within the current communication component, the first claim or post-conflict claim has a verifiable submission certificate, and the claim round is not lower than the local claim round fence. The coordinator can periodically send renewal messages that keep the claim identifier unchanged and increment the claim sequence number. If the lease expires and no fresh renewal is received within the tolerance window, the receiving device removes the valid claim.

[0024] The declaration round is used to filter late submissions, not to determine the coordinator candidate alone. In other words, a larger declaration round, a smaller node identifier, or a stronger link does not automatically grant a node preemption rights. Candidate nodes are determined by a predetermined ordering rule on a shared topology snapshot.

[0025] like Figure 2 As shown, the device can periodically or when observing changes, count the number C of different valid declaration identifiers and the number K of active nodes within the current communication component. When C equals 0 and K equals 1, the device enters or remains in the undeclared state UNCLAIMED; when C equals 0 and K is not less than 2, the device enters the initial generation waiting state BOOTSTRAP_PENDING; when C equals 1, the device enters the adopted state ADOPTED; when C is greater than 1, the device enters the conflict convergence waiting state CONVERGENCE_PENDING.

[0026] In the adoption state, subsequent nodes write a unique valid claim to their local state. As long as the lease and commit certificate of this claim remain valid and the coordinator is still in the current communication component, the addition of ordinary members, fluctuations in the quality of a single link, or changes in the node identifier order do not trigger candidate comparisons. This process avoids the group repeatedly re-electing a coordinator during regular expansion.

[0027] When entering the initial generation waiting state or the conflict convergence waiting state, the snapshot round module 130 sends a snapshot collection message with a snapshot round identifier and receives node reports within the collection time T_collect. The node report may include the reporting node identifier and its instance identifier, observation time, active observations of the node, side observations, the current claim identifier of the node, and authentication tag. Side observations may include two endpoints, link quality, side observation time, and evidence type.

[0028] The device only adds nodes and edges that satisfy the edge freshness (T_edge), authentication, and evidence rules to the topology snapshot. For undirected edges, the two endpoints can be required to report mutual reachability, or a preset trusted observation rule can be used. The device sorts nodes by node identifier and node instance identifier, and edges by endpoint and evidence version, filtering duplicate nodes, duplicate edges, and expired edges. The set of valid claim identifiers observed at the start of a conflict convergence round constitutes the original claim set, base_claim_set.

[0029] like Figure 4 As shown, the normalized set of nodes, edges, original claims, and snapshot parameters together constitute the topology summary input. The topology summary can be represented as H(group_id, snapshot_round_id, normalized_nodes, normalized_edges, base_claim_set, snapshot_parameters), where H represents the integrity summary function. The topology summary changes accordingly if any object in the summary input changes. Different nodes can only compute the same topology summary if they receive the same normalized input.

[0030] The candidate calculation module 140 calculates candidate tuples for each active node in the topology snapshot. In one embodiment, candidate tuples are compared in the following order: a larger number of reachable nodes is preferred; a smaller maximum hop count is preferred; a smaller total hop count is preferred; a larger direct connectivity is preferred; a larger minimum path quality is preferred; a larger average path quality is preferred; when all the aforementioned indicators are the same, the node with the smaller identifier is used for deterministic elimination of ties. The candidate calculation module can also calculate candidate score summaries for candidate tuples or scoring results. Other sorting rules that can produce unique and deterministic results on the same snapshot can also be used.

[0031] The stability window T_stable is not a waiting time after submission, but rather a hard condition that allows submission or confirmation of a commit. The device records the snapshot round identifier, topology summary, original declaration set, candidate coordinator, candidate score summary, and key node edge set as continuous over a period of time; if any of these changes, the stability window is reset. Therefore, the system will not form a committable result based on transient topology when there is link obstruction, high-speed node movement, or incomplete edge observation propagation.

[0032] like Figure 5 As shown, after the stability window is satisfied, the proposing node sends a consensus proposal. The consensus proposal includes at least the snapshot round identifier, topology summary, original declaration set or its summary, candidate coordinator, candidate score summary, next declaration round, proposing node identifier, and authentication information. The next declaration round can be the largest declaration round in the original declaration set plus one, or it can be generated by other preset monotonic rules.

[0033] Each node receiving a consensus proposal independently performs snapshot normalization and candidate computation. A consensus confirmation is sent only if the node obtains the same topology summary, the same set of original declarations, the same candidate coordinator, and the same candidate score summary, and the node's stability window has also been satisfied. Confirmation may include the snapshot round identifier, topology summary, candidate coordinator, next declaration round, confirming node identifier, confirming node instance identifier, the declaration identifier currently used by the confirming node, confirmation time, and authentication tag. If the computation results are inconsistent, the node can send a snapshot mismatch notification or wait for the next round of snapshot collection.

[0034] The consistency commit module 150 collects confirmations of the same snapshot round identifier, topology summary, candidate coordinator, and next declaration round. In the initial generation scenario, the original declaration set is empty, and the number of confirmations needs to reach the initial generation threshold Q_bootstrap. Small groups can require coverage of all active nodes in the snapshot, while larger groups can use the participation threshold in conjunction with member failure rules.

[0035] In conflict convergence scenarios, the submission conditions include a threshold Q for the number of acknowledgments, and for each claim identifier in the original claim set, the acknowledgment set must include at least one acknowledgment from an active node currently using that claim identifier. The set of active nodes carrying the same claim identifier is called a claim domain. The above claim domain coverage condition ensures that each conflicting claim domain provides witness to the same topology summary and candidate coordinator, preventing a single claim domain from unilaterally announcing the merge result.

[0036] When the commit conditions are met and the stable window is not interrupted, the commit initiator generates a coordinator commit message. This message may include the group identifier, the coordinator identifier as a candidate coordinator, the coordinator instance identifier, the new declaration round, the declaration lease, the snapshot round identifier, the topology summary, the candidate score summary, the original declaration set summary, the commit certificate or the commit certificate summary, and the authentication tag. The commit certificate consists of the identity, instance, original declaration ownership, confirmation summary, and authentication information of each confirming node.

[0037] When a submitted certificate cannot fit into a single message, it can be divided into certificate pages with page numbers, total page count, and certificate digests. The receiving device will only proceed with verification and coordinator-submitted messages after collecting all certificate pages and verifying the certificate digests. This pagination method does not change the requirement that all acknowledgments must reference the same proposal content.

[0038] If the receiving device has the following conditions: the group and authentication information are correct; the new claim round is higher than or equal to the current claim round fence and the submitted content is exactly the same; the topology summary and candidate score summary are consistent with the round in which the receiving device participated; the number of submitted certificates reaches the threshold; the conflict scenario satisfies the claim domain coverage; and the candidate coordinator is an active node in the submission snapshot and the submission has not timed out, then the coordinator submits the message. After adoption, the candidate coordinator sends a renewal message with an incrementing claim sequence number, and other nodes become follower nodes or assume corresponding business roles.

[0039] If the receiving device receives a commit below the declared round fence, a commit from an old node instance, or a commit with an expired lease, it maintains the current verified state. If two commits with different content appear in the same declared round, the device does not directly overwrite them with the first or last message, but marks the commits as conflicting and re-enters the topology snapshot collection round.

[0040] Physically inaccessible partitions cannot share real-time state; each partition can maintain or generate valid claims within its own current communication component. This invention does not require any globally unique coordinator among any physical partitions at any given time; convergence is committed only when multiple partitions re-form the same communication component, through a shared topology snapshot, a stable window, and a claim domain overlay.

[0041] In the smart helmet group intercom embodiment, when device A starts up alone, the current communication component only contains A and there is no valid declaration, so A remains in an undeclared state. After device B joins, A and B enter the first generation round, form the same topology summary and calculate candidate A; after the stability window is satisfied, A proposes a proposal, B confirms it, and after the first generation threshold is reached, a declaration identifier (A, instance_A, epoch_1) is generated, and the two devices maintain the declaration through renewal messages.

[0042] In the new node addition example, devices A and B have already adopted a valid claim with device B as the coordinator. After device C joins, only one valid claim is observed in its current communication component, so this claim is adopted directly. Even if device C has a smaller node identifier or a higher quality link in one of its links, no candidate comparison is triggered.

[0043] like Figure 6As shown, the original group was physically partitioned, forming claim domains D1 and D2. After re-contact, each node observed two valid claims, which together constituted the original claim set. Based on the same topology snapshot, candidate node C was obtained. Nodes from D1 and D2 each sent acknowledgments, and the total number of acknowledgments reached a threshold. The commit certificate thus simultaneously proved that D1 and D2 had witnessed the same candidate result. Node C generated a coordinator commit message for a higher claim round, and the old claim became invalid after being overwritten by a new commit or after the lease expired.

[0044] like Figure 7 As shown, in a chained network, when a critical edge repeatedly appears and disappears due to short-term occlusion, each change in the critical edge alters the topology summary and resets the stability window. The system does not form a commit until the edge state is stable. When a node receives an old commit below the current declaration round fence after restarting, the declaration lifecycle module 120 refuses to use that old commit to refresh the current declaration lease.

[0045] The status output module 160 can output the coordinator status, including undeclared, initial generation pending, adopted, and conflict convergence pending, as well as the current declaration identifier, remaining declaration leases, snapshot round identifier, short topology summary value, number of observed declarations, candidate nodes, commit reason, and snapshot mismatch count. The user-facing interface can display only the group connection status, with detailed fields used for project logs, mobile diagnostics, or backend analysis.

[0046] The duration, threshold, digest function, authentication method, candidate score, and message name in this invention can all be configured according to network size, mobility speed, link capacity, and security level. Any implementation that achieves the linkage between time-limited declarations, common normalized topology input, confirmation after a stable window, and conflict declaration domain coverage submission falls within the scope of this invention's concept.

Claims

1. A coordinator convergence method in a mobile wireless group network, applied to communication equipment, characterized in that, include: Receive node identifiers, node instance identifiers, reachability observations, and coordinator declarations from nodes in the same group; Based on the authentication results, coordinator instance identifier, declaration sequence number, declaration round, and declaration lease of the coordinator declaration, a set of valid coordinator declarations within the current communication component containing the communication device is determined. When the set of valid coordinator declarations is empty and the current communication component contains at least two active nodes, or when the set of valid coordinator declarations contains multiple different valid coordinator declarations, a topology snapshot collection round with a snapshot round identifier is initiated or joined. The node observations and edge observations that are fresh and have passed verification within the topology snapshot collection round are normalized to obtain a topology snapshot and a topology summary, and candidate coordinators are determined from the topology snapshots according to a predetermined candidate sorting rule. After the topology summary, the candidate coordinator, and the original declaration set (which serves as the starting input for the topology snapshot collection round) have remained consistent and reached a stable window, a consistency proposal referencing the topology summary, the candidate coordinator, and the next declaration round is sent or acknowledged. Acknowledgments for the consistency proposal are collected. When the number of acknowledgments reaches a preset acknowledgment threshold, and when the original declaration set contains multiple different valid coordinator declarations, the acknowledgments cover the declaration domain corresponding to each valid coordinator declaration, a coordinator commit message carrying a commit certificate, the topology summary, the next declaration round, and the declaration lease is generated. And after verifying the submission certificate, the topology summary, and the declared round fence, the coordinator declaration corresponding to the candidate coordinator is adopted.

2. The method according to claim 1, characterized in that, The coordinator declaration includes at least a group identifier, a coordinator identifier, a coordinator instance identifier, a declaration round, a declaration sequence number, a declaration lease, the snapshot round identifier, the topology summary, and a submission certificate summary; a valid coordinator declaration also satisfies the following conditions: the coordinator is located within the current communication component, the declaration lease has not expired, the declaration sequence number is a fresh sequence number relative to the same declaration identifier, and the declaration round is not lower than the declaration round fence of the local machine that has adopted the submission chain.

3. The method according to claim 1, characterized in that, Also includes: The state remains undeclared when the set of effective coordinators is empty and the current communication component contains only one active node. When the set of valid coordinator declarations contains only one valid coordinator declaration, the valid coordinator declaration is directly adopted during the duration of the declaration lease of the valid coordinator declaration, and candidate comparison is not initiated due to the addition of ordinary members, changes in the quality of a single link, or the node identifier sorting result of newly added nodes.

4. The method according to claim 1, characterized in that, The normalization includes: sorting the node set according to the node identifier and node instance identifier, sorting the edge set according to the two endpoints of the edge and the evidence version, filtering duplicate edges and edges that exceed the edge freshness threshold, and incorporating the original declaration set and snapshot parameters into the summary input; the topology summary is calculated from the group identifier, snapshot round identifier, normalized node set, normalized edge set, the original declaration set and the snapshot parameters.

5. The method according to claim 1, characterized in that, The predetermined candidate sorting rule compares candidate tuples in lexicographical order. The candidate tuples include, in turn: number of reachable nodes, maximum number of hops, total number of hops, direct connectivity, minimum path quality, average path quality, and node identifier. Among these, a larger number of reachable nodes is better, a smaller maximum number of hops and total number of hops is better, and a larger direct connectivity, minimum path quality, and average path quality is better. The node identifier is only used as a deterministic distinguishing factor when the aforementioned indicators are the same.

6. The method according to claim 1, characterized in that, The stability window is reset when any of the following changes are detected within the stability window: the set of nodes, the set of key edges, the set of original declarations, the topology summary, or the candidate coordinator. A node receiving a consensus proposal sends the confirmation only when it has independently computed the same topology summary, candidate coordinator, and set of original declarations, and its local stability window has also been satisfied.

7. The method according to claim 1, characterized in that, Each acknowledgment carries the node identifier of the acknowledging node, the node instance identifier, and the currently used declaration identifier; the submission certificate includes multiple acknowledgments referencing the same snapshot round identifier, topology summary, candidate coordinator, and next declaration round, and includes at least one acknowledgment issued by an active node currently using that declaration identifier for each declaration identifier in the original declaration set.

8. The method according to claim 1, characterized in that, The submitted certificate is divided into multiple certificate pages carrying page number, total number of pages and certificate digest. After the receiving device collects the multiple certificate pages and verifies the certificate digest, it uses the coordinator to submit the message. When a coordinator submits a message with a declaration round lower than the declared round fence, or when a coordinator submits a message with the same declaration round but different content, the current verified status is maintained, or the topology snapshot collection round is re-initiated.

9. A communication device, characterized in that, The device includes a wireless transceiver, a memory, and a processor. The memory stores a program that, when executed by the processor, causes the communication device to perform the coordinator convergence method as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the coordinator convergence method as described in any one of claims 1 to 8.

Images